Upgrade issue - zero length config.pb

Hi. I’ve inherited the management of some CheckMK infrastructure that I’m trying to upgrade to get to a supported version. One central and 9 x remote instances in a distributed setup, all running the docker image. Originally they were on 2.0.0p16 and were upgraded to 2.0.0p39 with no issues. Now when I try and upgrade to 2.1.0p49 the upgrade fails. The docker logs show the upgrade starting:

2025-10-14T12:42:18.929921987Z 2025-10-14 12:41:59 - Updating site 'ny_1' from version 2.0.0p39.cee to 2.1.0p49.cee...

and goes through all steps finishing with:

2025-10-14T12:43:05.161518422Z -|  30/32 Fix registered hosts symlinks...
2025-10-14T12:43:05.162137827Z -|  31/32 Update license usage history...
2025-10-14T12:43:05.261438584Z -|  32/32 Synchronize automationuser secrets...
2025-10-14T12:43:07.643787764Z -| Done
2025-10-14T12:43:13.545067046Z Generating configuration for core (type cmc)...
2025-10-14T12:43:13.621893702Z Starting full compilation for all hosts Creating global helper config...OK

“omd status” then shows nothing is running. If I then try and restart it fails with:

Starting cmc…Failed (Config /omd/sites/ny_1/var/check_mk/core/config.pb missing, run “cmk -U” and try again)

The “~/var/check_mk/core/config.pb” file does not exist, I just have a 0 length config.pb.new.

If I run a “cmk -U --debug”, I get “Configuration Error: Other restart currently in progress. Aborting”.

I saw something similar mentioned here: Werk #16407: omd update: Don't Delete "config.pb" During Pre-Update but am a bit stuck on how to move forward. Does anyone have any ideas as to how I can resolve or debug this further?

I did setup a test install and the same upgrade worked fine, so I’m assuming there may be something in the configuration of the live systems that is causing the failure.

Many thanks.

As a test, I did a backup of the site, installed the deb versions of CheckMK (outside of Docker), then tried the process again. Restored 2.0.0p39, all started OK, then updated to 2.1.0p49. This works OK and the upgrade completes and the config.pb file is correctly written.

Still no clues as to why the docker based installation fails the upgrade.

It looks like the problem was Docker version related. The host in question was still running Ubuntu 20.04LTS. After upgrading to 22.04LTS, the container upgrade to 2.1.0p49 worked as expected.