Bug #7218
closedDisplaying wrong number of pools with ceph -s after removeing a pool
0%
Description
$ceph osd lspools
0 pool0,1 pool1,2 pool2,3 pool3,
ceph osd pool delete pool0 pool0 --yes-i-really-really-mean-it
$ceph osd lspools
1 pool1,2 pool2,3 pool3,
$ ceph -s
pgmap v1800: 730 pgs, 4 pools,
Updated by Dan Mick over 10 years ago
I looked, and this is reporting a hash_map .size(); I'll bet .size() isn't an accurate count of currently-defined keys, but rather the allocated size
Updated by Greg Farnum over 10 years ago
Where are you looking?
I'm assuming that since this is PGMap output, the OSDs haven't all finished deleting the pool and updating the monitor's stats.
Updated by Dan Mick about 10 years ago
The experiment I did yesterday still shows the inconsistency. I was just looking at the output, in PGMap::print_summary, and assuming what probably happens with hash_map.size(). I don't know. Maybe the issue is that pg_pool_sum is stale, but it hasn't updated itself in nearly a day.