If Immich stops after a ZimaOS upgrade and the server log says getaddrinfo EAI_AGAIN immich-postgres, the immediate failure is name resolution between containers. It is not evidence that the PostgreSQL data volume itself is corrupt.
In the original 1.6.1 thread, renamed storage paths broke several applications, but MariaDB, Plex and Nextcloud recovered after their paths were corrected. Immich was different: PostgreSQL was healthy while immich-server could not resolve the hostname immich-postgres.
Read the Immich Error Literally
EAI_AGAIN is a temporary DNS resolution failure. The server was trying to reach a database host named immich-postgres, while the database container itself reported that it was ready to accept connections.
That makes Docker networking and service discovery the first layer to check. Do not start by deleting the database or recreating the photo library.
Verify That Immich Services Share a Compose Network
Docker's Compose networking guide explains that services on the same Compose network can reach one another by service name. Container IP addresses can change after recreation, while the service name remains stable.
Check that immich-server, PostgreSQL and Redis are attached to the intended shared network and that the database hostname in Immich still matches the Compose service name. Avoid replacing the hostname with a fixed container IP as a permanent fix.
Treat the Volume Rename as a Separate Migration Check
The upgrade also changed some storage names by replacing spaces with underscores in this user's environment. Correcting those host paths was necessary for other apps, so review every Immich bind mount as well, especially library, upload and database paths.
The Immich storage path guide and app data migration guide can help separate storage-path problems from network problems.
Recreate the Stack Only After Backing Up Configuration
After confirming the Compose file and paths, recreating the Immich stack can rebuild the intended Docker network and DNS entries. Preserve the database volume and Immich library paths; do not use a destructive reset as a networking test.
ZimaOS 1.7.1 later listed improvements to Docker network configuration and app migration compatibility in its 1.7.1 release notes. That is a reason to update, but the changelog does not specifically claim that this exact Immich EAI_AGAIN case was fixed.
Bottom Line
When PostgreSQL is healthy but Immich reports EAI_AGAIN immich-postgres, troubleshoot service-name DNS and shared Docker networks first. Then separately verify every renamed host volume path introduced by the ZimaOS upgrade.
