Re: How to set osd_max_backfills in Luminous

[Jean-Charles Lopez <jelopez@xxxxxxxxxx>]

Hi,

to check a current value use the following command on the machine where the OSD you want to check is running

ceph daemon osd.{id} config show | grep {parameter}

  Or

ceph daemon osd.{id} config get {parameter}

What you are seeing is actually a known glitch where you are being told it has no effect when in fact it does. See capture below

[root@luminous ceph-deploy]# ceph daemon osd.0 config get osd_max_backfills

{

    "osd_max_backfills": "1"

}

[root@luminous ceph-deploy]# ceph tell osd.* injectargs '--osd_max_backfills 2'

osd.0: osd_max_backfills = '2' rocksdb_separate_wal_dir = 'false' (not observed, change may require restart)

osd.1: osd_max_backfills = '2' rocksdb_separate_wal_dir = 'false' (not observed, change may require restart)

osd.2: osd_max_backfills = '2' rocksdb_separate_wal_dir = 'false' (not observed, change may require restart)

[root@luminous ceph-deploy]# ceph daemon osd.0 config get osd_max_backfills

{

    "osd_max_backfills": "2"

}

Regards

JC

On Nov 21, 2017, at 15:17, Karun Josy <karunjosy1@xxxxxxxxx> wrote:

Hello,

We added couple of OSDs to the cluster and the recovery is taking much time.

So I tried to increase the osd_max_backfills value dynamically. But its saying the change may need restart. 

$ ceph tell osd.* injectargs '--osd-max-backfills 5'

osd.0: osd_max_backfills = '5' osd_objectstore = 'bluestore' (not observed, change may require restart) rocksdb_separate_wal_dir = 'false' (not observed, change may require restart)

------------

------------

=============

The value seems to be not changed too.

[cephuser@ceph-las-admin-a1 home]$  ceph -n osd.0 --show-config | grep osd_max_backfills

osd_max_backfills = 1

Do I have to really restart all the OSD daemons ?

Karun 

_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com

_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com