Only restartChecks if pod is master for configReloader

[joshulyne]

Fixes #925

[integrii]

Great find! I feel like this one should go into a different place, though.

Check out the control code here. This control loop is responsible for doing various check starts and stops and reloads. It feels like the configReloader func should be refactored to return a channel, then after reloading configuration, it should notify the channel that configuration has changed. That signal would go into the switch in the control code loop and result in this reload (but only if master!).

This way we have a central place to manage all reloading and check start/stopping control flow...

[integrii]

You could probably move the config reloader goroutine to be just above the control flow... up where there call to monitorKHJobs() is...

[joshulyne]

Great find! I feel like this one should go into a different place, though.

Check out the control code here. This control loop is responsible for doing various check starts and stops and reloads. It feels like the configReloader func should be refactored to return a channel, then after reloading configuration, it should notify the channel that configuration has changed. That signal would go into the switch in the control code loop and result in this reload (but only if master!).

This way we have a central place to manage all reloading and check start/stopping control flow...

makes sense! will refactor to do so!

[joshulyne]

@integrii should we also be restarting the Reaper -- like in the case above?
https://github.com/kuberhealthy/kuberhealthy/pull/942/files#diff-6fe2eb1bda3003e83c281ea43553e293915389e31314d523116afe24efe877e3R244

[integrii]

I would restart the reaper just in case. This way we are more likely to make any new configuration options that have been set for it take effect.

[jonnydawg]

thanks!

resolved changes!