Update failed from 4.0.0-22 to 4.0.0-26

Infos:

  • Used Zammad version: 4.0.0-22
  • Used Zammad installation type: docker-compose
  • Operating system:
  • Browser + version:

Expected behavior:

  • update docker-compose to latest version
  • init container updates whatever required
  • zammad starts

Actual behavior:

  • elastic search container failed on missing codec. loops in restart
  • zammad does not start

Steps to reproduce the behavior:

  • install 4.0.0-22 with docker compose
  • create a few tickets
  • update you install to latest (-26)

I have assumed that elastic search is new and not compatible with the existing data volume.

I re-created the data volume and now elastic search seems to start.

However, now the init container restart-loops after one minute. I thought that indices where re-created on every start?

Any idea what could be going wrong?

I think I know what happened:

I noticed that when pulling latest zammad I was not receiving the same image as when pulling the -26

repo.vpc.vtk.lan/zammad/zammad-docker-compose   zammad-4.0.0-26                 0cb919d2c69c        20 hours ago        604MB
repo.vpc.vtk.lan/zammad/zammad-docker-compose   zammad-latest                   997f598b9c95        5 months ago        615MB

How is it possible that I am not getting the latest image, but instead a 20 months old image?

Checking docker hub:

Then, what is wrong?

Most Probably the issue is with our Sonatype Nexus 3 Docker repo that acts as a cache from docker hub. But I cannot figure out why this is happening here, as we get regular updates for other images, such as gitlab, etc.

Any ideas?

I dont understand why but even after evicting the images from nexus cache I still get old zammad-latest images when pulling.

It does not happen when pulling directly from the docker hub.

Unfortunately I will have to pull specific versions until I find a solution. Perhaps Nexus treats :latest differently to zammad-latest.