Weekly Community Telecon - openbmc/openbmc GitHub Wiki

brad: I'm out, canceling for next week.
brad: george, how is warrior going?
andrew: I can cover.
andrew: features from 2.7.  backport redfish validator fixes?
andrew: something like that...do we backport them?
andrew: rest/redfish logging?
andrew: I think the rule is unless it is critical?
andrew: but we don't have any rules?  can we make our own?
andrew: kurt could you weigh in?

gunnar: we have a customer that wants a longer root password than 20 chars.
gunnar: by default root is in ipmi group.
gunnar: so by default you can only have 20 chars.
brad: remove them from the group?
tom: need to create a different user for ipmi
gunnar: didn't see a way to do that using the gui?
brad: does it work?
gunnar: the customer can't create a new user

kurt: two weeks into freeze.
kurt: release cycle.  have not seen any test results.
kurt: would be nice to see some test results.
kurt: no documentation updates.
kurt: no release github issues have been updated.
kurt: I know everyone is testing.  please post results.
kurt: please update your docs.
kurt: please have a look at the release update issues.

sivas: ldap I tested and updated the results in the wiki.
sivas: I put them in the change log wiki.
kurt: we decided to use github issues.  please put them there.
joseph: it is so much easier to put it in the change log wiki.
joseph: i have a todo.
kurt: i'm fine, as long as we decide that.
kurt: why is the wiki easier?
kurt: everybody needs to use it, is the key.

jayanth: pushed code for csr
jayanth: for rsa 2048 takes 25 seconds, 4096 more than a minute
jayanth: for ecc < 1sec.
jayanth: do we want to support for 2048?
jayanth: ed suggested to create private key at the start.
jayanth: two points to discuss
jayanth: do we want to support rsa?
ed: can we limit the key size?  we are only limited for csr things.
ed: if you don't like what openbmc chose, you can upload your own private key.
ed: if you wanted to only support 2048, I'd be ok with that.
ed: the only code I don't want is the code that doesn't meet response time requirements.
jayanth: the 2048 is taking close 25 seconds.
ed: that isn't acceptable.
ed: third option is generate key on startup.
jayanth: only issue is if key compromised, you can't replace.
ed: you can upload a new one.
ed: and you should factory reset anyway.

kurt: a week into freeze.
kurt: no test results posted.
kurt: no documentation patches.
kurt: no feature cleanup.
kurt: I hate to nag.  but for freeze we need to test asap.
kurt: we need bug cleanup.
kurt: please put some time and effort into it.
sivas: some testing is WIP
kurt: I have seen some testcase activity for ldap, thats great.
ratan: you haven't tested ldap on master?
ratan: you can try it on master too.

nancy: quick update on osfc
nancy: prez submits were reviewed last week.
nancy: they should be notifying people shortly.

brad: tsc did not meet last week.
brad: tsc resuming normal schedule 7/16.

brad: doing release tags today (-dev and -rc1)
brad: kurt anything on release?
kurt: we'll keep doing development and merges in master.
kurt: release candidate tagged 2.7-rc1 today, please start testing
kurt: keep in mind for the freeze - test asap
kurt: please post back to the list or irc.
kurt: good to know if it works everywhere.
kurt: I'll remind everyone to write documentation
brad: anything else on the release
andrew: I'll get a job going on Gerrit
brad: oh like a daily (not a gerrit trigger)

brad: any other topics?
brad: see you next week.

brad: should we have a meeting next week?
brad: everyone ok with canceling next week?
joseph: only thing coming up is 2.7 release
brad: release planning next week is still on.
brad: will go ahead and cancel.

brad: <summarized the tsc meeting minutes>
gunnar: so we are not pursing dmtf?
brad: no, benefit unclear, push back.  right nancy?
nancy: yep, pretty much

brad: anything on nmi?

brad: any new topics?
anoo: adding diagrams to docs repo?
anoo: ascii diagrams?
anoo: does the gui have any experience?
anoo: I will send out something to the list.
spinler: fwiw the design template says to use ascii art.
brad: Indeed, you should follow the existing process and allow the reviewer to push for a different tool.
anoo: ok

brad: do we want to consider moving the call again?
brad: participation has been weak lately.
ratan: evening time?
andrew: maybe scale it back to once a month?
brad: this seems like the best thing to do for now.
brad: not taking any action yet.

jayanth: allowing the user to upload certs.  no validation there.
ed: it does include verification.
ed: there was no certificate service.
ed: what is the issue with that?
jayanth: assume user uploaded one.
jayanth: cert manager validates, and says good.
jayanth: bmcweb, restart, fail, and restart with self-signed cert.
jayanth: is assumes you have the root certificate.
ed: ok, sounds like a bug.
ed: that code did get merged late last week.
ed: if the bug still exists, reopen it.
jayanth: I'll update the bug I think.

jayanth: nmi stuff?
nancy: do we want to mention osfc and openbmc
nancy: I just go the email about it.
nancy: last week I finalized with osfc.  we will have an openbmc track at osfc in september.
nancy: this will be the openbmc hackathon.
nancy: anyone going to osfc so we might have a wider audience.
nancy: there will be a call for presentations at the end of the month jun 30th.
nancy: we might want to shuffle things around.  our hackathons have been a bit self organizing.
nancy: ticket regs are open already.
nancy: two days at google and two days at facebook.  two days of presentations, two days of hacking.

ed: jason is trying to walk through the whole chassis ipmi commands and power state schema in redfish.
ed: and get an interface that supports each one.
ed: one is nmi.
ed: at least on x86, we just assert a gpio
ed: I will check on it and come back next week
brad: will leave on the agenda,

sivas: issue 3557
ed: the ntp daemon needs to do the validation.
ratan: testing of single ip address is possible or not?
ed: no.  redfish clarified that.
ed: they updated the section of the spec that deals with arrays.
ratan: rajesh opened a dmtf forum thread and got a different answer.
ed: put the link in the bug or on irc.
ed: I could have misread the spec.
ed: I thought we were doing what the spec said.
ratan: we can discuss on irc.

ratan: ldap service can be used?
ratan: ad, ldap are type of account provider.
ratan: I have opened a thread with Jeff A on dmtf.
ratan: do you see something else I should do?
ed: the most important thing is that we are impmlementing to the spec.
ed: it looks odd that its called ldap service within ldap directory.
ed: but if that is how redfish defines, it, I can't change it.
ratan: I understand the concern.
ed: is there a redfish forum thread?
ratan: I paste it into the gerrit thread.
ed: I seem to vaguely remember that.  I don't remember specifics about which properties to setup for AD.
ed: I have not fully dug into the AD patches.
ratan: it would be good if you can look at others too.
ed: I got the bmcweb patch list down to less than 50 last week.
ed: I will get there.
ed: just make sure it complies to the spec.
ratan: the validator passes.
ed: the service validator isn't going to test error conditions, bad inputs, etc.

brad: [went through the TSC minutes]
brad: any topics?

joseph: please go review my security documents.
joseph: a second attempt at a formal risk model for OpenBMC.
joseph: don't want to discuss here.
brad: did you talk about this in security working group?
joseph: yep, it seemed well received, they said write a document.
joseph: OpenBMC is pretty abstract, so it is hard.
joseph: it has useful information as is.
joseph: it is a work in progress.

brad: anything else?
kurt: we are nearing freeze time for the release.
kurt; talking about branch and test vs test and branch in the release planning WG.
kurt: I have a drawing.  I'll post it.
kurt: please have a look.
kurt: please jump in the mailing list.
brad: when is freeze day?
kurt: week of the 8th of July.
kurt: release week is the week of the 5th of August.

derick: working locally and pointing at a bmc, only certain requests are working.
derick: we are getting an xsrf token generated when on the bmc directly, but not locally.
derick: we can login, and that post is ok.
ed: right, thats how it works.
derick: even with the INSECURE option?
ed: oh.
ed: there are only 4 spots where you can get kicked.  do you know which?
ed: do gets suceed but posts fail?
derick: posts work for login but nothing else.
ed: it sounds like the xsrf flag is not working.
derick: without that flag I can't login.
ed: login is a special handler, because it defends other urls, it has slightly different semantics.
ed: it is on the whitelist, but isn't itself whitelisted.
ed: if I had to guess, something with the flag is not working.
ed: its the only place where we have significant difference between get/post.
derick: thanks!
joseph: so you have enough info to go forward.
joseph: and the cross site scripting tokens aren't working? with the flag?
ed: oh - are you getting any content security policy warnings.
derick: I'm getting a 500 when I call the account service.
derick: there are a few different ones I can never get.
ed: 500 or 401?
derick: I'm getting unauthorized.

brad: can I take off oem?
andrew: it sort of an ed discussion.
andrew: he hasn't been joining
andrew: we'll need to do it eventually.
andrew: hopefully the answer isn't fork bmcweb.
brad: will remove then until we really need to do it?
andrew: sounds good.

brad: can do an openbmc oem command?
brad: do we need to talk about it here?

ratan: agreed.  can do the same for dump.
brad: ok, wiil take this off the agenda.

sivas: whats up with 3547?
ratan: we can make the changes.

brad: 3547 is a bug report - please bring it back up if someone disagrees that its a bug.
brad: for 3522, have someone at your company propose a design to implement the feature.

brad: ed joined.  should we re-visit oem?
andrew: can we do compile options?
andrew: you didn't seem interested in dynamic libraries.
ed: I'm fine with maintainer folders.

ed: can we have a mapping?
andrew: don't you still need an oem mapping?
ed: not really.   you could but it isn't required.
andrew: cool, I'll have to look into that.

ratan: we store data on the bmc.
ratan: how can we do that?
ratan: is it ok if we have our own schema?
ratan: we can have a rest way?
brad: so, if we make a non-redfish api, and it can be compiled out, that is ok?
ed: yeah, that works.

brad: open call for topics?

brad: no meeting next week

brad: what is the value of this call?  Do we still want to have it?
joseph: you are pushing for work to happen in other forms
joseph: value is for people that get stuck in other channels.
brad: so if we don't have attendance people aren't getting stuck?
joseph: work ebbs and flows
anoo: I've seen cases where there is a topic discussed on irc
anoo: bring it up on the call and things get resolved in five minutes
anoo: maybe this is a place holder.  sometimes having a call is more efficient.
vernon: its worth having on the calendar.  can always get time back if there are no topics.
brad: great, will keep it going.  thanks for the feedback.

brad: no meeting next week
brad: memorial day, US holiday.

brad; should we remove oem topic?
andrew: the piece that wasn't fully resolved - how can we do it downstream?
brad: will keep that on for when ed joins

brad: open call for topics?

andrew: three pieces
andrew: making some good progress with upstream on dmtf.
andrew: i'm ok with that.
andrew: but second piece -what would that look like.
andrew: sounds like everyone needs to agree it makes sense to do it.
andrew: we don't want to do oem commands as a community.
andrew: what does a company do.
andrew: dynamic library loading?  config flag options?
brad: not much to talk about without the maintainer.
brad: its just ibm on the call today, except richard.
brad: lets try again next week.

andrew: logging: changing phosphor logging again
andrew: infra: we have a new system coming at ibm - witherspoon 128
andrew: initial pldm and mctp support.  should we use community resources to build that?
andrew: hardware CI running internally.
andrew: it tests the mctp/pldm stack - pro
andrew: using community resources - con
andrew: please respond to my list email or in the infrastructure workgroup

ed: jeremy wants to write it as a kernel module.
deepak: that was a longer term approach.  I don't think jeremy is chasing that right now

ed: backlog of code reviews, design reviews
ed: so if its correct or error free
brad: are you seeing code that doesn't have +1s or isn't reviewed.

ed: there aren't a lot of patches.
ed: it takes a lot of times when I have to write 50 comments per patch
ed: I don't want to impose a policy of ed has to have a +1.

joseph: would it be fair to get a list of reviews?
joseph: we could tell the ibm team to run that query.

brad: can we set your priority?

ed: give me 48 hours and I'll get to that email.

kurt: here is a thought.
kurt: planning a workgroup would be good.
kurt: you can talk for specific subjects for an hour at a time.

brad: Terrible minutes today folks, I was heavily into the disucssion.  Sorry ABT.

brad: (recapped TSC meeting minutes)

brad: should we talk about oem?
ed: dont like ipmi
ed: all dynamic llinking
ed: hard to get guarantees at compile time
ed: everyone seems to have their own brand
ed: makes it hard for distributed change
ed: because of how seperate it is.
brad: willing to chat with others?
ed: yeah no problem
gunnar: invite them to the community call?
joseph: what are the redfish choices?
brad: an oauth api?

ratan: we have management controller that needs other oem things

joseph: eclypsium is a security org that targets bmcs is going to talk at security workgroup
ratan: another topic for oem snmp client solution - no client configuration

brad: will do general tsc minutes

brad: andrew?
brad: ed isn't here
andrew: we are adding function that doesn't have a schema?
andrew: hp, dell, supermicro, they oem things all the time
andrew: in upstream OpenBMC are we going to allow that?
andrew: or do we only want to support standard non oem things?
nancy: I'm ok with oem commands that are OpenBMC specific
nancy: but can we work with redfish community to make them more standard.
andrew: the problem I'm having is getting response from dmtf
andrew: no feedback.  weekly meeting is always canceled
andrew: struggling with dmtf process
nancy: I'm setting up an alliance partnership between dmtf and openbmc.
nancy: orgs work together to mutual benefit.
nancy: how that happens I'm not sure.
nancy: google just joined recently.
nancy: we should go ahead with oem
nancy: but still work with them.
nancy: maybe we just need a bigger voice.
andrew: how do we decide?
andrew: anything we can do to speed the process up.
nancy: I started an email thread with people at dmtf.
nancy: I'll cc you and brad on it.
joseph: I had proposed that instead of everyone proposing oem, we get an OpenBMC OEM agreement.
nancy: I think that might work really well with our alliance partnership
ratan: if multiple companies have the same interface, its really good.
maury: getting OpenBMC community support of an OEM interface would give leverage to influence the DMTF.
supreeth: tianacore edk2 we have the same problem - implementation is faster than specification
supreeth: tianacore we can submit the implementation and then change the spec.
supreeth: if we have alliance with dmtf, we should raise that approach with dmtf alliance and see what they say.

