Fix flushing of unclean objects
This was preventing OSD failover when there were some unclean objects.
Bug was introduced in aa436027c8
This commit is contained in:
+2
-2
@@ -27,9 +27,9 @@ void osd_t::handle_peers()
|
||||
misplaced_objects += p.second.misplaced_objects.size();
|
||||
// FIXME: degraded objects may currently include misplaced, too! Report them separately?
|
||||
degraded_objects += p.second.degraded_objects.size();
|
||||
if ((p.second.state & (PG_ACTIVE | PG_HAS_UNCLEAN)) == (PG_ACTIVE | PG_HAS_UNCLEAN))
|
||||
if (p.second.state & PG_HAS_UNCLEAN)
|
||||
peering_state = peering_state | OSD_FLUSHING_PGS;
|
||||
else if (p.second.state & PG_ACTIVE)
|
||||
else if (p.second.state & PG_HAS_DEGRADED)
|
||||
peering_state = peering_state | OSD_RECOVERING;
|
||||
}
|
||||
else
|
||||
|
||||
Reference in New Issue
Block a user