Closed agadelshin closed 1 week ago
More I think about it - if requeue is still required for some reason - it's better to increase duration between runs to some reasonable period like 10-30 seconds.
Removed requeAfter changes because it's already a part of #1016
Closed in favor of #1102 and #1103.
@agadelshin Thank you for the proactive proposal!
Added status update on fail. There is a way to just run it in a defer, but it's only 2 places so I guess it's more straightforward solution.
Also changed Update to Patch, added request of virtualmachine before to avoid conflicts.
And a last one - removed requeue after succesful reconcile, it's not required to reconcile if nothing happens, reconcile will be queued if there will be a change in a VirtualMachine or Pod with VirtualMachine owner. If there is another reason to keep it in a constant loop - please tell, will update it. Otherwise it'll save some CPU cycles and make logs more readable.
Tested it manually on a virtualmachine inside GCP, if there is a way to run autoscaling on Silicon Apple, I'd be happy to know (x86-64 emulation in QEMU is too slow, took hours to build everything).