Bug #63334: Recovery starts while norecover flag is set when PG splitting occurs - RADOS - Ceph

Actions

Copy link

Bug #63334

open

Recovery starts while norecover flag is set when PG splitting occurs

Added by Aishwarya Mathuria 6 months ago. Updated 4 months ago.

Status:

Pending Backport

Priority:

Normal

Assignee:

Aishwarya Mathuria

Category:

Target version:

% Done:

Source:

Tags:

backport_processed

Backport:

reef

Regression:

Severity:

3 - minor

Reviewed:

Affected Versions:

ceph-qa-suite:

Component(RADOS):

Pull request ID:

54708

Crash signature (v1):

Crash signature (v2):

Description

This was observed while client I/O was happening on a pool and PG splitting occurred.
Some PGs went into degraded state and when we try to write to a degraded object the current recovery work flow attempts to recover that particular object even if the norecover flag is set.

Steps to reproduce on vstart cluster:
1. Set norecovery and nobackfill flags
2. Create a pool
3. Use radosbench to write objects to this pool
4. While objects are being written to this pool, increase the number of PGs (pg_num).

Related issues 1 (1 open — 0 closed)