gunnar: your change will go in, its just maybe bad timing?
andrew: we'll see.
andrew: whenever I go Lenovo has a lot of time.
andrew: I think a lot of it is just getting on the queue.
andrew: it can take weeks/months to get on the agenda

brad: what about value add?
nancy: having the ability to add oem commands in convenient.
nancy: I don't think we should disallow.
deepak: even some existing schema have oem properties
deepak: you can get ipmi sels out through the log service
deepak: in those instances we'd need a plugin framework
deepak: is not allowing that a deviation from the schema?

gunnar: I have a question for nancy about the alliance
gunnar: what will it entail?
nancy: hoping it will allow openbmc members participate more easily
nancy: you have to setup terms
nancy: we can work to figure out what those terms are.
nancy: short term goal was to make it easier for OpenBMC community members to have a voice.
supreeth: there are existing members in dmtf as well as OpenBMC (companies)
supreeth: arm is part of redfish workgroup
supreeth: once the alliance gets finalized would you suggest individual companies stop participating and participate as a single OpenBMC group?
nancy: let me find out.

brad: open call for topics?

supreeth: question for deepak
supreeth: did you get a chance to present the pldmtool?  make any progress?
deepak: no we haven't talked since the last pmci meeting.  we are working on it.
supreeth: guess we need to cooporate on that.
deepak: we should post on the list and pmci working group mailing lists.

brad: nothing new on tsc

brad: deepak you have the floor

deepak: we use rsyslog
deepak: some customers wanted some kind of audit logging.
deepak: not audit logging proper, but logging rest api calls.
deepak: the old rest server could log the rest api calls.
deepak: they'd be logged to the journal and streamed out via rsyslog
deepak: can we continue to do this via bmcweb?
deepak: or should we come up with some other audit logging solution?
ed: so I missed some of the context
ed: if done correctly in a way that meets gdpr requirements and doesn't cause a
performance impact it should be fine to put in bmcweb.
ed: did I understand?
deepak: I think so.
ed: need to scrub passwords and user data
ed: would be nice to be async
ed: username
brad: username?  how does that work
ed: you have to opt in.
ed: off by default and disableable might be ok.
joseph: what user data could we collect?
brad
deepak: without the username there might not be much to audit.
deepak: what is the best way to verify?
ed: one other thing - you mentioned rsyslog, have a look at redfish
deepak: we are interested in redfish logging too.
brad: leave on the agenda?
deepak: yeah, leave for one more week.

brad: gunnar, go ahead
gunnar: wanted to talk about gui sensor page.
gunnar: it displays all sensors in the system.
gunnar: there is a feature added so you can select which chassis
gunnar: that is fine, but we should really look at the new sensor collection?
gunnar: and not thermal/power redfish schemas.  those are a subset of the
sensor collection.
gunnar: do you agree?
ed: yes, I think.
ed: we still need to be careful of the power/thermal schemas.
ed: you have power, you have thermal.  sensors collection are things that don't
fit.
gunnar: I thought of it as all your sensors.
ed: we should ask on the redfish forum.
ed: you would think they'd define that.  superset or something else.
gunnar: apulis change...I thought I had looked, but it seems to be missing some
sensors.  we've regressed.
gunnar: we've lost some of the sensors on the page.
ed: is that because they aren't in Redfish?
gunnar: yes.
gunnar: it has been merged.
gunnar: I'll ask the question on the redfish forum.
gunnar: I read something that I thought meant it was for all the sensors in a
chassis?
gunnar: how do you feel about reverting?
ed: not great, but understand.
ed: the thing we need to get the bottom of it, is the one chassis patch
gunnar: I don't see how you are getting current sensors.
ed: I'll hop on IRC.
ed: but I guess I don't have a problem with reverting it.
ed: well if we go to sensor collections, its the same thing right?
gunnar: you'd have something where you select what chassis youu have.
ed: I thought he was doing a flat list.
ed: lets hop on IRC and put together a plan.

brad: open call for topics?
brad: mapper?
gunnar: matt did a bunch of work on the associations.
gunnar: aren't associations going to stay around?

joseph: previous topic - user account added - that is where you need to get access.

richard - submitted doc, it got merged.  please go through.
joseph - that just says what we want to do.
joseph - are we going to have a design based on that?
richard - adding new openbmc test user.
richard - implementing oem command
brad - is that in what got merged?
richard - the doc covers basic design, and how the flow looks.
brad - can I take off the agenda.
richard - will work with CI team, to go through phases.

joseph - I went ahead and sent a couple emails.
joseph - I created two new wikis
joseph - one is a work in progress.
joseph - the security wg has had one.
joseph - we'll have to be aggressive about moving things off the list.
joseph - when its done, it would move to the changelog wiki.
brad - is this in a security context only or project -wide?
joseph - project wide?
kurt - why wouldn't we just use the github issues list?  like we do for release planning?
joseph - github issues are too messy.
brad - why not use the github issues list for work in progress list.
kurt - worried about having two steps for users.
joseph - the alternative to the wip wiki is to do a query over the issues?
kurt - yeah, if we had a big list of issues, it might be a problem, but we only have 13.
kurt - if there is a security concern, we can tag them somehow.
brad - do we want small issues in github then?

joseph - where should I open the issues.
kurt - start with openbmc/openbmc.  long run, we can open in per-repo projects.
joseph - so the direction I'm hearing is, I should work to put wip wiki items
joseph - I should create openbmc/openbmc issues.
joseph - so for the changelog - what do we think of that?

brad - open call for topics?
richard - mfg mode?  do we require any special modes.

joseph - does it have to be mfg mode?
joseph - are you changing it back?
joseph - multi-level security.
joseph - you have to be authorised (discretionary)
joseph - second level - labeled security (system owner)
joseph - lock people out even if someone wants to give you access.
joseph - would it be cleaner to do as a different form of multi-level security?
richard - by default it won't be in mfg mode.
joseph - you enter the mode by adding a certificate?
ed - more like signing a csr that says yes I want to be in this mode
joseph - if you forget that you are in that mode?
joseph - you would look for that certificate?
ed - there tends to be timeouts
richard - it requires the knowledge of the user.

brad - is it just mfg mode?
richard - there are few modes.  deployment.  deployed.

brad - open call for topics
jason - question on qemu and ci
jason - just started looking wolfpass qemu model.
jason - looking for pointers on what kind of things are done on ci.

joseph - infrastructure wiki - it has lots of details.

ratan - during ldap configuration we have password that needs to be changed?
ratan - don't need to send the password over dbus?
ratan - is that a security issue?
ratan - we can change it via dbus property

brad - can bmcweb write the password?
ed - that doesn't work because bmcweb won't always run as root.

brad - how do you do it?
ed - you pass a file descriptor
ed - send the file descriptor over dbus.

ed - the other option is we send the ldap secret over dbus, and document it.

brad - no update on the TSC report
brad - richard not here

shawn - working on getting fan and temp support in redfish
shawn - we need to return accurate status
shawn - health field and state field
shawn - I'd like to return as absent if the hw resource is missing
shawn - I need to find a way for the redfish thermal sensors to return accurate health state and present state
shawn - I wanted to use existing dbus object paths interfaces and properties
ed - can you send the slides to the list
shawn - sure
shawn - sensor health.
shawn - we could obtain from the functional property of OperationalStatus.
shawn - this interface isn't usually implemented on sensor object paths.
shawn - there are two apps that do inventory.
shawn - hoping the concepts are the same.
shawn - here are a couple inventory paths.
shawn - that represents the physical hardware.
shawn - we could link sensors to an inventory item.
matt - for sensor health. say you have a fan that burned out (rotor doesn't spin)
ed - most redfish implementations don't differentiate between a failed sensor and failed item.
matt b - how does anyone decide what is broken then?
ed - for i2c sensors, we have other sensors for that.
matt b - but nothing from a fan perspective to de-lineate between fan, wire, controller chip.
ed - not that I'm aware of.
ed - is it not better to get the ball rolling?
ed - we've talked about a sensor status interface.
ed - everyone was looking at adding something new.
shawn - would it be on the sensor OP or the inventory item?
shawn - assume its good, then if an interface exists, override.
matt b - the tricky part is it ties to whether a fault file exists.
ed - in the case where you have parallel redundant fans, and you pull two?
matt - no, in the case of witherspoon, we have an external fan controller chip
matt - it only exposes a fault file for half of the rotors.
matt - because that fault file exists, it will expose that property for only those four rotors.
matt - my issue is, how do I decide if the others are functional?
shawn - what does Redfish expect?
ed - those are two different conditions
ed - the rotor spinning at twelve - that should be in the sensor interface.
ed - whether or not the fan is present, is a function of the inventory.
ed - the overall redfish status is a merge of the two.
shawn - if the sensor is able to read status at all, its ok?
ed - no, we can still trigger a fault in redfish.
ed - in Dbus, those are two different interfaces.
shawn - so merge them?
shawn - thats a different function than what I was thinking about.
ed - looking at the warning and critical interfaces.
gunnar - shouldn't you also look at functional, not just the presence.
ed - implementations tend to not.
shawn - functional status on the sensor.
shawn - functional status on the inventory item.
shawn - we've got missing - (presence)
shawn - comparing the actual value to the thresholds
ed - redfish shouldn't be thresholding.
ed - it depends on the fail case.  there is no rollup.
ed - you are aggregating different dbus conditions into one dbus property.
shawn - whatever we need to do.

susan - design wg meeting moved.  we've moved it back to 7:30 central time.  I know thats hard for west coast.  I felt like that was a good compromise.
susan - my team would like to talk about the network panel.  also working on engineering design for remote logging.
susan - I know that Intel is working on a couple panels.  More than welcome to share those.

brad - shawn back to you.
shawn - the sensor state - enum with 10 or 15 values.
shawn - bmcweb always shows enabled.
shawn - proposal was to have an association.
shawn - is this a good way to determine if the sensor is present?
ed - that sounds reasonable to me.
shawn - might be weird edge cases.

shawn <gave a summary>
shawn -  am I close?
ed - sounds pretty close.
shawn - do we need to effect the health too?
ed - oh, we are talking about using two properties.
shawn - so there is a status object in the redifsh output.
shawn - health and state are returned - these are the two I was trying to set.
shawn - for state, that is the one where I was going to have it show absent if absent.
shawn - for health, it is one of three things, ok, warning, or critical.
gunnar - I'm trying to look at other things (processor, dimms)
ed - and I'm looking through the spec.
ed - best thing is to take this to the list.
ed - we can go to the redfish forum to find the intent.
ed - I'm sure they've thought of it and just not documented it.
shawn - I will just try to write down what I said.
gunnar - in the sensors - there are dbus properties, there are critical/warning.  look at the status.a
ed - step one, figuring out what we want long term.
ed - step two, cut it up.

brad - do we want to talk about focused hackathons?
maury - tempted to not push on this too hard
maury - what would help me is to make sure we are talking about hackathon and
        not a meetup
maury - A meetup is meet, talk about designs
maury - a hackathon is lets crank out some code.
maury - as long as we know the intent, that would really help me out.
maury - I won't pursue at this time.
nancy - no solid plans for the fall.
brad - will drop from agenda?
maury - sounds good.

brad - richard?
richard - no response to my note
richard - quick check with who is on the bridge?
brad - what is the impact?
andrew - from a CI we'd need to get a workaround in place?
richard - two points
richard - do we have systems where you want to ship with default user name and
          password?
richard - instead of root as default, can we create something else?
andrew - does that impact running commands from the shell?
brad - is there a high level design?
ed - one conflict we are going to have is logging in as root.  we don't have a
     good solution for that.
richard - I want to get rid of exposing uid 0
brad - will leave on the agenda to make sure we are getting traction going
       forward.

richard - net ipmi commands not exposed over dbus.  sol config parameters.
          same with session information.
brad - <gave a long rant about why design templates are useful>
deepak - isn't this something that the ipmi architecture solves?
kurt - is there an issue for this in github?
richard - yes, one was opened today?
kurt - can you post a link?
kurt - will that be a release 2.7 feature?
brad - can they be marked as 2.8?
kurt - yep

brad - ratan, you have the floor
ratan - thanks ed for reviewing
ratan - as part of get request, ldap property would not be there.
ratan - in ldap there account provider type.
ratan - doing patch for ldap property it works ok.
ratan - ran validator for ldap config there and not there and it passed
ratan - need some inputs on more changes
ed - the schema doesn't change whether or not there is a configuration present?

ratan - system comes up and there is no ldap configuration
ratan - we have default values
ratan - user patches with valid values

ratan - need to blow up the properties under ldap

ratan - backend supports one type of configuration
ratan - either ldap or AD
ratan - in the schema we have ldap and ad property
ratan - both props are type EAP
ratan - EAP type has property account provider type - which can be either LDAP
        or AD
ratan - can we use any type of ldap configuration?
ed - we would not support patching type
ed - not supported message - or resource not supported.  action not supported
     is what you want probably.
ratan - suppose user disabled ldap service
ratan - and user enables AD, what happens to ldap

ed - can we agree it would be better to get a series of patches that gets ldap
     in first?
ratan - yes

brad - nothing new on tsc report
brad - focused hackathons - maury not on

gunnar: I saw the video
brad: anyone want to say anything
brad: <gave an overview of what he saw at the summit>
gunnar: how many people from OpenBMC at the meeting?
brad:
gunnar: will OpenBMC keep going to OCP since it is a LF project
deepak: FB had an interesting demo of the tiogapass system running OpenBMC 2.6

brad: keep an eye out for fallout from ocp

brad: maury - do you want to talk about focused hackathons?
maury: not today - lets revisit next week.

