Upgrade fails 2025.2.4 => 2025.4.0 (not enough time to complete migration of session to database) - Solved by removing redis sessions

[francescor]

Hi,
We've just upgraded from 2025.2.4 => 2025.4.0

In our docker swarm setup services "worker" and "server" where not able to finish the "Migrating Redis sessions to database" (reached 70% then the container was killed), so upgrade fails.

We still do not understand what was killing those services (we do not have healthchecks on those services), apparently 2 minutes after start.

We solved by cleaning up Redis sessions with this command

# enter redis running container/services 
docker exec -it authentik_redis...whatever  /bin/sh
# Bulk delete all cached sessions
redis-cli --scan --pattern "public:authentik_cache:1:django.contrib.sessions.cache*" | xargs redis-cli DEL

it was a quick and dirty fix we needed (users where waiting --- maintenance windows was over), not sure if it's a proper solution, but we could go on

[rissson]

Yeah that migration can take quite a bit of time if you have many sessions. Deleting those sessions or temporarily removing the healthchecks are both valid workarounds.

[francescor]

great, thanks!