Option to exclude persistent volume storage size from reconciler

[terryyanko]

Proposal

Currently there is no way to leverage an external volume autoscaler for Percona operator-managed MongoDB nodes. The Percona operator should provide an option to exclude persistent volumes from reconciliation in the case that you are using an external volume autoscaler to scale your PVs automatically. Alternatively, built-in PV autoscaling could work as well but would likely take more effort.

The Percona operator currently throws the below error when it recognizes that a PV has been scaled beyond it's initial spec...

ERROR	Reconciler error ... requested storage (50Gi) is less than actual storage (1142Gi)", "errorVerbose": "reconcile StatefulSet for rs0: resize volumes if needed: requested storage (50Gi) is less than actual storage (1142Gi

This could be remedied if there was an option to exclude the volume from the reconciliation loop and allow the external autoscaler to assume control of the PV sizing moving forward.

Use-Case

MongoDB persistent volumes could scale automatically without the need for any regular monitoring / manual intervention / config updates.

Is this a feature you are interested in implementing yourself?

Maybe

Anything else?

No response

[hors]

hi @terryyanko, you can disable it via enableVolumeExpansion: false option. Will it work for you?

[terryyanko]

Hey @hors I haven't tested it yet, but assuming it will bypass the reconciliation and alleviate the above error then sounds good to me!

[terryyanko]

@hors per the docs, isn't this already disabled by default? https://docs.percona.com/percona-operator-for-mongodb/scaling.html#automated-scaling-with-volume-expansion-capability

You can enable automated scaling with the [enableVolumeExpansion](https://docs.percona.com/percona-operator-for-mongodb/operator.html#enablevolumeexpansion) Custom Resource option (turned off by default)

Also looking at the volumes.go, I don't see how this would solve the issue...

[terryyanko]

@hors #1897