maury: speaking of the hackathon - was there any discussion about an upcoming hackathon
brad: the TSC did talk about it privately, everyone seems more flexible

gunnar: so more meetup and less code writing
brad: didn't see any hacking or pair programming
nancy: there wasn't critical mass

gunnar: I would like to see a hackathon this year
brad: what would we hack on this fall?
gunnar: some presentations and getting everyone together
gunnar: we could poll the community for things to hack on

andrew: set up test gerrit to gerrit 2.15.  go play around with it
andrew: will do the upgrade tomorrow.
andrew: small outage window tomorrow.  before west coast is online.

andrew: going to stop testing meta layers that aren't tested.
andrew: but Nancy if you have a system that builds, we'll add it.
nancy: I'll go back and take a look.
andrew: just a general fyi - I put the affected meta layers in the note thread.1

brad: open call for topics?
brad: trying to get us starting on time
kurt: I'll be doing the same

gunnar: lots of design docs out there.  really looking for some reviews before merge.

richard: sent default user account email.
brad: how does this impact testing?
richard: that is a problem.  have some ideas.

brad - going to start meeting at 11:04 next week....getting back to on-time

richard: can bring back up next week

andrew: lots of autobumps aren't working
andrew: biggest one are ipmi
andrew: user manager
andrew: state manager
andrew: obmc-console
andrew: should we do more to highlight this?
ratan: for user manager, that was failing because of dependencies
richard: need to merge with phosphor-dbus-interfaces (for user manager)
richard: for ipmi - vernon provided fix that hasn't gotten +2
andrew: sounds like we are ok with andrew/brad adding people to the review?
richard: autobump should be first priority
andrew: host ipmid has been broken since feb 27th
brad: very small sample size of maintainers on the phone...
ratan: we have not faced this problem before
brad: we are updating new interfaces instead of just adding new ones
ratan: is there a better way
richard: is it ok to rebase patches?
brad: yeah!
brad: vernon did you understand the question?
vernon: yeah, that works for me
vernon: it can be hard to dig through the links but Andrew helps me.

brad: open call for topics
joseph: when the commit goes into the repo, then its in the repo when its merged
joseph: then autobump when openbmc repo is updated.  is that correct?
joseph: you don't need to have the autobump finished right?
joseph: you could update you own metadata for it?

joseph: we talked in security workgroup...could go over that
brad: sure, go ahead
joseph: other people might want to join the security wg.
joseph: we agreed to have additional security workgroup meetings
joseph: there was some resistence to that
joseph: if we have a single workgroup, we have a single wiki for it.
joseph: we can have multiple meeting times.
brad: meetings aren't good for big OSS projects.
anoo: kurt set up meeting bot in IRC.  we haven't used it ever.  we could give it a try.
anoo: hurdle is people don't know how it works.  might be worth a try
kurt: don't have the bot setup
kurt: we do have an openbmc-meeting channel.
kurt: anyone can jump in there and have a gathering.
kurt: go for it.
ratan: pldm and mctp - hoping we aren't having workgroup and still we are collaborating.
brad: how is it working?
ed: its a much more concrete problem?
ed: not much to discuss
joseph: will try to push agenda items into the email.

brad: any more topics?

kurt: plug for designs
kurt: beginning of 2.7 cycle.
kurt: quite a few work items.  not everything is there.  need to get list finished up
kurt: if you are working on a major feature, without a design doc, please get one started.
kurt: let me know if you have any questions.
kurt: lets get these work items, issues, designs created.
kurt: just a plug for that.
brad: is it written down somewhere?
kurt: sent email, talking in irc, and release planning meeting.

joseph: firewall design, was a little light on why we needed it
joseph: what I want to do going forward is have a network vulv assessment.
joseph: bmc is in the lab network.  what kinds of things openbmc is doing and needs to do to protect itself.
joseph: we have things in place already.  passwords, etc.
joseph: we don't have a document that explains all that.
joseph: that is the piece I think I want to put in place as a prereq to putting forward a firewall design.
ed: sounds reasonable.  same as security wg last week right?
ed: no objections in that meeting.

susan: just a reminder
susan: looking at changes to firmware panel
susan: I would like to see changes to local user management
susan: so it follows already established patterns.
susan: these are what IBM wants to show.
susan: if any other teams want to show, would be great for time.

richard: want to talk about manufacturing commands in the factory
richard: button workon ok.  led working fine?  testing before leaving factory
richard: do you provide special commands.  how to make sure they aren't not executing in field.
brad: ibm certainly does manufacturing testing
joseph: function that is more interesting when bmc is being serviced?
richard: right - to test that hardware is ok?
joseph: turn off function so it can't be used?
richard: fan testing.  test at 25% pwm.  see if temp range ok.  test at 50% pwm, check again.
richard: but at the same time, these commands can't be exposed to the end-user.
richard: security risk to expose those.
tom: how do you it richard?  using test binaries.  or special ipmi commands?
richard: enter mode through special commands.
joseph: are those built into the image.
richard: built in.
richard: how to secure it even more.
ratan: to answer - we disable certain users.
ratan: we have a firewall, turn it on.

richard: next topic - exposing root user.
richard: by default root user is exposed and default password.
richard: we don't want any users to be in the system
richard: we don't want root in redfish ipmi
joseph: I assume when you grab a copy of the system, and you fork, you can turn off root sign-on.
brad: anyone have issue with this direction?
george: from a test perspective, what do I have to do?
ed: there is a california law coming that bans default passwords.
ed: it is specific to network interfaces

ratan: when we say default password is illegal - can we have default password during mfg?
ed: what matters is what you ship.  the defaults should be shippable.
brad: why?
ed: secure by default.
brad: ok so secure by default and not shippable by default?

maury: one dilemma with hackathons is getting all the right people together
maury: and getting travel ( we have a global team )
maury: I've gotten the sense that the hackathons might be less effective if you
maury: don't pair up the right people
maury: one of the thoughts was lets target more focused topics
maury: lets reach out, find out who are the experts.
maury: pldm for example.
maury: get those people together and let them focus on the work, design, hacking.
maury: if you break it up, you could be sending lots of people on airplanes.
maury: we use webex for this - its a way to establish webex.
maury: do something first in a webex?
maury: talk about requirements, design, refactoring
maury: lets get the people together first.
maury: would that be a more effective way of doing hackathons.
maury: in the end, we'll have a reason to travel (focused on an individual topic)
ed: how is that different than the working groups?
maury: could be the same.  we have higher level ones.
maury: maybe we have to target even more.  with just the tech experts interested in that topic.
ed: you want the topics to be more specific.
maury: if you are just focused on pldm.  lets get them together.
maury: they could be very short-lived workgroups.
maury: the intent was to minimize hackathons and focus more on specific topics, trying to accomplish what a hackathon might have accomplished.
nancy: not hearing anything different than the other techniques today.
maury: something about a hackathon is its about building relationships.
maury: build relationships without traveling.
maury: maybe its just extending the working groups into more specific topics then.
gunnar: I don't think the working groups replace a hackathon.
nancy, ed: me too.
maury: what do you get out of the hackathon?
nancy: I've found them valuable for building relationships.  I find it useful from a lead perspective.
nancy: my management also finds it valuable.  our team is smaller.
maury: whats the definition of a hackathon then?
maury: maybe we should have this discussion again after the next hackathon.
nancy: I have been thinking about the next one.
nancy: google hosting a related event, it might/could include OpenBMC.
nancy: problem with workgroups is someone has to lead it.
nancy: not everyone attends.
nancy: can you add next hackathon as a topic
nancy: I wanted to review based on what Google is planning for the fall.

george: all this hackathon sounds more like a development thing.
george: for the community is there anything with test?
ed/maury: there is a test work groups.
maury: beyond that I'm not aware of anything
joseph: test should do test-driven development

ratan: question regarding bmcweb review?
ratan: do you have too much code coming in?
ratan: how can we expedite?
ed: I don't understand
ratan: too much code coming in?
ratan: can we expedite somehow?
ed: what code reviews need to be expedited?
ed: I check every day
ed: if there are some you think I missed, you can point them out.
ratan: I can send the list.
ed: can you check that all comments have been addressed?

brad: tsc met last week
brad: tsc agreed to record meetings.
brad: want to make sure thats legal first :-)
brad: just need one more tsc meeting to wrap up.
brad: will add a docs/tsc page.

brad: andrew/kurt - release update?
kurt: yes, we are done.  we have tagged.
kurt: off and running for next release
kurt: interesting thread on tagging the documentation repo
kurt: version of documentation that aligns with release tag.
kurt: any additional comments?
andrew: don't like the tag.
kurt: talked about in release planning - we 2.6.240
brad: 2.6.240?
nancy: so what can we do?
ed: something to tag the master branch.  so we can have incrementing numbers?
kurt: the heartburn is not necessarily the tag name, but the rc1.
brad: we could do like yocto.
kurt: this goes away if we test against master.
brad: so long term, we maybe try tag and then branch
brad: short term, I can do a dev-2.7 tag
brad: andrew, can you propose a tag in #openbmc?
andrew: sure.
brad: can take off the agenda?
kurt: yep, I will add to release planning meeting agenda

brad: I have Maury/Kurt for demos.  Anything to talk about?
maury: we are doing internal demos.  can we demo here?
brad: everyone ok with doing a demo here?
maury: we can do youtube.
ed: probably better to not do it in this call.
kurt: should be live.
ed: I just worry about audience.
kurt: like everything else, lets just try it and see how it works.
maury: it is going to depend on the content.
maury: andrew, maybe we can see if any items are worth making external.
andrew: seems like pldm might be a good candidate.

brad: whats up with pldm issues?
deepak: will send email to the list we could use help with, and describe what we are working on.
deepak: but long term we need to find tooling.
deepak: being discussed in release planning
brad: so what do we do with this?
maury: couldn't github issues work?
...: long chat about how to work around github permissioning
kurt: the goal - a rep from each org, be responsible for ccla
andrew: there is a github api for adding someone to an organization
kurt: lets take this forward in release management planning meeting
reed: I've done a lot of this automation too
kurt: cool
kurt: is anyone opposed to group management being automated over to github?
...: <crickets>
brad: so can this be tracked in release workgroup? taken off the agenda?
kurt: deepak - you ok with that?
deepak: as long as I get the minutes.  can't make the meet because of timezone

brad: open call for topics?
brad: kurt will be running the meeting next week.

brad: no tsc meeting last week - meets tomorrow.

brad: kurt - release update?
kurt: hoping we can tag the branch today
kurt: release notes are progressing nicely.
kurt: gunnar tagged some folks.
kurt: we should not wait on tag for the release notes
kurt: documentation effor tlooks lin good shape
kurt: release planning meeting today - company reps should bring 2.7 requirements.
kurt: also wants to pick a new management tool
gunnar: I just merged the release notes.
brad: within three hours I'll tag
kurt: by end of meeting.
kurt: there were questions about release tag formatting.
kurt: will update
andrew: are you going to tag the subrepos?
brad: lets wait until a need arises.
george: are more changes coming to 2.6?
brad/andrew: no

brad: anything on demos?
kurt: would be good to share.
kurt: table until maury can discuss?
brad: ed you want a reminder?
ed: I'm looking into opening those up.
susan: ed attended last gui workgroup where we shared networing page
susan: ed and the intel team helped us tweak the page to make it more accurate.
susan: by doing those things it really helps
brad: will leave on for now
kurt: idea would be to get something scheduled in the community call
nancy: are we just looking for demos or tutorials
kurt: a demo would be a high level pass with some function, where you are at with it.
kurt: a tutorial was not really the intent

brad: deepak wanted to talk about pldm
deepak: I got some feedback over the mailing list and on gerrit
deepak: interest to collaborate
deepak: there was a question how to avoid duplicate work
deepak: what does everyone think the best tooling would be?
deepak: github issues under the pldm repo?
kurt: we are talking about this in release planning
kurt: I was leaning towards one big epic/issue
kurt: individual repo epics/issues is an interesting idea.
kurt: will bring up in the release planning meeting today.
kurt: if anyone has any strong feelings, please come and discuss at release planning.
deepak: would also want to open github issues in both ibm-openbmc as well as openbmc/pldm
kurt: yep - we all have the same problem - how to reference upstream work in our product releases.
deepak: the description of the work can be upstream.  something like zenhub can attach milestones.
kurt: zenhub can't do that today, but that might be coming soon.
brad: leave on the agenda?
deepak: leave it on the agenda
deepak: subscribing to gerrit notifications for the pldm repo is a good way to keep aware of what is going on there
supreeth: sounds good to me.

