Consuming a large amount of resources during restoration from backup

[vborisov12]

Report

I am creating a PerconaServerMongoDBRestore to restore from a backup (logical) in a new environment. During the initialization of the restore, the operator starts a resync of metadata for all backups in s3. During this process, it consumes a lot of resources.

1. What causes such resource consumption?
2. Why does resync run for all objects in s3, and not just my recovery backup?

More about the problem

Logs:

...
INFO	Cluster state changed	{"controller": "psmdb-controller", "object": {"name":"some-mongodb-cluster","namespace":"stage"}, "namespace": "stage", "name": "some-mongodb-cluster", "reconcileID": "18075461-5982-4b89-a73c-c7645095c81c", "previous": "ready", "current": "initializing"}

INFO	Starting restore	{"controller": "psmdbrestore-controller", "object": {"name":"mongo-data-restore","namespace":"stage"}, "namespace": "stage", "name": "mongo-data-restore", "reconcileID": "10b7c882-3985-4f77-8dad-f0343a5ecd80", "backup": "2024-10-29T08:05:29Z"}
D [resync] got physical restores list: 0

D [resync] got backups list: 52
D [resync] bcp: 2023-11-08T06:35:21Z.pbm.json
D [resync] bcp: 2023-11-09T08:38:31Z.pbm.json
...

Steps to reproduce

Versions

1. Kubernetes v1.27.16
2. Operator v1.15.0
3. Database 5.0.14-12

Anything else?

No response