Closed RyuunosukeDS3 closed 7 months ago
The problem was caused by readiness check in plex and jackett. Removing those from the template fixed the problem.
Now we need to figure out if its a specific problem of mine, or if this is happening to everyone else.
The automated tests check the readiness of each app, and each release is tested against the probes, what error did you get while you deployed it? @RyuunosukeDS3
It said it couldn't reach... Also ingress didn't work for me... So i ended up using loadbalancers for everything and now it works.
I have also faced sporadic issues with the readinessProbe when I was trying to deploy today (when the server hasn't been claimed) it seems to return 401 unauthorized
, which is treated as a problem by Kubernetes, which marks this service as not ready. I'm not sure if readinessProbes allow you to do something about the return code, but 401 should not be treated as an error in this case I feel.
Same problem here, I am deploying with ArgoCD and plex is the only service that is "unhealthy" as it returns a 401 error. I both tried with and without claim, facing the same issue.
As mentioned above, it's due to the misconfigured readinessProbes
,
I ended up doing
helm template helm -n mediaserver --create-namespace -f values.yml > rendered.yml
And manually removing the readinessProbes
from rendered.yml
for the Plex deployment
As mentioned above, it's due to the misconfigured
readinessProbes
, I ended up doinghelm template helm -n mediaserver --create-namespace -f values.yml > rendered.yml
And manually removing the
readinessProbes
fromrendered.yml
for the Plex deployment
I've been struggling with this and I suspect it has to do with #84 changing the readiness probe from a simple TCP connect to looking for an HTTP response, but Plex does not always give a 200 response, which is what it appears httpGet
is looking for by default, and if it doesn't get that - even if Plex is working as designed - the readiness check fails and there is no ingress. Plex is still running and it can reach out to the internet, the Plex app on various devices can pull content and transcode/stream it, but I am unable to reach the Plex instance directly unless I proxy through an SSH tunnel into the internal cluster network.
I think this HTTP readiness check, at least in the case of Plex, is too aggressive.
Thanks for the great feedback, that's definitely something I'll need. I'll take the checks offline and see if I can end up with a consistent way of testing (that's the main point) during the integration tests without affecting the deployment.
I'm not able to reproduce this, even doing all the steps manually that's what I get, the readiness succeeds, and I am able to end the configuration.
mediaserver jackett-8676c88788-kq6kp 1/1 Running 0 111s
mediaserver plex-55c9677c47-ftkfw 1/1 Running 0 111s
mediaserver prowlarr-798ccd854-nmxw6 1/1 Running 0 111s
mediaserver radarr-6dd774c49d-bd8sl 1/1 Running 0 111s
mediaserver sabnzbd-566b78d7cd-xcgws 1/1 Running 0 111s
mediaserver sonarr-77d7449699-cvqbf 1/1 Running 0 111s
mediaserver transmission-f995f9769-kzm4d 1/1 Running 0 111s
I'm not able to reproduce this, even doing all the steps manually that's what I get, the readiness succeeds, and I am able to end the configuration.
mediaserver jackett-8676c88788-kq6kp 1/1 Running 0 111s mediaserver plex-55c9677c47-ftkfw 1/1 Running 0 111s mediaserver prowlarr-798ccd854-nmxw6 1/1 Running 0 111s mediaserver radarr-6dd774c49d-bd8sl 1/1 Running 0 111s mediaserver sabnzbd-566b78d7cd-xcgws 1/1 Running 0 111s mediaserver sonarr-77d7449699-cvqbf 1/1 Running 0 111s mediaserver transmission-f995f9769-kzm4d 1/1 Running 0 111s
I have a local NAS where the config is stored, storage is good. It seems like when I rm that entire config directory and Plex starts anew, it starts up and gives a 200, goes green in Rancher, good ingress. Readiness checks are always good with a fresh virgin config. After that, response codes change, I think it's doing a 302 Found
and redirecting to plex.tv, and the readiness check doesn't like that. Only way to get the readiness check to succeed is when it starts fresh with no config.
Even with a brand new config/pvc and token I was not able to get Plex to return anything other than the 401 to the probe on /
.
When setting the probe as /web
instead, this works fine, maybe we could use this instead.
fixed by #102 Thank you!
Describe the bug Plex wont let me connect in any form
To Reproduce Steps to reproduce the behavior:
Expected behavior Plex would connect to UI
Screenshots
Environment: Client Version: version.Info{Major:"1", Minor:"21+", GitVersion:"v1.21.13-3+b0496756fa948e", GitCommit:"b0496756fa948e718d67351ed8e5293c3a28f0b8", GitTreeState:"clean", BuildDate:"2022-06-08T10:21:43Z", GoVersion:"go1.16.15", Compiler:"gc", Platform:"linux/arm64"}
Additional context There is my values file:
Application file: