[bitnami/harbor] Image push to Harbor registry takes a minimum of 30min and can go upto 45min also

[nanarun13]

Name and Version

bitnami/harbor:22.0.8

What architecture are you using?

None

What steps will reproduce the bug?

We have a Harbor registry setup using bitnami charts in our Kubernetes cluster. It uses Traefik as the ingress controller.

As per the traefik documentation, we have configured the ingress values.yaml.

Initially for bigger images, when our Jenkins builds were executed, the stage where we run the image push to Harbor registry took a lot of time to do a docker push action and eventually would result in a timeout or throwing a "HTTP 499 Client Closed Request" error. After following a number of technical issues reported on similar issues in the internet, we made the below configuration changes to our ingress controller values.yaml file:

traefik:
------------------
some code here
------------------
  ports:
    web:
      address: ":80"
      forwardedHeaders:
        insecure: true
      http:
        redirections:
          entrypoint:
            to: websecure
            scheme: https
      transport:
        respondingTimeouts:
          readTimeout: 0s
          writeTimeout: 0s
          idleTimeout: 0s        
    websecure:
      address: ":443"
      forwardedHeaders:
        insecure: true
      transport:
        respondingTimeouts:
          readTimeout: 0s
          writeTimeout: 0s
          idleTimeout: 0s
-----------------------
some code here
------------------------

WIth the above configuration, the frequent timeouts/http error 499 seems to have considerably reduced, making the image push action also successful, but unfortunately with a huge time duration( approx 30min-45min), which is way too high!!

Also, we do not see any errors in the postgresql DB pod logs or any of the other Harbor pods (registry, job service, exporter, etc)

Could you please help us in this regard? Any other configuration changes either in traefik or Harbor part required?

What do you see instead?

The harbor push stage in our cicd pipeline gets executed, but takes a very long time (upto 45 min) for the stage to complete, this time is too high considering productions executions.

[nanarun13]

Hi @javsalgar , the initial issue was raised against Harbor github issues, but they suggested to open this issue against the bitnmai charts. Please find the reference to the issue link in Harbor issue repo: https://github.com/goharbor/harbor/issues/20904

[javsalgar]

Hi,

To confirm, what CPU / mem resources you set for your production environment? Does the issue happen without traefik? Please note that this is a very specific installation that is hard for us to reproduce

[nanarun13]

Hi @javsalgar , the CPU/mem resources you are asking is for the total prod environment or particularly for the Harbor pod? Please confirm.

Also, since we have configured Harbor to work with traefik as the ingress, yes, the issue is seen currently with traefik. We have not tried the same without Traefik.

Please suggest.

[javsalgar]

I was thinking more on the harbor pods

[github-actions[bot]]

This Issue has been automatically marked as "stale" because it has not had recent activity (for 15 days). It will be closed if no further activity occurs. Thanks for the feedback.