postgresql init issue after upgrade about zammad-helm HOT 11 CLOSED

Does it work if you add "--set zammadConfig.initContainers.postgresql.securityContext.readOnlyRootFilesystem=false" to your helm upgrade command or set in your values.yaml file?

from zammad-helm.

Hello !

I have the same error trying to upgrade to chart 9.0.2 from 8.2.2

from zammad-helm.

No change with the readOnlyRootFilesystem=false in values, still the same error.

Note that I now have to set the following in the value or the data-chmod container fails with a "permission denied". Not sure it's related, but since most of the change in the chart are volume related...

securityContext:
  runAsUser: 0
  runAsGroup: 0
  runAsNonRoot: false

from zammad-helm.

Does it work if you add "--set zammadConfig.initContainers.postgresql.securityContext.readOnlyRootFilesystem=false" to your helm upgrade command or set in your values.yaml file?

Yes; that allows the postgresql-init container to finish starting. Thank you!

(Note; this was needed for the elasticsearch init; railsserver; scheduler; and websocket container as well to get it started but the custom brand logo is still missing)

from zammad-helm.

maybe adding the fix from zammad/zammad#4684 would help.
hopefully i can find some time the next days to test it.

from zammad-helm.

Errno::ENOENT: No such file or directory @ rb_sysopen - /opt/zammad/storage/fs/3bc5/897d/001f9/61170/e3344f4/877996c/8854fe4f3194c99a9943f8b7c7f380a0
getting this error, just a nightmare, cause changed directory

from zammad-helm.

I've created an pr which hopefully workarouds the issue: #197

You might test it via my branch: https://github.com/monotek/zammad-helm/tree/storage

from zammad-helm.

fixed in: #197

from zammad-helm.

@monotek for me to finally get it working again - even after #197 - I had to manually make some additional changes:

changes to the postgresql and elasticsearch init containers:

• also "dual-mount" the fs under /opt/zammad/var AND /opt/zammad/storage here
• set readOnlyRootFilesystem: false

for the zammad-railsserver, zammad-scheduler and zammad-websocket container I had to set

• readOnlyRootFilesystem: false
  to prevent getting some kind of "Read-only file system" errors in all the mentioned containers

And there is still one thing I cant get to work - the logo on the login page ist gone. Even if I set a new logo, it wont show again...weird.

from zammad-helm.

@monotek I second that. Even with the #197 the upgrade failed.

from zammad-helm.

Possible fix in: #206

from zammad-helm.