Lot of errors in pods about zitadel-charts HOT 14 CLOSED

I've send you a mail with the attached har-file.

from zitadel-charts.

I can provide an update on this. As mentioned in #105 I've imported the whole CRDB cluster to a new setup and installed zitadel chart 4.6.0. The app loads in a normal way and I don't get any errors like before. So I can make a backup of the cluster and restore it on a fresh installation. That's good to know. You can add this also to the documentation how to backup and restore on a new (k8s-) cluster.

from zitadel-charts.

Can you share what version you used and deployed now?

from zitadel-charts.

I've used the Chart in version 4.3.0 and CRDB in 11.0.0. I've upgraded to 4.6.0 and CRDB to 11.0.1. So no huge changes at all. Very often the site doesn't load and I get a blank screen. Or some options doesn't load and I need to refresh the site. Sometimes I need to do this 4 to 5 times :-D

from zitadel-charts.

When I reload the page with opened developer tools in FF, sometimes when the pages blank, I can see that a JS-File main.ecd3aed263a08a21.js can't be loaded. Curious.

from zitadel-charts.

So in Firefox (LIbrewolf 114.0.1-2) Zitadel currently does not load at all. I have not changed anything else. The messages are the same. A login is also no longer possible on some pages where Zitadel was included as OIDC. A private window also makes no difference.

Chromium (UngoogledChromium 114.0.5735.106) on the other hand works without further or worse problems. There the page remains only rarely blank.

And before the update everything went without further problems. It happened that the page was not loaded correctly, but that is normal for me.

from zitadel-charts.

Hm that is a really strange behavior, can you quickly verify what zitadel version was deployed (the version in the k8s deployment).

This sounds to me like a http1.1/http2 issue

from zitadel-charts.

@fforootd the version in the deployment is image: ghcr.io/zitadel/zitadel:v2.23.1

Hope that helps. You say it, strange behavior.

from zitadel-charts.

What happens if you overwrite this version with a more recent one.

I.e. 2.27.2

from zitadel-charts.

Now the interface does not load at all on FF or Chromium. I think there are any corrupted DB entrys possible?
Yesterday I've set up another Zitadel Cluster and done the Upgrade like on prod. No problems. All very smooth. The only difference was that we deployed prod with fluxcd. I think its possible that flux is not the best way to deploy such a critical system like zitadel.

Can you set a more recent version in the helm charts? Only if I will setup a new cluster :-D

Edit: When I revert and set the old version the dashboard will load again in Chromium

from zitadel-charts.

Interesting, newer have seen this 😬

I think nothing really speaks against using fluxcd to deploy ZITADEL.

Is it possible that you dump a har file and send it to [email protected]?
I want to poke a little around on what exactly is replied by the server when the UI is breaking.

from zitadel-charts.

I looked at the har file and could not really figure out why there is a status code 0 😁

What kind of proxy do you run as ingress? Maybe we should look at this as well.

from zitadel-charts.

We use nginx as ingress. With the default config. Nothing special.

from zitadel-charts.

It looks like you have a running cluster now, and it is hard to find out what caused the errors before.
I will close the issue for now, feel free to reopen if needed

from zitadel-charts.