Closed realHarter closed 1 year ago
Hey @realHarter thanks for the addition!
It's odd that your deployment fails because of the livenessProbe. I remember there were some changes recently from the upstream project to add new scripts which were used during probe checking. Furthermore it is odd that none of my tests catch any issues with deployments of the recent versions.
Will check next week with my own deployment if I can find any issue. If you in the meantime also find some interesting logs or ideas on what could be wrong - maybe some weird thing on your Cluster? - I am keen for input.
Hey @dirsigler ,
so I did a little more testing, and it turns out that the default timeoutSeconds of the liveness probe of 2 seconds was set too low for my system.
When manually hitting the probe it turned out to work just fine, the liveness probe just needed about 3 secounds (k3s on a raspberry pi 4B 8GB).
It's unfortunate that the application doesn't seem to log anything about the probe.
So it's not really a bug, but depending on how long the probe takes on more powerful systems I'd suggest that we might increase the default value of the livenessProbe.timeoutSeconds
setting.
Description of the change
add enable flag to liveness and readiness probes. Default is set to be enabled.
Benefits
ability to disable probes for use in test environments
Possible drawbacks
Applicable issues
Additional information
The liveness probe of my deployment fails for no apparent reason. The application works just fine when I manually disable it. So I thought there should be an option in the Chart to do so, if one wants to run the application in a non critical environment where some downtime won't matter.
Checklist
Chart.yaml
according to semver.