Clean install has three pods not coming up about helm-charts

[karezza]

Modifying the default resources for rspamd got that one working, the default values must need to be updated, here's what worked from me taken from another ticket:

rspamd:
  resources:
    limits:
      cpu: 600m
      memory: 500Mi
    requests:
      cpu: 500m
      memory: 500Mi

[ma1ko]

I was just stuck with the same issue. The last log line from rspamd was:
rspamd_url_init: start compiling of 9270 TLD suffixes; it might take a long time
then it fails startup checks because it takes too long to start up. Increasing resources seemed to have worked for me as well (it probably makes the startup fast enough).
Another solution would probably be to give rspamd more time to start

[karezza]

closing, i'll submit a pull request later

[karezza]

Being in dev I didn't set things to work with cert-manager, this caused the mailu-front pod not to load. It started working once I manually created a tls secret called: mailu-certificates.

[karezza]

clamav seems to have started working now that rspamd is working, not yet closing ticket though as a placeholder to get the defaults of the resource for rspamd updated for the sake of future users

[micw]

Can you figure out, why the startup failed (checking the logs)?

[AndreKoepke]

@karezza I had the same problem. Can you open the PR? ;)

[lknite]

Been awhile, I think what I found was a new helm chart which didn't use a single pvc which required rwx, but rather multiple pvcs as needed to go along with each service and I thought the resources requirements were going to be updated at this time. I believe my earlier post of values for resources should work for you. I'm still using this helm chart at the moment and everything is working. If those don't work for you let me know. If you search here you'll find additional issues talking about these resources, that's where I got my answer, might find more there on efforts being made by others.

[lknite]

See if that other helm chart was finished up, rwx is generally a bad idea. I had to use a local share and node affinity / taints.

[AndreKoepke]

Nice to know. I'm came from "plain kubernetes" and migrating to helm.
But yes, the values you posted before works.