brad: open call for topics
ratan: ed is around, related to ldap deconfiguration
ed: I haven't had a chance to read.
brad: two mctp designs.
brad: ed, are we doing duplicate work?
ed: jk is looking at mctp from a new channel to communicate to the host
ed: that isn't a usecase of mine
ed: I need it to talk to add in cards
brad: I need that too.
ed: I didn't see detail on what transports we are going to build first
ed: I didn't see details on what the kernel interface
ed: I looked for jeremy on irc.
ed: it doesn't look like there is any wasted work.
ed: we should probably work to join the proposals
ed: not #1 on my priority list
supreeth: from what I gather from review of jks source, he is constructing a base layer
supreeth: for i2c need to add phys header for mctp packet itself.
supreeth: it is a modular approach.
supreeth: you should consume the library jk is creating.
supreeth: and then invoke the i2c driver / userspace.
ed: I've posted comments on jeremys review
ed: some of it is I might be missing details on the design doc
ed: right now it looks like its running over a serial port
supreeth: there is a mctp over serial
ed: some bmcs have hardware support for pcie mctp.
supreeth: now that lot of folks, we need to have a meeting where we discuss details.
supreeth: to avoid duplicate work
supreeth: other question involving mctp
supreeth: need a public one
supreeth: if it has the serial binding it needs to be separate.
brad: I will send a note - supreeth please respond with what you'd like to see
ed: is there a better forum to talk with jk on?
brad: i'll mention that people were looking for him on irc.

brad: any other topics?

ed: one thing for bmcweb - we are trying to get modern cipher suites again
ed: hoping they are less broken
ed: would be nice if people can check the patch out ahead a time.
ed: if you see weird connection issues.
brad: andrew who do we have for that?
andrew: joseph, george
ed: will ad joseph.  george is already on.
gunnar: yeah lets test beforehand.

brad: tsc met, made some progress on community interaction guidelines
brad: approval of a TSC mailing list.
ed: what is wrong with the main list?
ed: need to document when to send to what list?
brad: will push documentation eventually.

brad: release update kurt/andrew?
brad: its this week?
kurt: yes.
kurt: lets hop in irc and talk
kurt: posted release notes - would appreciate review
kurt: need focus on what we are doing for this release
kurt: more verbiage on what features were completed.
kurt: also need more input on what limitations we have
kurt: things that didn't get finished or partially implemented
kurt: bugs that won't be backported
kurt: please hop on IRC an comment
kurt: documentation.  anyone wants to make sure documentation is up to date.
kurt: last push for documentation.
kurt: probably won't complete the release today
kurt: really do need some attention on this.  please hop in IRC and help get this release wrapped up.
kurt: andrew?
kurt: brad backported the fixes we found
brad: should we grab aspeed 4.19.19?
kurt: lets talk about it on irc.

brad: open call for topics?
susan: reminder that our gui workgroup is wednesday
susan: going to be reviewing the network panel.
susan: hoping Ratan can join.

brad: susan - icons?
susan: I wanted to remove.  we had chat on irc and have direction
gunnar: was also sent to the mailing list.

andrew: ed, gunnar, anthony, please perk up.
andrew: the code review has been going on for a long time
andrew: irc chats a couple times
andrew: wanted to take a sec to pick a direction
andrew: LogServices - usually have a log1 - indication of a system log
andrew: you can have multiple log instances
andrew: you can different types - events, sel, or oem
andrew: I'm trying to understand within phosphor-logging we have dbus objects and sels.
andrew: it seems like both could fit?
andrew: Just wanted to talk this out.
ed: sure.  at the end of the day, the dbus logs are not fast enough.
ed: existing bmcs can store 4000 log events.
ed: we need something at least comparable.
andrew: they should be under LogServices events?
ed: no, it shouldn't be under events.
ed: some put it under 1.  most systems call it something with a unique name
ed: to dis-ambiguate from other event types.
ed: in redfish, an event has a specific meaning.
andrew: the phosphor-logging implementation using the dbus apis not maintainable?
ed: not sure about not maintainable?  it doesn't meet the need.
andrew: thats why we came up with the other implementation.
andrew: so how do we move forward?
ed: the todo list needs to be completed.
andrew: agreed.
ed: who owns that?
andrew: anthony.

brad: open call for topics

maury: I have one about showcasing topics.
maury: we try to do agile
maury: at the end we do some demos
maury: that might not be interesting, but as some things come togther might be interesting?
maury: would anyone be interested in doing some showcases?
maury: or seeing what we've done?
maury: something that got merged into master?
maury: we could do one function for the last 10 minutes of this call?
maury: flexible there?
maury: what do people think?
...: <crickets...>
maury: good idea?  bad idea?
ed: we do youtube videos internally
maury: that interesting
ed: we might still present the video in the meeting.
ed: we don't post externally today, but moving forward we could...
brad: we do this too.
kurt: anything we can do to communicate what we are doing is a win.
ed: anyone have a good cross platform app for recording desktop?
kurt: quicktime player on a mac.
kurt: can record screen and speak to it.
matt b: vlc does screen recording too.
maury: kurt, tie into release workgroup?
kurt: lets not limit to one meeting.  I will mention it.
kurt: but lets keep discussing it here too.
kurt: love the idea of once a month having a workgroup lead schedule a brief rundown on function
brad: I will leave on the agenda.

brad: any more topics?
gunnar: can we circle back to logging?
brad: go for it.
gunnar: the dbus logging service - entry type: event
gunnar: the redfish defined event.
gunnar: are we doing to convert the dbus logs to events?
andrew: does it map well?
gunnar: I'm not sure.
gunnar: resolved does not map well.
andrew: ok so openbmc oem and then upstream that?
ed: for redfish event type - you need a message registry
brad: are you supporting events
ed: I need to understand more.
brad: would you have the same number as sels?
ed: no, they are orthogonal.

brad: no update, tsc meets tomorrow
brad: kurt, andrew - release update?
kurt: general status
kurt: should have a release notes patch pushed out today
kurt: would appreciate eyes on that
kurt: started working 2.7 requirements, think on that
kurt: release planning today
andrew: our test team has identifed 3 bugs to be backported
andrew: if there are other problems, please let us know?
brad: what day are we shooting for?
kurt: week of feb 4th

brad: open call for topics?

brad: andrew?  required thud branches?
andrew: put support for thud into the CI scripts
andrew: its all in the email
andrew: did discuss in the infra workgroup last week
andrew: discussed library versioning
andrew: taking a reactive approach.
andrew: everyone ok with that?
andrew: is it worth the effort of going through what I had in the email?

brad: gunnar or iffy - did you know what susan wanted to talk about
gunnar: not sure what she wanted?

gunnar: are we merging bmcweb today?
andrew: yeah!
susan: is ed on?
ed: yep
susan: we talked in a bunch of forums
susan: ed and gunnar both had concerns about committing some code
susan: we are not going to integrate code now
ed: what is the plan on an actual translation
susan: we don't have plans for at least a year
susan: don't quote me on that
gunnar: does anyone have plans sooner?
ed: we could do a chinese translation
ed: but was going to wait for infrastructure merge
susan: I thought the concern was performance related?
gunnar: the only thing translated was the header?
susan: so if we change all the words to varables so they could be?
susan: would that help?
ed: the patch needs to be carried forward to completion
susan: so resource issue is not a problem?
ed: no
susan: the only issue then is finish up the conversions?
gunnar: also need to make runtime selectable

brad: any more topics?
susan: I was supposed to do some homework, didn't get that far
susan: will do it next week
ed: sounds good to me

brad: tsc met, but did not get to discuss interaction guidelines for TSC.
brad: any general 2.6 update to share?
kurt: we've branched
kurt: should be focused on testing
kurt: should be focused on getting documentation updated
kurt: I'm about 3/4 way through release notes
kurt: need eyes to look at that please
kurt: anything else that needs to be added, please let me know.
kurt: sent out a list email - please reply
kurt: andrew?
andrew: from infrastructre pov everything is set up
andrew: haven't turned on for individual repos, if there are changes
sivas: unit test
ed: aren't these already run by CI
ed: we have a tested by flag
ed: patrick and william are trying to get things unit-testable
brad: yep - so plenty of examples for improving
brad: open call for topics?
susan: can we talk about icons
ed: sure
susan: accessibility is the motivation for svg
susan: eu has some laws (gdpr)
susan: eu has new laws coming for accessibility
ed: when you say accessible what do you mean?
susan: tabbing through a table header (screen readers, etc)
susan: same happens for buttons or icons
susan: accessible settings are based on ids in the html
susan: they don't relate to the image format
ed: once compressed fonts aren't that big
ed: we can tree-shake too
susan: so can glyph icons be compressed smaller than svg?
ed: they aren't necessarily smaller - both vector graphics
ed: the difference are negligible
ed: tree-shaking we get rid of the icons that aren't used
susan: and that is what is happening with the font icons?
susan: how many font icons are being used?
ed: a dozen or so?
ed: the big thing from my pov is those are really helpful because they are there and you can pull from them.  all consistent.  all responsive.
ed: if we go to pure svg icons, now you lose the ability to pull in new icons
susan: you need someone to create them, is that why?
susan: from a design view, the icons might not be consistent
ed: if the goal is consistency, use an oss icon pack
susan: they are limited, not consistent
ed: what was fontawesome missing?
ed: fontawesome accepts submissions
ed: I'm not saying we can't have any svgs
susan: we can use both - is that an ok compromise
ed: I don't think svg vs font impacts accessibility
ed: today the size issue is javascript, not icons or fonts
kurt: if ibm is creating a set of custom things that are made available and open source
kurt: we aren't differentiating ourselves with that
ed: svg vs fonts is orthonginal to library vs custom
susan: just wanting to set a guideline for new contributors
susan: when a user goes to a panel, it behaves the same regardless of who designed the page
ed: is there a difference today?
ed: the arrow is from glyphicons(serverinfo)
ed: there are others
ed: are we finding accessibility problems today with glyphicons?
ed: can we get them documented in a bug
ed: that makes it easier
susan: sure, I will have my team help open bugs with that
susan: should we meet after that?
ed: it will probably be an ongoing discussion
kurt: can we just use the existing icons?
susan: in Riot there was a discussion about propsing designs using issues
susan: just wanted to make sure Ed that you feel good about the answer to that
susan: these were designs we wanted the community to provide feedback on
susan: these are things that IBM is planning on coding.
susan: do you want changes made?
ed: I clicked on one that said 'you are not an IBM employee'
ed: I don't look until there is code
susan: it is more about do you have the right content
susan: is it an ntp server?
susan: we want a test button?
susan: those sort of things - are those ok?
susan: next week I do want to ask some questions about date and time
brad: please try on the mailing list first

brad: tsc report
brad: work in progress

brad: CoC update?
kurt: CoC was accepted
kurt: We already had most of the suggestions
kurt: go ahead and remove the agenda

jason/tom: ipmi-sel?
brad: tom what are you working on?
tom: moving some ibm specific things into the open power oem library
brad: should we remove?  can we make progress on IRC?
tom: we should be able to make progress on mailing list / irc
brad: anyone else?  otherwise removing from agenda
brad: going, going, gone - removing from agenda

brad: ocp update
brad: dropping from agenda
brad: ask questions on list
brad: going to drop this from the agenda

kurt: release update
kurt: we tagged
kurt: everyone should be testing on their platform
kurt: making sure release candidate is sound
kurt: freeze ends today
kurt: should talk in IRC about show-stoppers or defects found
kurt: should talk in IRC about how we tag a fix needing to be backported
kurt: how should we do that?
brad: went over the yocto process
brad: what do you think?
kurt: I’ll summarize:
kurt: developer submit to master as usual (HEAD:refs/for/master)
kurt: can also submit to thud (HEAD:refs/for/thud)
kurt: maintainer can ack/nack thud backport
kurt: or
kurt: developer submits to master only
kurt: maintainer suggests back port (or does it themselves)
kurt: lets try it

brad: open call for topics
andrew: infra workgroup
andrew: one action item was to bring to the community call
andrew: official upstream system ( reference platform)
kurt: requirement that reference platform be tested and how to do that

andrew: going over the meeting minutes from the last infrastructure workgroup meeting
andrew: reference platform(s)
andrew: its how the community verifies that changes work
andrew: it gives concrete meaning on what it means to be upstreamed
kurt: systems that haven’t been updated for this release
kurt: should we mark as deprecated?
brad: send a mail to the original submitter and the list a month after 2.6?
brad: let them know platform is being moved out of openbmc/openbmc?
kurt: I like an openbmc-contrib idea.  could stage things out
kurt: could also put new platforms here initially
ed: is there somewhere written down what problem we are trying to solve?
kurt: there was an email
ed: I remember that email
ed: It was around orchestration and management
ed: seems like we are adding more of that without solving any problems
ed: it doesn’t seem like it solves the problem
kurt: we have scaling problem
<long problem statement discussion from brad>
brad: I should reply to the know with my thoughts on the motivation for reference platforms

brad: tsc report
brad: work in progress

kurt/emily: code of conduct
brad: Kurt, any word back from TLF?
kurt: the code of conduct was approved!
kurt: was accepted.  a couple tweaks were recommended
kurt: will submit a patch hopefully today
brad: will leave on until patch is out there

jason/tom: ipmi-sel proposal
brad: will bring back up in January
brad: what’s new here?
brad: are jason or tom on?
brad: does anyone want to discuss this?
<no response>
brad: will bring back up next week
brad: if no reply, will drop from the agenda

brad: ocp update
brad: some emails from sai
brad: have a look

brad: open call for topics
kurt: cla reminder process
kurt: release coming up

andrew: release plan - critical fixes list, branch all repos?  criteria for putting fix in branch, testing plans
andrew: known issues with master
andrew: mapper problem (3450) - must have
andrew: webui - user management changes
andrew: gunnar what is up with that?
gunnar: having trouble with user management page on ibm systems
gunnar: could move back two commits
andrew: ipmitool
andrew: did the default ipmi user id removal get merged?
richard: things should be good now.
andrew: any other hot issues?
matt: do you care that fan control is broken on witherspoon?
andrew: we should fix that
matt b: I have a fix up
matt b: it will temporarily fix that
andrew: any other hot issues that should try to get in?
andrew: are we going to branch all the repos?
andrew: what are the criteria?
vernon: should be merged in master first
brad: I agree.  are there additional criteria?
vernon: additional testing bits?
kurt: no new function
maury: is there an expectation the failing testcases?
brad: no autobump on release branch
andrew: got most of the infrastructure changes done
andrew: can still run ci
joseph: question about release branch criteria
joseph: same criteria before/after release?
kurt: good question.  haven’t fully defined that
kurt: its really hard to get something backported.
kurt: back ports are typically very bad problems (security)

brad: didn’t get to yocto 2.6
kurt: its freeze week and release week.
kurt: it isn’t hard and fast as today
ed: what is left?
brad: anything with yocto26 topic
andrew: can a maintainer keep merging code?
kurt: maintainers should be selective with what is being merged
kurt: now is not the time to be merging major new function
maury: look for critical fixes
andrew: brad or kurt please send out an email?
brad: kurt can you handle the communication?
kurt: any other function that needs one last push?
ed: quick last push not sure but what about kvm?
andrew: don’t think we are there
andrew: but even then, there is a webui commit
ed: it isn’t stable
kurt: I just posted the gerrit link for yocto 2.6.

brad: anything else on 2.6?
kurt: today was the target for a 2.6 tag.
kurt: in the wg we agreed to a 3 day window of sanity testing
kurt: when we tag, will enter into 3 day sniff test
kurt: you pull and test on your platform
kurt: if we don’t hear back, we will branch and enter window until feb 4th
kurt: ->4th for updating documentation, etc
kurt: communicate

brad: cla process
kurt: everyone has signed
kurt: TLF has a tool
kurt: no specifics on how it works
kurt: going into the beta pool
kurt: have been working with TLF rep
brad: is this 3rd party gerrit?
kurt: I think so

brad: tsc report
brad: no meeting last week, no update

kurt/emily: code of conduct
brad: Kurt, any word back from TLF?
kurt: nothing back from TLF

jason/tom: ipmi-sel proposal
brad: what’s new here?
vernon: Jason out for the holidays
brad: will bring back up in January

brad: ocp update
brad: nothing new here
brad: will get Sai to send something out
brad: leave on the agenda for next year

susan: user testing for date/time
susan: confusion about 3 options for setting time bmc/host/both?
susan: Do we need to offer this?
susan: if we do, is the command line enough?
ed: works for me

susan: wanted to confirm last weeks decision of style guide updates
susan: would it be better to change it in the website first, then in the gui panels?
ed: don’t have a strong opinion on which order things are done
ed: important thing is that you aren’t cuasing work for others.
ed: made a style change, but the gui wasn’t updated

susan: where should the website go?
susan: docs, phosphor-webui?
susan: can we vote?
susan: any strong opinions?
ed: who would maintain?
ed: I don’t want to maintain
ed: I don’t have experience maintaining a style guide
susan: what to do then?
ed: propose a maintainer for the style guide
susan: this was invented so the style guide could be open
susan: should we just go back to style guide in envision?
ed: that works for me
susan: ok, lets do that.  we can revisit if needed.
brad: this was a communication fail
brad: it seems like we’ve been talking about this for a couple weeks
brad: now we are just scrapping the idea?
brad: what happened?
susan: the right people were not at the design team workgroup meetings
brad: ok

iffy: went through a proposal for internationalization
brad: these kinds of things should go in development best practices?
ed: yes
ed: ng-translate defines a pattern for doing translations
iffy: yep, I have that in my slides
ed: this all sounds great to me!
iffy: ng-translate?
ed: I think so yes.  I’m not aware others.  ngtranslate seems to be the standard.
ed: do you have a plan for translating things that are based on the bmc?
ed: the sensor names for example are coming from the bmc.  how to handle?
iffy: haven’t thought about this.
brad: what about our dbus api?
ed: on other bmcs they don’t translate
ed: vocab for sensors aren’t that big
ed: should be ok

minutes from susan:
It was agreed that we will NOT create/maintain an open-style-guide-website. For now,
the IBM Design team will continue to maintain our InVision style guide, and review GUI
panels for consistency. Ed (Intel) and Gunnar (IBM) will remain maintainers of the code.

brad: tsc report
brad: tsc met, minutes are late, will do this week
brad: started community interaction discussion
brad: good discussion

kurt/emily: code of conduct
brad: Kurt, any word back from TLF?
kurt: have not heard back from TLF.

jason/tom: ipmi-sel proposal
brad: what’s new here?
brad: anything to talk about?
vernon: james has been slowly getting stuff in the intel-oem layer
brad: leave it on?
vernon: leave it on for next week, and revisit then
brad: tom or deepak - anything to add?
deepak: handling moving things to an oem library
deepak: should help with keeping the sel design common

susan: style guide proposal
brad: still needed on the agenda?
iffy: susan is still working on it

brad: ocp update
brad: nothing new here
brad: will get Sai to send something out

adriana: redfish and virtual media
adriana: putting the api - redfish virtual media schema
vernon: not terribly familiar with redfish.
vernon: send an email?

andrew: reference hardware
andrew: what should the community upstream supported hardware be?
andrew: minimum set?  for testing?  maximum function?
andrew: some ideas - ibm has upstreamed the most software.
andrew: from a reference perspective for IBM it is the Romulus system.
andrew: maybe a Raptor system?
andrew: raspberrypi?
andrew: a variety of servers are supported.
andrew: would be good to have coverage on all three
adriana: trying to test hardware only?  or qemu?
andrew: end goal is to make sure function works
vernon: would be awesome that wanted to get a system upstream had the possibility to?
brad: can you put the patch in IRC?
vernon: a futures patch to host-ipmid
brad: could that have been handled differently or better?
james: downstream we would do a bitbake of ipmi
james: is there a way to test one repo?
andrew: yeah, sure
brad: how often does sstate get cleared?
andrew: older than a month

susan: proposal of a style guide went out
susan: one person from Intel thought it was fine
susan: SG addition - new behavior
susan: would communicate to the community (irc, mailing list)
susan: would make an issue to change in phosphor-webui
susan: SG change
susan: would communicate to the community (irc, mailing list)
susan: would make an issue to change in phosphor-webui
susan: would make an issue to change in gui itself
joseph: how does the website review work?
susan: the thinking is it will be the     same as everything else?

susan: where to put the website?

brad: tsc report
brad: tsc met, ratified code of conduct
brad: next meeting is 12/11
brad: still targeting january for community roll-out of tsc interation process

kurt/emily: code of conduct
brad: signed off by TSC.
brad: kurt, how is it going with TLF?
kurt: sent to TLF.  Have not heard back yet.  Asking about CCLA tooling.
brad: about ccla tooling - I have a note.  will forward.
kurt: lets leave on agenda until next week.

jason/tom: ipmi-sel proposal
tom: jason responded to email
tom: he is coming up with a proposal for fixed sensor numbers
tom: we are heading in the right direction
vernon: we are headed in the right direction.  but it is moving slowly.
vernon: considering pushing to the intel repo.  until the rest of the project can catch up.
brad: how does that work exactly?
vernon: can install our handler first.

susan: style guide proposal
brad: any resolution here?
susan: two directions being discussed
susan: hosting through docs and phosphor-webui
susan: not so much of a split where it would be split
susan: we can talk this week wednesday or next week.
susan: anyone that wants to participate can come wednesday
susan: the envision flows are now interactive
joseph: its like a design preview?
susan: the style guide itself?
susan: these are flat jpegs.
susan: what we are showing wednesday is the website from a navigation standpoint.
joseph: posted 4 gui based reviews for ldap and time server
joseph: are those based on function we have or are going to have?
susan: it depends.
joseph: as a review it would help to know what function is present and what is blue-sky thinking.
susan: would like some feedback on the envision stuff.

andrew: infrastructure workgroup kickoff wednesday
brad: how’d it go?
andrew: it went well.  had some action items.
andrew: get a facebook system added to CI.
andrew: can remove from agenda.

vernon: default ipmi credentials
richard: I sent an interim patch to live for a few days
richard: once CI infra is done,  patch can be removed
vernon: waiting for comments from the ipmi maintainers
vernon: who is doing the CI change.
richard: I sent a patch.
vernon: can remove from agenda

gunnar: ocp update?  Track?
gunnar: saw TSC meeting minutes.
brad: have a look at sais note and respond if you are interested in OpenBMC @ OCP summit.

sivas: test sign up for community release
sivas: just a reminder.
brad: just a reminder - please join the workgroup meeting today
maury: have two rest server implementations
maury: in general we want to move to bmcweb strategically
maury: impression the community wants to do that too
maury: to do that, we have to update a number of functions that we already delivered
maury: there are issues on that front
maury: the question comes up  - do we merge the previous functions with the old ?
maury: or keep going towards the longer term goal
maury: may not be able to make the community release
maury: that is the general discussion we’d like to have.
kurt: actually we could discuss now.  this is a technical meeting.
… long conversation about ldap and certificates implementation
… the plan condensed on not including ldap / cert support in the 2.6 release
maury: will review this plan at the release planning workgroup
joseph: should send a note to the list to solicit problems?
brad: who will send out the email?
maury: we’ll find someone.
maury: will also discuss at release planning workgroup.
kurt: if you know of content that isn’t going to make it, let everyone know.
joseph: need to know which work items are being done so we can run test against them.

brad: holiday meeting schedule
brad: thinking about canceling the 24th and 31st.  sound right?
… many affirmatives
brad: ok, will cancel 24th and 31st
brad: have a good week

brad: tsc report
brad: tsc did not meet 11/20
brad: tsc will meet 11/27, but won’t be ready
brad: new target will be end of January
brad: will keep posting minutes at a minimum
andrew: did you get a submission in for OCP?
brad: tsc consensus was that brad will do a single OpenBMC talk
brad: brad to submit this talk this week

kurt/emily: code of conduct
brad: merged!  thanks!
brad: thanks emily!
kurt: yep, merged
kurt: I volunteered to make sure it is properly approved per the OpenBMC charter
kurt: the charter says the TSC will adopt a new code of conduct.
kurt: we can assume the review and approval was satisfied by the gerrit review
kurt: not a bad idea to have the tsc confirm that
brad: I will add that to the tsc meeting agenda
brad: will leave on agenda until sign off from TSC and TLF

jason: ipmi-sel proposal
brad: are we making any progress?
jason: I think we’re stuck
jason: we’ve diverged on ipmi sensor infrastructure
… long technical discussion between Tom and Jason here …
brad: jason - the expectation would be that you propose an abstraction point
brad: implement your side of it
brad: and we will implement the OpenPOWER side of it
jason: sel - we seem agreed - just need to fix sensor mapping
jason: have some ideas on sensor mapping
brad: how will you submit that idea?
jason: code review and will reply to Toms thread.
brad: are we still stuck?
jason: no, I think I have a path forward now

susan: style guide proposal
susan: not much to add - just a refresher
susan: effort to get consistency between Designers
susan: eds initial reaction was that we should not do this
susan: ed might be interested in GitHub pages, but don’t want documentation in different places
susan: these seem like different types of documentation
joseph: when coming into OpenBMC, you try to figure out where the docs are
gunnar: is there is enough in the style guide that it warrants its own repo?
susan: yes.  it will grow
brad: what are we doing different here?
iffy: GitHub pages - looks like a website
brad: I think we need ed to make progress
susan: I will reach out on IRC
brad: I will reply with an idea about generating our current docs in a GitHub pages site.

andrew: infrastructure workgroup kickoff wednesday

vernon: default ipmi credentials
brad: can automated CI be done without them?
richard: yes - can use an unauthenticated channel like -I dbus to set the password
brad: can someone help Vernon get the automated test updated?
andrew: I will show Vernon what script needs to be updated.
brad: will touch on this next week, make sure things are progressing.

brad: call for topics
brad: didn’t get to call for topics this week.
brad: please feel free to add to the agenda in the wiki.

brad: skipping to the front with meeting next week topic
brad: any feedback?
andrew: probably good to cancel
brad: going to cancel next week

brad: tsc report
brad: was able to get started on a process proposal
brad: target presenting to TSC 11/20 (if there is a meeting)
brad: also posted minutes, agenda from the last meeting
brad: will leave on the agenda until the process has been presented on this call

brad: susan, design team update?
brad: susan is on mute
joseph: good discussion
joseph: good process to review the designs
joseph: adding something to release planning spreadsheet, to find issues the design team is working on
brad: anyone not from IBM that can agree it was a good process?
gunnar: richard was there
richard: need confirmation from Ed on envision
richard: we should allow others without envision to submit patches
richard: but discussions are going in a good direction
joseph: posted a link to Kurts release planning in the webex chat, line 14 is gui stuff
brad: can I take off the agenda?
gunnar: yeah
joseph: yeah, we can revisit if need be
brad: ok, will remove from agenda

brad/gunnar/ed: redfish login
brad: can we continue to defer this for a bit?
ed: some of the webui patches moving to redfish are going to back up.
ed: making it a configuration option, supporting both for some amount of time could work
brad: can we do that gunnar?
gunnar: it is some work, and is it worth it?
gunnar: how far out are we?
matt: not this year
brad: what is the work exactly?
ed: there is a config file?  not much work?
gunnar: a little bit of work to put it in the layer
brad: remove from agenda?
ed: yeah, sounds good

kurt/emily: code of conduct
kurt: making progress, getting reviews.
kurt: everyone should go review.
brad: will leave on agenda until merged

jason: ipmi-sel
brad: has your pull request been merged?
jason: it has gotten some +1s…so could be ready?
tom: now we have mapping within error log entries to ipmi
tom: how to get back to feature parity?
brad: tom & deepak need to have at least one real-time conversation with Jason this week.
brad: tom & deepak to understand Jason’s design, identify our concerns with it, and work with Jason to identify something that works for everyone.
brad: will check back next week

brad: call for topics
susan: question about style guide
gunnar: 30 second update on the tag

andrew: infrastructure workgroup
andrew: ad-hoc thus far
andrew: more companies would be good - just ibm and facebook thus far
andrew: initial goal is to document what we have
andrew: should I do a poll?
brad: I did.  What happens is the time doesn’t end up working for a lot of people.
andrew: this is your chance to make a change in the infrastructure

susan: considering making style guide open
susan: considering a new repo
susan: wanted to get more feedback
ed: what is the impetus behind the change?
ed: we have a process.
susan: the style guide for the gui is a flat mock-up
susan: you can’t interact with it, can’t copy code snippets
susan: any change would have to go to the IBM design team
ed: why would we not just check it in with the existing documentation?
ed: why are we creating another documentation repo?
susan: if we had a style guide repo it would be easy to find?
gunnar: agreed, my preference is it would go in the docs repo or webui repo.
susan: can we still do git pages like that?
gunnar: yep
brad: what are git pages?
ed: is it like makedocs?
gunnar: yep
gunnar: what is the format of the style guide?  scratch pages?
susan: right now they are sketch pages.
susan: delay to next week?  could show examples?
ed: could we get something written up on the mailing list?  link to ideas, tool we are going to use?
susan: sure, can do.
ed: swapping things out - please explain why?
ed: two things going on?  make a style guide.  move docs to git pages.
joseph: gunnar - is this just a bunch of static webpages?  or is it more complicated
gunnar: derrick talked about this in the design workgroup.  scratch format not supported by github

brad: tsc report
brad: no time in last two weeks - nothing new to report
brad: new outlook would be two weeks from tomorrow
gunnar: can yo1u send out minutes tomorrow?
brad: I will ask

susan: design team workgroup update
gunnar: meeting wednesday
joseph: gunnar pointed at me to a webui issue which pointed at envision
joseph: this seems workable.
gunnar: I think this is Susan’s proposal.
gunnar: Will be discussing this wednesday.

