Operator panics when scheduling a backup if crVersion is not present

[Kajot-dev]

Report

When a resource is missing spec.crVersion property operator panics when scheduling a cron backup.

Expected result:

• Either set crVersion mandatory on the CRD side so no cluster without it can be created or make it actually not required
• If it's required operator errors in clear way, not crashing for all users

More about the problem

Logs:

panic: Malformed version: goroutine 385647 [running]: github.com/hashicorp/go-version.Must(...) /go/pkg/mod/github.com/hashicorp/go-version@v1.7.0/version.go:105 github.com/percona/percona-postgresql-operator/pkg/apis/pgv2.percona.com/v2.(*PerconaPGCluster).Version(0x22e4931?) /go/src/github.com/percona/percona-postgresql-operator/pkg/apis/pgv2.percona.com/v2/perconapgcluster_types.go:367 +0x47 github.com/percona/percona-postgresql-operator/pkg/apis/pgv2.percona.com/v2.(*PerconaPGCluster).CompareVersion(0x0?, {0x22e1e7d, 0x5}) /go/src/github.com/percona/percona-postgresql-operator/pkg/apis/pgv2.percona.com/v2/perconapgcluster_types.go:371 +0x1d github.com/percona/percona-postgresql-operator/percona/controller/pgcluster.(*PGClusterReconciler).createScheduledBackup(0xc0008c0420, {{0x2780588?, 0xc0097e0db0?}, 0x0?}, {0xc0082782c0, 0x1c}, {0x22e137d, 0x4}, {0xc00916a890, 0x5}, ...) /go/src/github.com/percona/percona-postgresql-operator/percona/controller/pgcluster/schedule.go:121 +0x2f6 github.com/percona/percona-postgresql-operator/percona/controller/pgcluster.(*PGClusterReconciler).reconcileScheduledBackup.(*PGClusterReconciler).createScheduledBackupFunc.func1() /go/src/github.com/percona/percona-postgresql-operator/percona/controller/pgcluster/schedule.go:82 +0x91 github.com/robfig/cron/v3.FuncJob.Run(0x0?) /go/pkg/mod/github.com/robfig/cron/v3@v3.0.1/cron.go:136 +0x12 github.com/robfig/cron/v3.(*Cron).startJob.func1() /go/pkg/mod/github.com/robfig/cron/v3@v3.0.1/cron.go:312 +0x55 created by github.com/robfig/cron/v3.(*Cron).startJob in goroutine 120 /go/pkg/mod/github.com/robfig/cron/v3@v3.0.1/cron.go:310 +0x90

Operator crashes for all users

Steps to reproduce

1. Create a cluster without crVersion or migrate from PGO from crunchydata (in my case was reproducible on ~50% clusters that I have migrated, on the rest crVersion was added automatically)
2. Create scheduled backup
3. Observer operator panic when trying to schedule a backup

Versions

1. Kubernetes 1.24
2. Operator 2.6.0
3. Database - postgres 15, but happens on all versions

Anything else?

No response