Bug #42346
closed
Nearfull warnings are incorrect
Added by Anonymous over 4 years ago.
Updated about 4 years ago.
Description
OSD_NEARFULL 2 nearfull osd(s)
osd.53 is near full
53 hdd 9.09470 1.00000 9.1 TiB 1.3 GiB 287 MiB 0 B 1 GiB 9.1 TiB 0.01 0 0 up
OSD53 got purged and formatted and readded to cluster.
osd.96 is near full
96 hdd 10.00000 1.00000 9.1 TiB 6.8 TiB 6.8 TiB 84 KiB 17 GiB 2.3 TiB 74.57 1.20 31 up
Could you add some more explanation to this issue?
- Which commands did you run?
- What was wrong`
- What did you expect instead?
I dont expect to see nearfull warnings if they are not over threshold.
Commands run: I see this every time there is a rebalance, and others see, it's a common thing asked in IRC channel.
Certainly dont expect to see a near-full warning for an OSD that's empty ;)?
- Project changed from mgr to RADOS
- Priority changed from Normal to High
- Assignee set to David Zafman
- Status changed from New to In Progress
- Pull request ID set to 31954
Spurious nearfull warnings caused by backfill reservation mechanism during rebalancing. The nearfull ratio was compared against the backfill reservation adjust space. The fix is to use the actual unadjusted value.
- Status changed from In Progress to Pending Backport
- Status changed from Pending Backport to Resolved
While running with --resolve-parent, the script "backport-create-issue" noticed that all backports of this issue are in status "Resolved" or "Rejected".
Also available in: Atom
PDF