Hello,
I'm running a (very) small cluster and I'd like to turn on pg_autoscale.
In the documentation here >
http://docs.ceph.com/docs/nautilus/rados/operations/placement-groups/ it
tells me that running
ceph config set global osd_pool_default_autoscale_mode <mode>
should enable this by default, however running the below:
$ ceph config help osd_pool_default_autoscale_mode
Error ENOENT:
seems to tell me that the setting isn't correct.
By grepping around I found osd_pool_default_pg_autoscale_mode (with _pg
after default) to be the correct setting, however it seems this can't be
changed at runtime as docs suggest:
$ ceph config help osd_pool_default_pg_autoscale_mode
osd_pool_default_pg_autoscale_mode - Default PG autoscaling behavior for
new pools
(str, advanced)
Default: warn
Possible values: off warn on
Can update at runtime: false
So after enabling this via ceph-ansible by overriding the conf I tried
to set it also on existing pools.
This is the output before my attempt:
$ ceph osd pool autoscale-status -f json-pretty | jq '[ .[] |
{'pool_name': .pool_name, 'pg_autoscale_mode': .pg_autoscale_mode }]'
[
{
"pool_name": "cephfs_metadata",
"pg_autoscale_mode": "warn"
},
{
"pool_name": "default.rgw.meta",
"pg_autoscale_mode": "warn"
},
{
"pool_name": "cephfs_data",
"pg_autoscale_mode": "warn"
},
{
"pool_name": "default.rgw.control",
"pg_autoscale_mode": "warn"
},
{
"pool_name": ".rgw.root",
"pg_autoscale_mode": "warn"
},
{
"pool_name": "default.rgw.log",
"pg_autoscale_mode": "warn"
}
]
Then I tried executing the following:
$ docker exec -it ceph-mon-sagittarius ceph osd pool set cephfs_metadata
pg_autoscale_mode on
set pool 2 pg_autoscale_mode to on
However, while it seems the operation was successful, the following
command returns:
$ docker exec -it ceph-mon-sagittarius ceph osd pool get cephfs_metadata
pg_autoscale_mode
pg_autoscale_mode: warn
And since pg_autoscaler module is turned on, this is the output of ceph
status:
ceph status
cluster:
id: d5c50302-0d8e-47cb-ab86-c15842372900
health: HEALTH_WARN
5 pools have too many placement groups
services:
mon: 1 daemons, quorum sagittarius (age 5m)
mgr: sagittarius(active, since 2m)
mds: cephfs:1 {0=sagittarius=up:active}
osd: 4 osds: 4 up, 4 in
rgw: 1 daemon active (sagittarius.rgw0)
data:
pools: 6 pools, 184 pgs
objects: 242.08k objects, 862 GiB
usage: 1.7 TiB used, 27 TiB / 29 TiB avail
pgs: 184 active+clean
What am I doing wrong?
Thank you.
Daniele
_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
