fix: set default request timeout to 120 seconds

[maxsxu]

Motivation

Fixes #3207

Modifications

• Change default request timeout to 120s from 5s. Since 5s is too short for most of requests, especially for large clusters in production environment.

And according to kubectl docs, the default kubectl request timeout is 0 which means don't timeout requests. Maybe we should do so too for k9s, but let's see.

[derailed]

@maxsxu Thank you for this update! The current default timeout is 10s.
I am hesitating making it any bigger, since you can easily override it with a cli arg.
What do you think?

[maxsxu]

@maxsxu Thank you for this update! The current default timeout is 10s. I am hesitating making it any bigger, since you can easily override it with a cli arg. What do you think?

@derailed Thanks for reply!

And yes we can override via cli arg, but if this is something users' frequent use case, then we can make it a common option.

So wondering what's the backwards if making the default timeout bigger? If that won't break us, then I suggest making it bigger.

[robert-openai]

@derailed Could you at least make it something you can set in config.yaml? I didn't see anything mentioned in the documentation about it. The K8s clusters I deal with regularly are far too big to reliably reply within the default request timeout.

[vitali-raikov]

We are in the same boat, we are dealing with a pretty big clusters (~20K pods) so a lot of k9s functionality is just not working unless we set timeout to something like 20-30 seconds.

Setting CLI arg is a workaround and I could probably alias my k9s command to be something like k9s --request-timeout=30s to k9s but it would be nice to be able to override it via config.yaml as @robert-openai proposed.

[MrVinkel]

I work with crossplane clusters which has 100k+ resources and I have to specify the --request-timeout for doing anything in those. Working with any bigish cluster a timeout of 10s is just too small.

I agree with @vitali-raikov that it would be nice if it could be configured in the config.yaml

[gottschd]

Please allow me to share my user experience with K9s, in the spirit of honest feedback.

As a casual, non-professional user, I've found K9s incredibly user-friendly, particularly when installed and updated via Chocolatey on Windows. It has consistently worked seamlessly for me with the default settings, requiring no additional configuration, which has been greatly appreciated.

However, with the recent change in the default request time, I encountered some unexpected challenges. This led to a bit of a troubleshooting journey to identify the source of the issue. My DevOps colleagues (who were using an much much older version of K9s 😄 ) , initially speculated about network infrastructure and "TCP keep-alive" settings, topics that are quite complex for me to grasp.

I am grateful for the timeout fix that has been proposed, as it allows me to return to the straightforward experience of having K9s work effortlessly out of the box for my needs.

Thank you for your effort and continued dedication to improving the software.