brad/gunnar/ed: gui - redfish login
gunnar: we’ve held off on merging redfish into the gut
gunnar: Ed agreed to hold off until we move to bmcweb and mapper
gunnar: Ed that ok?
brad: don’t see Ed on the call - we’ll ask next week
gunnar: Matt are we getting closer on bmcweb/mapper
matt: still some things in enumerate method that are funny
matt: tried to run robot regression
matt: Ed did say he’s looking at some things.

kurt/emily: code of conduct
kurt: some updates - email sent
kurt: just waiting for review at this point

jason: ipmi-sel proposal
brad: was there any activity on your pull requests?
jason: not much movement other than phosphor-logging
brad: will ping deepak and tom on the reviews
brad: jason could you set the topic?
jason: three are set to ipmi sel
brad: will bring back up next week

brad: call for topics
gunnar: what is the status of tagging
brad: meeting time
jason: sel automatic monitoring automatic setting default
joseph: two wikis that might be useful
joseph: review for how bmcweb is configured

brad: tag schedule looks ok?

richard: late for India
kurt: we could try an irc meeting
brad: would you advocate for not having a dial-in?
brad: will bring it back up.
kurt: checkpoint topics should just be discussed in irc
kurt: workgroup reporting makes sense on the phone
kurt: alternating call time
gunnar: vote with a dial-in
brad: why?
gunnar: better participation with a dial-in meeting
gunnar: some problems are easier to communicate with talking rather than typing
nancy: agreed
joseph: could try adding an irc meeting - not as if we have to do one or the other
brad: more going on that how easy it is
kurt: meetings are quicker and more efficient
nancy: I find it confusing who is talking to who
kurt: that is an etiquette problem
gunnar: this is a lightly attended community call
gunnar: worry that we would have five people talking in irc
kurt: that is the problem
nancy: harder to convey certain nuances in writing
gunnar: agreed
kurt: it boils down to what people get used to
jason: I’ve seen a virtual call for topics, sometimes they get discussed and resolved before the meeting even begins
kurt: if anyone wants to try it - use #openbmc-meeting
james: is that in docs?
kurt: I will fix that

jason: the sel-daemon has an interface for manually logging sel events
jason: automatically monitor for events (threshold sensor events)
jason: proposed to move to meta-phosphor and enable by default
jason: which default for automatic monitoring?

vernon: like the idea of minimal default config
vernon: good docs explaining what you can turn on
ben: can we have minimal and full?
brad: I like this idea

joseph: I pitched my two new wikis on IRC
joseph: collect terminology, disambiguate acronyms
joseph: trying to find a way to tie designs together
susan: I noted this idea on the agenda for wednesday.
brad: this is design with a lower-case d I think
gunnar: I think a wiki makes sense for things that don’t need review
gunnar: for things like acronyms, etc, don’t these need a code review?
joseph: ok so a review then.
susan: design team working on a style guide
kurt: each workgroup can have its own glossary
kurt: don’t want everyone defining their own version of what someone means

Community Call

10/29

brad: needs to propose a tsc process to the tsc
brad: was anyone at the design team workgroup?
gunnar: met last wednesday - special one off meeting
gunnar: talking about how to do reviews
gunnar: taking about how to work together
gunnar: interest from ibm/intel on creating new pages, moving to redfish, etc
brad: will leave on the agenda to make sure we get a good process

gunnar: matt made some progress moving to mapper/bmcweb
matt: want to run robot test suite on it
ed: ObjectManager might be getting filtered out
brad: lets checkpoint next week, leave out Redfish for now

brad: how is the code of conduct going?
emily: thanks for volunteers
emily: the review - contributor covenant vs linux foundation
emily: should we use LF version with our escalation path?
kurt: agreed
kurt: will take as a todo to try to find better public landing page for the project
kurt: so we can make this more prominent
gunnar: copying TLF and then tweak it some?
emily: really just going to change the escalation path
kurt: make that available in docs directory?
emily: agreed

brad: did anyone review your work?
jason: emily looked at ilmi change, adriana looked at phosphor-logging change
brad: deepak do you have any concerns?
deepak: nope
jason: only other thing is concerns about remaining dbus error logging objects?
jason: how to access those?
brad: can the power esel oem command just add a new call to the new addsel api?
deepak: that should work
brad: deepak, can you make sure Tom has a look at what Jason is doing, understands it, and make sure we don’t break
deepak: sure thing

ed: doesn’t care when the tag occurs
jason: wait for renames to finish
anoo: do people care about the freeze before the tag?
brad: is anyone planning on doing any testing for this tag?
gunnar: are we going to do any?
brad: yes
brad: will send a plan to the list

brad: call for topics
ed: moving to dbus broker
brad: found a bug, will fix

brad: made some new repos.
brad: any other repos pending
brad: will carry tsc questions to the tsc meeting tomorrow
brad: update api proposal?
deepak: did discuss at hackathon
deepak: came up with an alternative - autogenerate a create method?
brad: urgent?
deepak: can work on this in the background
deepak: are there other examples of dbus projects?
brad: umm…freedesktop
brad: can I remove from the agenda?
deepak: sure, will bring it back if necessary

sivas: what should be the forum requirements, design, test cases for function?
kurt: was the question test-plan documentation?
sivas: design document with requirements and use-cases
kurt: the design template
kurt: the test plan can be referenced
sivas: where will the documents be?
kurt: the docs repo for now.
kurt: there could be and should be docs in the individual repo
kurt: set the topic to design
jason: subfolders in docs repo?
kurt: whatever works - we can change it?
brad: can I remove from the agenda
sivas: sure
kurt: please! write a few sentences around what you are working on

brad: how did the workgroup go?
susan: got into two rabbit holes
susan: rabbit hole 1:
susan: Intel is doing gui design work, has a FED team
susan: Need to figure out how to connect IBM and Intel teams
susan: rabbit hole 2:
susan: how to review gui mock-ups with the community
susan: design is used broadly - visual team is different than the backend team
susan: what is design?
susan: process is cumbersome - design doc in git and push that document out to gerrit
susan: what is the expected purpose of the design doc
gunnar: the workgroup meeting this coming wednesday?
susan: this week is about the two design teams working together
susan: the following week is about process
brad: will leave this on the agenda to check in
ed: are there meeting minutes?
susan: they are in the wiki

ed: wanted to chat about moving over to doubles
brad: go for it!

brad: gunnar, what are we talking about?
gunnar: moving IBM systems to bmcweb
gunnar: moving the gui over to redfish
gunnar: andrew said a few weeks ago had tried out nginx removal and it didn’t work?

ed: moving webui over to redfish is orthogonal
ed: is there a proposal for a sane path through the minefield?
gunnar: can we hold off moving to redfish auth until we find a sane path?
gunnar: what is the next step exactly?
ed: we have to find out what is broken
brad: leave on the agenda?
gunnar: yes

brad: open call for topics
jason: ipmi-sel
kurt: code of conduct

jason: there were concerns in the last discussion in losing out on dbus errors and dbus log objects
jason: pushed a change to phosphor-logging, new template specialization
jason: please have a look
brad: does everything up for review provide the complete picture

kurt: Emily pushed a patch for community code of conduct
kurt: it is a valid request
kurt: urge everyone to go have a look, get involved
emily: read the mailing list discussion and the LF CoC

brad: next week?
emily: yes please
ed: is there a summary?
emily: I did a summary on the mailing list

brad - ipmi sel - I have to make repository

joseph: test workgroup, proposal was incremental improvement for how openbmc collaborates when we make designs
joseph: that happens today on the email list, docs repo review, or in code review, or even in testing
joseph: a lot of information is being lost
joseph: proposal - part one - identify what problem is trying to be solved
kurt: we have a design template, feel free to use that
brad: why should I?
kurt: I don’t have to explain it 50 times
brad: I could just not answer?
kurt: cross participation will lead to a better design
joseph: It shouldn’t take much time.  Not asking for tons of information.
gunnar: couple paragraphs?  couple sentences?
kurt: going forward, most content will _have_ to have design docs.
kurt: but for now, a couple sentences will work
susan: identify the problem which came up with a design.
susan: thought there was a problem statement when we open an epic?
brad: no, we don’t have a shared planning tool (beyond “the spreadsheet”)
andrew: matt spinler put out a design document for review on Gerrit
joseph: proposal - part two - two much detail.  set of requirements allows you to focus.
ed: is all this stuff in a review somewhere?
joseph: there is the design doc template
james: what is the diff between the readme and the design doc?
kurt: once we come up with a feature, the design doc gets filled out early.
kurt: design template was a stake in the ground - feel free to modify
kurt: lets give it a try, and fix what doesn’t work as we go
brad: ask the question - where is the design?
gunnar: the docs reviews are slow.  go look when you are added to reviews for those
sherad: reviews by nature cause focus on things that don’t matter
 
susan: design workgroup meeting - fyi
susan: first topic - process for reviewing and documenting designs
susan: second topic: heuristic changes to the gui (pretty minor - blue color overlap)

gunnar: what is the tsc up to?
kurt: boards are more closed - tsc is more open
kurt: this project tsc is more like a board?
kurt: maybe need a once a month meeting?

gunnar: how is membership determined?  how are people added?  how are people removed?
andrew: we need to figure out the openbmc branding.  a lot of people are saying they have openbmc system
kurt: there is an openbmc logo.  and a built-on openbmc logo.

No discussion but two topics raised for next week.

brad: sorry about agenda/minutes from last week
nancy: didn’t get as much participation as would have liked
nancy: some good work
nancy: split over two days made it hard
kurt: agree, maybe to a retrospective
kurt: fewer topics?
kurt: went well, could do better
brad: pair programming?
nancy: how would that work?
nancy: advance notice is key
kurt: this is the nice thing about an irc meeting
adriana: less topics or team formation
kurt: might need to be some education on using IRC
kurt: did not have a chance to get meet bot setup
kurt: how often?  quarterly?
nancy: sounds good
brad: taking off agenda

brad: ipmi sel daemon update?
ed: not much on ipmi sel daemon
ed: he’s working on it
ed: hit a segfault in journald
brad: will leave on the agenda for a bit

brad: update method update?
brad: resolve at hackathon?
deepak: yeah, sounds good
deepak: want it for rsyslog
ed/deepak/brad: a discussion around the update api ensued

susan: web ui workshop
susan: introduced herself and the design team
susan: talked about what the design team is working no
susan: mentioned the gui workgroup, every other week
ed: are the changes the ones coming through gunnar/rebecca
susan: yes and iffy
ed: are you going to be on the reviews?
ed: how can we get more engagement
susan: didn’t realize there wasn’t much engagement
derick: fed lead on the design team
derick: will make sure to do some testing
joseph: try to spoonfeed your audience
ed: like devtool modify + cherry-pick
susan: designs will not be code - they will be mock-ups
susan: any and all feedback is desirable

brad: we’ve struggled in cross-company collaborative design work
kurt: design template process is there.  you can use that to get review comments in gerrit
ed: envision is not free
susan: you can review for free
kurt: propose everything comes back to gerrit at some point
kurt: too many places to do reviews is bad
kurt: the community doesn’t use Github or epics for planning work
kurt: right now we use a spreadsheet
kurt: its not great but its a start
ed: could we say that when we open an envision proposal, we open a gerrit review
susan: derick and susan can try and come up with a process and propose next week

brad: hackathon?
ed: its going to be awesome!
joseph: how would video access work?
nancy: James was going to look into video
sherad: james working it - he will be back thursday or friday

ed: will we have a call next week?
brad: yes

kurt: kicked off fix-it last night
kurt: two full global days
kurt: if you leave, please stay in the channel
kurt: will have logging, working on meet-bot
kurt: overcommunicate, update frequently where you are
nancy: google team will be online in a few hours
brad: removing from agenda

brad: anything going on with ipmi sel work?
jason: like it better having the sel daemon monitor for events
jason: might add a bit of complexity into the sel damon
jason: can deal with it
jason: other path still exists
brad: other path is dbus api?
jason: yes
brad: ok to leave on agenda for a couple weeks?
jason: yes
jason: keep an eye on gerrit reviews for ipmi command changes
jason: current sel entries will go away
tom: where does the sel daemon go?
jason: haven’t found a home

vernon: changes have been committed, ok for now
vernon: moved further into async stuff
vernon: boost use header only - pull in boost::coroutines and boost::context
vernon: can I bring these in?
brad: can I remove from the agenda?
vernon: sure, remove it

deepak: update api proposal
brad: should we talk now?
ed: what problem does it solves?
ed: you can’t do individual properties?
deepak: the other example is rsyslog server configuration
deepak: ip address and port
deepak: the application can react after both properties have been set
ed: just a way to avoid race conditions?
deepak: still need to support individual properties being set
ed: don’t want a non-standard interface
ed: what about an apply method?

