Chart v9.0.1 breaks existing redis config #472
Reference in New Issue
Block a user
No description provided.
Delete Branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Installing/upgrading to chart v9.0.1 with configuration working in v9.0.0 breaks redis with this in the log:
I'm using external GCP redis with this config:
Do I have to use redis-cluster instead of cache in v9.0.1? If so, what should it look like? I'm a bit confused.
Without
redis-cluster.enabled=true
you don't have any provider forcache
orsession
(as shown in the log above). You need one and if you don't want to useredis-cluster
you need to configure an external provider for these (e.g.memcache
).Hmmm... I'm confused. Are you saying that I need something like this? If not, can you show me an example of what the currently working config (in v9.0.0) should look like for v9.0.1?
An example for a
cache
config withredis-cluster
is in the upgrading notes.There's no change for the config between 9.0.0 and 9.0.1. I am not sure we're understanding each other right now 😅
The only thing you need for 9.0.0 is an explicit definition of
cache
and friends (again see the upgrading notes) due the bug shown in #356I'll release 9.0.2 in a few seconds, it's best to try with this one - but you will still need to configure the above mentioned values in your
values.yml
.My current Gitea config works in v9.0.0/1.19.4. If I update to (or install from scratch) v9.0.1/1.19.4 with no changes to the Gitea config, I get the error (PANIC: dial tcp 127.0.0.1:6379: connect: connection refused) as seen in the log above.
I don't know how to get the config used by Gitea when it can't start with v9.0.1 as the pod is in state CrashLoopBackOff. I would like to compare it with the config used when I install with the v9.0.0 chart.
Ah sorry, I think I misunderstood you. If you have an external working redis, you shouldn't set
redis-cluster.enabled=true
.In 9.0.1
session
has been set toredis
by default, I think this is where you're failing.Can you try definining
session
with your existingredis
instance as described in the Gitea docs?if you don't define
session
and also haveredis-cluster
disabled, it might be that a non-working configuration is written toapp.ini
. I need to check that.9.0.1 configured
session
by default withredis-cluster
- or expects a working external redis configuration. Which is missing in your case if you haven't set it explicitly before.@tobiasbp Did my suggestions help to solve your issue?
I'm afraid don't understand what you are suggesting :(
This is my current config with v9.0.0/1.19.4. If I change the chart to v9.0.1 (also new install) Redis breaks as sin in initial post.
These are the values used with the v9.0.0 chart:
Resulting /data/gitea/conf/app.ini from running pod:
I have the same issue, after upgrading to gitea-helm 9.x I can't run a simple gitea-instance. I don't need or want a redis-cluster, nor do I have a big enough instance to need something more than a sqlite-db.
https://docs.gitea.com/installation/install-with-docker
Gitea itself also doesn't need redis, nor an external database to work.
Are you saying it's impossible to use this helm-chart without external dependencies?
For context, I have tried the following (this is just the relevant parts):
It still looking for the service for the embedded redis-cluster.
@tobiasbp
You need to configure https://docs.gitea.com/next/administration/config-cheat-sheet#session-session with your existing redis installation.
And please use the latest release (9.0.3) and not 9.0.1.
I was able to get Gitea 1.19.4 running on chart v9.0.4 by adding these:
Should I still have these in the cache config??
cache
andsession
are two different settings, so yes - you need both. And this is unrelated to 1.19 or 1.20. Why do you want to go with 1.19?Before v9.x of the helm chart,
session
was not auto-configured but onlycache
.Because S3 storage appears to be broken in 1.2.x.
https://github.com/go-gitea/gitea/issues/25984
The chart will also work with 1.19 but you would need to issue a DB downgrade if you already upgraded to 1.20.