docker container resource requests and limits

[uzdik-student]

https://hub.docker.com/r/redpandadata/console We use the above docker image to run redpanda-console as a pod in kubernetes cluster, as you can see from the screenshot, we have 107308070 messages in our topic, everything works fine until we select START OFFSET = Oldest:

It starts consuming messages from the beginning and stops after several seconds with the following error:

From the logs, we can see that the pod was restarted 5 seconds after redpanda started consuming messages.

Here are the requests and limits we set for our pod:

Limits:
  cpu:     100m
  memory:  700Mi
Requests:
  cpu:      50m
  memory:   600Mi

Could that be the reason of pod restarts? What are the recommended memory and cpu limits for redpanda-console container?

[weeco]

Hey @arman-sydikov , yes the resource constraints may be the reason. I'd recommend 2Gi of Memory and a burstable CPU configuration, since the only resource intensive task is a message search such as you were performing. If acceptable in your environment, you may want to remove the CPU limit entirely, but a limit of 1vCPU sounds reasonable too. The number of messages it can process per second is likely to be CPU bound. In our experience it is around ~50k messages / second for each core that it can use.

[uzdik-student]

@weeco

thank you for the advise, I have increased the limits to 2 cpu and 2.5Gi memory

Limits:
  cpu:     2
  memory:  2560Mi
Requests:
  cpu:      50m
  memory:   50Mi

now it doesn't restart, but stops after consuming about 45 Gib / 50 million messages with the following error:

[weeco]

What Console version are you using?

[uzdik-student]

What Console version are you using?

Redpanda Console (Platform Version v24.1) (built May 30, 2024) 788406B

[weeco]

Can you tell how long this search was running? I think we have some hardcoded upper limit like 35minutes or so.