brad: fix up update?
nancy: kurt sent out spreadsheet - 24-25
kurt: good involvement - good topics flowing in, please put ideas out in the spreadsheet
kurt: getting irc meeting setup

brad: ipmi sel update
brad: conversation on irc last monday - intel happy with plan?

brad: does anyone know what vernon wanted to talk about on boost / sdevent question?
vernon: have two io frameworks
vernon: best depends on the application
vernon: most Intel apps used boost-asio
vernon: there are three mechanisms for timers
brad: are you happy in short-term?
vernon: should we push in one direction or the other?

vernon: the plan settled along the lines of a ipmi-sel daemon watching for events
deepak: phosphor-logging client library to react to specific metas for ipmi

brad: open call for topics

joseph: documentation dash
kurt: release planning
deepak: new dbus interface proposal (multiple dbus properties)

kurt: fyi the release planning working group meetings are bi-weekly.
kurt: start next monday.  No call today.
kurt: the planning meetings are now open to the public.

joseph: documentation.  how do we get better?
joseph: tried to write it and ran into two problems
joseph: added a new wiki called documentation dash
joseph: problem 1 - development owners don’t have interest or time to write the documentation
joseph: problem 2 - others try to write it later, and they don’t have the required information 
joseph: documentation dash
brad: target for the first one?
joseph: external interfaces come first
joseph: try and use ideas from this for the fix-it or hackathon
kurt: yes - this is great for fix-it and hackathon
brad: does anyone think this is a bad idea
brad: is anyone concerned about resource for anything?
sharad: support the idea
sharad: not sure how easy or hard it will be
nancy: it can’t hurt!
joseph: the ask is really a couple hours from the lead developer

brad: vernon, any chance you can speak to Eds concerns with the Update API proposal from Deepak?
vernon: good idea on the surface, have not spoken with Ed
brad: lets keep discussion going on the list, will checkpoint next week

brad: that a wrap, see you next week

brad: fix it update?
kurt: talked with Nancy - kurt will be helping get it setup.
kurt: fix it v-sprint
kurt: details will be coming in an email
kurt: two full days to ensure global participation
kurt: Google fix it week will be the week before
kurt: OpenBMC part will be 24-25.
kurt: Will send out call for topics in the coming days
brad: how do people like the new workflow?
andrew: its a pain, but understand the need
andrew: probably better to have it not immediately opened against openbmc/openbmc CI
adriana: merging into openbmc/openbmc cron job maybe 30min to reduce CI load
andrew: try a squash
brad: is this urgent?  Can we keep discussing on it?
andrew: yeah that is probably fine

brad: open call for topics
brad: ipmi add sel?
vernon: curious about direction sdevent vs boost asio

brad: described the proposed view
brad: bmc applications?
ed: yes
brad: will we do the same for redfish?
ed: its just a wrapper around syslog
brad: can we add an a PEL call?
vernon: is there a pel for every error log?

nancy: OpenBMC fix-it
andrew: space contstraints update
brad: openbmc/openbmc restructuring update

brad: openbmc fix it - come back if Nancy joins
brad: space constraints update?
andrew: working on Eds suggestion of LTO
andrew: also working on removing python pieces ( removing py/pyc? )
brad: can I remove from the agenda?
andrew: yep
brad: gave a restructuring update
anoo: should we resubmit old ones?
brad: sit tight on old ones.  I will follow up with further instructions.

new business:
richard: pam recipe
brad: why is pam in initramfs in the first place?
richard: allows recovery login
brad: do we need recovery login?
brad: the crux of the issue here is fixed 4MB kernel partitions.  We need to engage our kernel maintainer.
brad: I will send a note to try and engage Joel on this.

nagaraju: review update for new interface for ldap config review
nagaraju: just wanted to remind richard of review comments
richard: I made some comments.
tom: use getgrpnam to figure out which group ldap user is a part of, and have a role associated with that group
richard: really just looking for some more documentation

brad: private emails
brad: please don’t do it.  please copy the list. same goes for irc

richard: pushed lots of changes related to user management
brad: tom and ratan to look as soon as possible

nancy: nothing to report on openbmc fix it week - week of sept 17th
nancy: waiting to get with Kurt

anoo: next week is holiday in the US do we have a call?
brad: no call next week due to the US holiday

sivas: did the test workgroup convene?
andrew: sivas rescheduled, and moved out this week
brad: can I remove from agenda?
andrew: yeah
joseph: there is a wiki with meeting information

kurt: irc rollup / community member presentation update?

nancy: OpenBMC fix-it?
brad: leave on agenda?
nancy: sure, it doesn’t hurt

andrew: how to screen bugs?
brad: was this discussed
andrew: yeah, not much input
brad: action plan for this agenda item?
andrew: will start screening
joseph: better set of documentation is needed to figure out where to put bugs
brad: agree and disagree

brad: around the room
andrew: space constraints

andrew: lots of new function, reaching limits
andrew: are others having this issue?
joseph: can you give some background?
adriana: gave some background
andrew: do we -Os?
ed: I think we do -O3
ed: LTO is where we get real improvements
brad: how about an analysis
ed: don’t think we have a problem - don’t pull in nginx, don’t use phosphor-hwmon
andrew: will open an issue for enabling LTO, enable builds without python
richard: is there a temp solution?
brad: will add to the agenda for next week to check-in
andrew: time permitting I will look at yocto features wrt size

old business
brad: will leave test workgroup formation on the agenda until formed
brad: anything more on irc
maury: was going to put something together
kurt: was looking at irc clients
kurt: will roll irc client info up
andrew: what is going on with irc
kurt: getting spam everywhere
brad: will turn off and see what happens
brad: what do you have in mind kurt for IRC rollup?
kurt: trying to finish modifications to the ‘good community presentation’
kurt: will start with an email

new business
deepak - hosting webui from nginx
nancy - google is doing another fix it week, hoping to work with Kurt
james - removing scale from sensor interface
joseph - security working group



brad: deepak take it away
deepak: have a patch on gerrit
deepak: simpler to host content with nginx
deepak: to do this need try_files, gzip_static
deepak: a workaround exists
deepak: anyone know of a better way of doing this?
deepak: where is the right place for this?
ed: if your browser doesn’t support gzip, you get an empty file
deepak: check client headers, another gzip module exists?
brad: can we untangle redfish from the webui

joseph: quick update on security working group
joseph: meetings are open to all now, created a wiki
joseph: call-in number has changed, no using a google hangout number
brad: you have some old documents
joseph: looking to get security response team review
brad: gunnar please make sure expectations are apparent

james: most daemons convert to double
brad: I think phosphor-fan-control uses integers
james: two dbus calls are required to get the threshold scale
sachit: loss of precision?
james: no
brad: what is the urgency?
james: no rush, just want to work on the upgrade path

Old Business

sivas: where is the test plan located? What are the project planning tools going to be?
brad: sivas sent out some notes\
maury: do you have a list of participants?
brad: google polls work here too
brad: will leave on the agenda until the first wg meeting occurs

maury: not an expert - what do companies use, recommend
andrew: mess - riot/irc wasn’t working - conversation was missed
brad: riot network isn’t quite enterprise ready
kurt: multiline sends a link
kurt: the common denominator is an IRC client, can talk about a proxy
andrew: could see everything
andrew: we use slack at IBM, we like it
kurt: I was on the hook to do some research to setup a proxy service
kurt: chanserv - we could log
brad: where does it go?
kurt: creates a set of html pages
brad: freenode doesn’t host this then?
brad: does anyone want to do something different?
maury: interested in what non-ibmers do?
richard: using riot, thought it was nice, internally we use slack
brad: weird to use slack on an open source project
maury: what was the issue with riot vs irc?
maury: lets make sure we are being efficient
kurt: there are web interfaces for irc
kurt: maybe do a survey of irc clients?

brad: lets make a list of pain points
nancy: too many knobs to turn
andrew, nancy: history is important
andrew, nancy: lots of channels, too many threads
kurt: multiple conversations at once are a good thing actually
matt: notifications?
kurt: yes most clients can do this
joseph: learning curve - documentation could be useful here
brad: is your presentation public yet?
kurt: yes very soon.
maury: add to your list of clients - what platforms are supported
kurt: sure thing





New Business
Around the room


brad: call for topics

brad: out of office
nancy: don’t care - ok skip
sachit: light of topics vote to skip a week
brad: skip this week and result the 6th

joseph: formation of security response team
brad: list created, please hash out who goes on the list in the security workgrouup
joseph: create a webpage to submit problems


maury: get a feel for best practices around communication? riot/irc/etc?


brad - how to merge code faster
james: richard are you being blocked by anything?
richard: review response is slow
sachit: dedicated hour is good

brad - what is going on with this testing work-item?
sivas - put it under release management?
kurt - initial discussion we could have there, should split off later
sivas - will set up a test workgroup
brad - can I take off the agenda

sharad - leave on the agenda one more week?
sivas/brad - sure thing.  checkpoint next week

tom - discuss password management
joseph - private way for security reporting?
sharad - overall feature plan?
maury - get a feel for best practices around communication?  riot/irc/etc?
brad - how to merge code faster

tom - what will the interfaces look like for changing the password?
richard - gave an overview of sssd
ratan - had some questions about how sssd works
ratan - how to configure things?
richard - haven’t defined that yet
ratan - ipmi with ldap?
ed - there is a flow and a way to do it
brad - leave this on the agenda for next week
brad - please work on having these conversations on the list or irc

joseph - security reporting
brad - replied on list
joseph - who gets to be on the list?  what are the expectations?
sharad - tsc, or one person delegate?
brad - sounds good?
brad - joseph please take to the security working group and get it ratified there
joseph - creation of list, who is on it
sachit - we need some kind of notification mechanism
brad - will leave on agenda to checkpoint security working group ratification

brad - kurt, can you describe the planning workgroup?
kurt - got a group of people together that wanted to talk about release content, dates, etc
sharad - aware of the workgroup, and James attends
sharad - question is more general - how do we collaborate more generally?
kurt - it all boils down to communication
sharad - Intel shared a plan with what we plan to work on in next couple of quarters
maury - kurt - have we gotten feedback in that workgroup?
maury - how will the plan be communicated?
kurt - workgroup will have to decide - right now its a spreadsheet
joseph - what about GitHub?

maury - leave communication on the agenda for next week
maury - please come back next week having thought about what you think works for them

old business

brad - yocto 2.5 in, will pick up updates frequently
joseph - openssh was updated, we are using dropbear
kurt - last version of dropbear was ???

brad - lets revisit test workgroup next week, sivas not here today
joseph - slightly interested in testing workgroup
brad - watch the list

new business

brad - around the room

ed - gerrit/github integration?
kurt - release planning workgroup update?
andrew/gunnar - 2.3 tag?
joseph - security workgroup update?

ed - a couple repos, the github/gerrit sync doesn’t work anymore
brad, andrew to have a look at what is going on, will update on IRC

kurt - had a kickoff meeting, was good, was well attended
kurt - don’t have a lot of company reps on this call….
kurt - need everyone to go fill out the spreadsheet
kurt - please ping your rep to fill out the spreadsheet

andrew - got kernel update, yocto 2.5, test team testing

joseph - security workgroup update
joseph - documenting hw, fw (os), bmc team lifecycle, what a downstream team would do

ed - has anyone tried to boot bmcweb?
deepak - dhruv tried but got distracted

brad - sivas topic - tool choice - will be discussed in release planning group
kurt - agreed
kurt - test plans, results, etc - this covered by the test workgroup
brad - to put out a call for test workgroup formation
brad - leave this topic on agenda to see if a workgroup gets formed

kurt - a few replies to planning workgroup
kurt - please reply on preference on time or agenda items
kurt - will cover initial release process, and date
kurt - talk about design freeze, milestones, tools, etc
brad - can I remove from this meeting agenda?
kurt - can be brought up in community call as needed
brad - ok, will remove from agenda

james - did the poll
james - date will be october 9th to 11th
nancy - how many replies to the poll
james - 27 replies
brad - can I remove from the agenda
james - go ahead
james - watch out for registration
nancy - when might that be?
james - hoping for this week
kurt - could a rough agenda be proposed?  might help with travel planning
james - an agenda was starting to form on the list
james - will capture and send-out

brad - yocto 2.5 - really close
brad - will leave on the agenda

brad - call for topics
matt - inband code update

andrew - 2.3 tag.  just fyi, we are hoping to do a  2.3 tag
andrew - will freeze on wed/thursday
gunnar - will just quit merging bumps?
andrew - will be nice to quit repo commits too, maybe not required
brad - what is the ask?  test on thursday, raise and fix any issues
brad - when to tag?  end of the day friday?
andrew - sounds good.  stop merging wednesday evening
andrew - this is tentative
brad - this is a short term process
brad - there will be more time to test in the next iteration of the process

matt - we need to start looking at inband code update
matt - there was discussion on the list last november
matt - if you plan to put something up, would be good to know
matt - design thinking is oriented around virtual media
matt - will start the conversation on the email list
matt - possibly google has a host implementation?
nancy - it requires some oem commands
matt - what about the part that runs on the host?
nancy - planning on upstreaming that too - just a small c++ utility
matt - they were in the openbmc main repo?
brad - we don’t do that
brad - where are we stuck?
matt - review 9497.  would like to see this move forward
james - do we have a group extension?
brad - check with Tom on what we have registered
james - will check to see how to get a group extension
vernon - recommend that anything is common use the common the number
brad - what should this repo be called?
vernon - openbmc-ipmi-fw-update?
james - works for me
james - have you looked at this?
vernon - no, will take a look
matt - sounds good.  the other piece? oem router? review 8210
vernon - I don’t understand the purpose of this.  the router that is there allows you to do this.  don’t need another router to sub-route.
brad - where is the code?
vernon - can I just toss code somewhere?
brad - I think gerrit is fine?
nancy - matt, can you reach out to patrick venture about oem routing?
matt - sure
brad - take off agenda?
matt - sure, will bring up again if necessary

ed - talk about sdbusplus asio patch?
brad - should be good

nancy - hackathon
brad - what went well
nancy - needed more organization
nancy - interest in another one
brad - yes
kurt - more chatter in IRC
nancy - agreed
brad - will do webex next time
nancy - was a whole week of fix-it
nancy - one day per quarter would be nice
kurt - very good success with multi day events - more interaction
brad - will drop from agenda

brad - kurt, how is roadmap
kurt - still having trouble
kurt - workgroup will kick off this week
brad - leave on agenda

brad - hackathon?
nancy - when is the registration coming out?
brad - where is the survey?

