proxy_timeout in k3d-proxy nginx configu for API server should be increased

[fsimmend]

When i attach to a pod to follow the logs with kubectl, the proxy_timeout in the k3d-proxy is set too low and sometimes if nothing happens in the pod, the connection is closed. This timeout should be much higher or configurable.

• How was the cluster created?

  • k3d create -w 4

• What did you do afterwards?

1. kubectl apply -f mydeployment.yaml
2. kubectl get logs --follow <pod of a job>
3. error: unexpected EOF

What did you expect to happen?
i expected that kubectl would follow the logs until the job is completed but instead at a point that takes about 3-5 minutes without any new log messages, the error appeared. I then attached to the k3d-proxy container increased the timeout, reloaded the nginx config and
retried my long running job and id did work.

• MacOS

Which version of k3d?

k3d version v3.0.0-rc.2
k3s version v1.18.3-k3s1 (default)

Which version of docker?
Docker version 19.03.8, build afacb8b

[iwilltry42]

Good catch @fsimmend :)
Fixed 👍

[PuppetA17]

Has the problem been solved?
How do I change the k3d-proxy proxy_timeout when creating a cluster

[iwilltry42]

@herozabbix, you cannot change it on cluster creation. It has been increased to 10 minutes.
If you really need more than that right now, you can workaround this issue by

• exec'ing into the loadbalancer container
• changing the value in the nginx.conf manually
• reloading nginx (via SIGHUP)