Node not starting, failing unit "ovsdb‐server.service" - gpillon/k4all GitHub Wiki

Troubleshooting: ovsdb-server.service Failed on Boot

Problem Summary

During system boot or when starting the Open vSwitch (OVS) services manually, you may encounter the following systemd error:

[systemd]
Failed Units: 1
  ovsdb-server.service

This typically occurs in K4all deployments using Open vSwitch as part of the network setup. The root cause is usually incorrect file ownership on the OVS database files, which prevents the ovsdb-server process from accessing them.

Root Cause

The ovsdb-server process runs as the openvswitch user. If the ownership of the database files located in /etc/openvswitch/ is incorrect (e.g., root-owned due to system provisioning or image changes), the service fails to start properly.

Affected files include:

• conf.db
• .conf.db.~lock~
• .conf.db.tmp.~lock~

Solution

Fix the file ownership by assigning them to the correct user and group (openvswitch):

chown openvswitch: /etc/openvswitch/conf.db
chown openvswitch: /etc/openvswitch/.conf.db.~lock~
chown openvswitch: /etc/openvswitch/.conf.db.tmp.~lock~

Then restart the service:

systemctl stop ovsdb-server
systemctl reset-failed ovsdb-server
systemctl start ovsdb-server

Then, reboot the node.

You can verify that the service is running correctly:

systemctl status ovsdb-server

Recommendation

If you're building custom images or provisioning K4all nodes programmatically, ensure these files are created or restored with the proper ownership during post-install configuration steps. This prevents service failure after reboots or updates.