brad - open call for topics

nancy - would like to talk about gerrit plugin
ed - yocto 2.5 update
maury - update from redfish workgroup

nancy - looking into configurations for owners plugin.  can we try it?
brad - do we want a test repo?
andrew - test repo or test instance?
adriana - tools type repo?
brad - make a test repo
kurt - does openstack use the owners plugin?
brad - nancy can you find a project that uses the owners plugin?
brad - what happens after we make a test repo?
nancy - need access to the gerrit instance
kurt - looking at owners plugin, path based
brad - thats a good thing
kurt - acls for gerrit - assigned by tsc?
andrew - need a test instance
kurt - access list should be small
brad - replicate gerrit instance for development?
brad - lets table for a week
brad - two issues at this point - acls process for services (gerrit), and development gerrit instance?

brad - should have yocto 2.5 up for review this week

ed - work has been slow
ed - reduced down to two different implementation, bmcweb and go-redfish
ed - Michael won’t be able to work on go-redfish as much as he would like
ed - openbmc callbacks were removed from go-redfish master
brad - what was the motivation for taking it out?
ed - had to refactor, could not refactor the openbmc support
ed - met with Hari, sounds like IBM is hoping to making a decision on redfish stacks
ed - bmcweb, a bunch of reviews going on
ed - computersystem schema
maury - still trying to do an evaluation ourselves
maury - it will probably be a bit longer than two weeks for IBM to make a call
ed - if measuring performance use some of the newer patches
maury - how far along in terms of implementing services?
ed - wip - ocp basic server profile?
ed - upstream - account/thermal/chassis
ed - will post targeted redfish schema review roadmap to the docs repo
brad - can we take out legacy rest api?
ed - just flip compiler flags
brad - are you going to rewrite the boost licensed code in bmcweb?
ed - yes

sivas - discussion from moving from GitHub to Trello
sivas - test plan, where to put it?
brad - will add these to the agenda next week for a checkpoint

joseph - security updates are on-going
joseph - have some topics for next security workgroup meeting
nancy - will set up another

maury - 2 weeks from now is the week of the forth?
brad - no meeting the week of the forth

brad - fix it week looking good, have been some sign-ups
nancy - last chance to put anything down

kurt - little progress on release planning, focused on downstream work, hopefully starting next week
maury - have you received input?
kurt - no updates on spreadsheet
maury - general plea to update the spreadsheet with plans
kurt - was going to have a call or email thread that want to plan
brad - what happens if no-one does this?
kurt - work will still continue, but to reap the benefits of OSS you need to organize

james - request from Nuvoton to push out
maury - how far out?
james - october 8th timeframe
james - will look into 8th internally
kurt - maybe a survey?

brad - call for topics
tom - user management topics
gunnar - recipe clean up.  how normal are we?
ed - yocto 2.5 progress?

tom - how much development has occurred?
richard - have done ipmi user management, pam module
tom - what about pam_unix?
richard - gave brief pam overview
ratan - had some pam questions
richard - two pam ipmi modules
ratan - how to set password over rest?
richard - rest server calls pam
ratan - plan for bringing ldap
richard - yes, this is the first step
richard - will try to upstream pam module
brad - create a repo to hold pam module
ratan - why do we need a new delete interface?
brad - move errors to base xyz Object interface
richard - we can move to xyz

brad - gave an overview of how ‘conventional’ our bitbake metadata is(not very)

brad - docs process - drop from agenda - anyone can pick this up
joseph - security docs would dovetail into user operations guide
kurt - customer information comes from the product deployer
kurt - suggest a security workgroup, might kick off in a couple weeks

brad - fix it week?  Anyone planning on going?

kurt - busy with downstream things
kurt - intent is to start release planning workgroup
brad - did you get james roadmap
kurt - yes
gunnar - will we have one sooner for yocto 2.5?
brad - no
kurt - targetting november

brad - drop hwmon from agenda, been submitted to upstream
brad - drop ut framework - link seams

james - hackathon update, space limited < 100, sign up early
james - will put together a package on hotels

james - working on mapping CC model
brad - are we doing CC?
james - don’t care about CC
nancy - not useful for google
sachit - not familiar with CC
joseph - chose CC, familiar with it
james - want more clear, concise actionable
joseph - can you give an example where there is too much meat?
nancy - what does Intel do internally?
james - security architecture spec
brad - how does this work out?
james - what hardware features can be used maliciously?
vernon - debug features, etc.
brad - what is the problem statement?
joseph - CC covers everything, but not deep
joseph - protection profile, talks about product, users, threats, objectives
joseph - how the team has coded the requirements into the product
joseph - gave an overview of CC
brad - move to security workgroup?
nancy - will schedule a follow up meeting

gunnar - yocto 2.5?
brad - I am working this

richard - user management
brad - has anyone looked
ratan - I started looking
ratan - can we include ben in user management?
tom - how much development has happened?
richard - user management works pretty good
ratan - how can we contribute in user management?
ratan - can you open an issue?
richard - sure

brad - docs update - no update
nancy - a couple responses, not sure how many are interested.  going to put out a shared document for sign up
brad - a few ibmers might come
tom - had a discussion on the commit.
ed - optional dbus interface?
tom - should be ok.
tom - a long conversation ensued on how IBM uses IPMI/MRW
brad - take topic off the agenda?
tom - yeah

kurt - background: got task of what we are working on
kurt - sorry, was busy with some downstream things
kurt - created a spreadsheet - showing a bigger problem
kurt - looked at what is going on in Github
kurt - confused by what is going on in Github
kurt - question for community - what are we working on and how do we track that?
kurt - need project releases, move to a different tool than Github (jira, trello, etc)
kurt - created a free trello account, will setup a kanban page for a proposed release.
kurt - need list of people to add permissions for creating content
kurt - how do we start marching to a release schedule?
ed - how to avoid the yet another tool problem?  will IBM keep using GitHub?
ed - James is putting data together in a spreadsheet, should be done in less than a week
nancy - not a fan of github
kurt - starting with something that is free
kurt - short term - lets get what we are expecting to do in a spreadsheet
kurt - have a meeting with TSC on this.
brad - lets do every 6 months
kurt - will propose 6 months

oshri - will we write code without upstream?
brad - probably not?

brad - new framework - link seams
brad - leave on the agenda

joseph - no update with security framework documentation
joseph - james working on protection profile

sachit - brought it up, will post to the list.

joseph - question in review.
joseph - security assurance requirements
joseph - requires collaboration: devs + requirements + test and use
joseph - how much do we expect in terms of writing down security designs?
nancy - who is the dev team?
joseph - the team writing the code
joseph - we could probably make level 4
ed - how would we do that as a project?  as a whole?
joseph - secure defaults
ed - this could use a working group
brad - start the security workgroup please

ed - can we chat about sdbusplus + asio?

brad - nothing to report on documentation process

brad - anything to report on fix it week?
nancy - no

tom/james - sensor scaling
brad - tom and james aren't here, will leave on agenda for next week.

kurt - roadmap
brad - kurt isn't here, will leave on agenda for next week

andrew - made decision on domain, both domains go to same IP.  
andrew - get rid of xyz later.
andrew - will send a mail to list.  
andrew - remove from agenda

brad - what is going on with Redfish?
brad - I will follow up with Ed.
sachit - the workgroup is working through performance and size concerns
sachit - and cached vs non-cached
vernon - where are the bottlenecks?
andrew - consensus on implementation?
sachit - no, but some consensus on goals
brad - did not know redfish working group was still happening
brad - does anyone view this as an issue?
sachit - to ask the redfish team for more minutes

brad - asked oshri's question about hwmon
sachit - volunteered to have a look at the request to see if someone could have a look
brad - I will forward the email with the questions

sachit - cache/not cache - not ready…a couple weeks

sachit - memorial day?
brad - no meeting memorial day

james - security, additions
joseph - too detailed - this might not have been the correct assessment
james - presentation to community at some point.
maury - week after memorial day?
brad - James that OK?
james - yes
brad - 30 minutes?
james - yes

maury - kurts note about what is going on?
james - figuring out how to respond
sachit - will have a look

nancy - venture mentioned new UT framework?
brad - not aware of a new UT framework
brad - will move forward with patricks work
maury - will ask ARJ about other UT framework

james - hackathon update, looking for September, in Hillsborough

joseph
   - got started
   - objectives into requirements
   - protection profile modules (read only fs, admin access, etc)
brad - what do you need
joseph - find out what we have
joseph - more involement
sachit - do you want more general documents?
brad - is this project based or product based?
joseph - the method of documentation handles this?
brad - different profiles for for different bmc use cases
ratan - how do you handle per repo security requirements
brad - can I take this off the agenda?
brad - please avoid wasted effort - check with James.

new user guide - drop from agenda
security documents - nancy - drop from agenda

documentation
sachit - consistent form, where it is, lifecycle - lets come up with some guidelines
brad - anyone can add documentation - go nuts pushing process proposals to docs
kurt - go ahead and write a guide - we'll figure out organization later
sachit - careful with random documents all over the place
brad - will try to add some guidelines for review

nancy - fix it week - will send out a post
kurt - prospose topics on the IRC channel
nancy - proposed testing, docs internally

tom - generating sdr from dbus
tom - explained how sdr, scaling factors work on openpower
tom - explained a proposal from James Feist
tom - how do you get the blueprint
ed - get the inventory

kurt - project roadmap - next week

andrew - what to do with gerrit instance?
  run it on both?
brad - do whatever you want
nancy, ed - yeah, go for it

Old business:
deepak:  remote dbus:
still researching feedback, questions
tried systemd api - works
brad: will take off list
sachit - looked redis

user management
richard: from community point is design ok
brad: has anyone looked?  (crickets….)
richard: gave an overview
richard: is delete method ok?
brad: gave overview of why we did it this way
richard: ok, sounds good
ratan: had some questions about the design
richard: answered ratans questions
brad: asked Ratan to go look in Gerrit

roadmap:
brad: how do we share?  Do people care?
andrew: yes, people care
nancy: a spreadsheet?
sharad: start small, sharing with what others are working on
brad: why doesn’t GitHub work?
sharad: verbal should work better
sachit: formalize sprints in Github?
sharad: good idea, but not ready
nancy: teams, devoted to upstream
maury: interested in project planning.  start small.  knowing the next three months.
sachit: use what we use now?
sharad: tsc will know what is going on.
sachit: if we all go to one model - and then switch?  switching habits is hard.
nancy: propose it and see how it goes?
maury: kurt will propose a plan

new business:
joseph reynolds - new to the call, works for Maury.  security concepts document?
sharad, nancy - new user guide would be nice
nancy - security documentation
brad - what are we talking about with document
sachit - how are we going to do it?
sharad - yes, process also important, but how do we do it?
nancy - fix it week.  want to devote a few days to OpenBMC
tom - sensor model, scaling factor

overview:brad
welcome
community driven agenda
add topics to the wiki
around the room
will try harder to add agenda topics

agenda
around the room

topics
remote dbus (deepak)
sachit - interested
sivas - who is working on what
user management

deepak:
  recap - high level problem - multi bmc system
       possible system designs - multiple hosts, single hosts
       may need to cooperate (to power on for example)
       external clients interact with point of contact BMC
     solve the problem in different layers
       discovery, trust, rpc, clustering

   want to talk about rpc - propose dbus
   how to remote dbus?
   tcp sockets? - described in spec, support in libdbus
   ssl tunnel - systemd supports this.

sachit:
   what about auth?
this is the trust layer

for tunneling - independent of the transport

deepak:
    mirror dbus model to each bmc
    calls can be routed
    applications make same dbus calls as before

ratan:
   how does this scale?

deepak:
    same number of dbus objects are needed in either case
bus cost is 8x for an 8 sled
    
active - active cluster
   multi sled - single chassis
   
ratan
what alternatives exist?

sachit
 redis - oss memory replicator?

deepak
how to do for objects dynamically created?
error logs?
proxy dbus object managers
brad - too many error log objects anyway