Physical restore cannot start if a node is in RECOVERING state
General
Escalation
General
Escalation
Description
Environment
None
relates to
Activity
Show:

Jan Mynar September 12, 2024 at 10:03 AM
duplicate of this feature - https://perconadev.atlassian.net/browse/PBM-1335
Duplicate
Details
Details
Assignee
Unassigned
UnassignedReporter

Needs QA
Yes
Priority
Smart Checklist
Open Smart Checklist
Smart Checklist

Open Smart Checklist
Created June 6, 2024 at 12:40 PM
Updated September 12, 2024 at 10:03 AM
Resolved September 12, 2024 at 10:03 AM
Had a sharded cluster with 2 shards, each is a 3 node replica set.
One of the replica set members for shard1 is in RECOVERING state (e.g. due to small oplog window)
If you try to do a physical restore and one of the nodes is in a state not able to process queries, then restore will be stuck:
This happens because the pbm-agent tries to issue following query to every node:
however since this particular node is in RECOVERING state, it cannot accept any queries.