Set the min so that the pg_autoscaler won't scale it back down. Choose
something large enough that even for a large FS we'll do reasonably well
distributing the metadata workload across OSDs.
Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit
b567b110cb6286d4676653179cb86ae1fe861f15)
r, outb, outs = self.mon_command({
'prefix': 'osd pool create',
'pool': mdp_name,
- 'pg_num': 8
+ 'pg_num': 16,
+ 'pg_num_min': 16,
})
if r != 0:
return r, outb, outs