Gonzo take a long time to parse kubectl logs

[Ga13Ou]

First of all, nice tool! It’s a very practical utility to have.

Description:
When using gonzo to display Kubernetes logs together with k9s, I noticed that it takes a long time to load and parse the logs. This makes it difficult to use for quick debugging or tailing in real time.

Am I using the tool incorrectly? I installed the plugin following the documentation.

[destari]

Hmm I doubt you are using it incorrectly- we have had a few other issues opened about slow load in some cases, and I cannot yet reproduce it (making me pull my remaining hair out). Can you share details? How are you using it, any params, how many logs, and specific type of logs or anything? Does it work if you pipe logs in, or pulling from a file?

[j0nr33v3]

Thank you @Ga13Ou for the report here - the other thing we might try is if you could manually grab the logs and pass to Gonzo via kubectl e.g. kubectl logs -f $NAME -n $NAMESPACE --context $CONTEXT | gonzo to see if the delay is being introduced by the K9s -> Gonzo plugin.

[Ga13Ou]

Thank you for the quick response. I sadly can't share the logs. However, it happens whenever we have a lot of logs which is almost always the case for pod that have been running for a couple of hours. I am using this plugin config.

 gonzo:
    shortCut: Ctrl-L
    description: "Gonzo log analysis"
    scopes:
      - po
    command: sh
    background: false
    args:
      - -c
      - "kubectl logs -f $NAME -n $NAMESPACE --context $CONTEXT | gonzo"

I did try to run kubectl logs -f $NAME -n $NAMESPACE --context $CONTEXT | gonzo and it is the same behaviour. I can see that gonzo is trying to parse the logs at a rate that I vaguely estimate at 40 lines/sec (I didn't measure it). And in my case the pod had around 28k lines of log (a normal Nginx pod) which will take a couple of minutes to get to the last line of code (this is without enabling the -f flag in kubectl)

[j0nr33v3]

@Ga13Ou - I'm wondering if we should add something like --tail=0 to the plugin command - since the intent here is to tail/follow logs from "now"...? maybe that helps with your scenario?

[Ga13Ou]

It would still be useful to have some old lines from the logs, but, indeed, that can only be practical if the parsing was a bit faster

[j0nr33v3]

@Ga13Ou I updated the default K9s instructions to include the --tail=0 for now; for your use case, do you typically want to look at "older" log lines - are you looking at pod logs primarily or also files etc...?

[Ga13Ou]

Thank you @j0nr33v3, in my case, I will more likely be using it through K9s