On 05/06/2013 12:14 PM, John Wilkins wrote:
Guido, My apologies. I seem to have omitted the PG troubleshooting section from the index. It has been addressed. See http://ceph.com/docs/master/rados/troubleshooting/troubleshooting-pg/ Ceph OSDs peer and check on each other. So running a cluster with only one OSD is not recommended. Operationally, it's perfectly fine to bootstrap a cluster that way, but an operating cluster should have at least two OSDs running. See http://ceph.com/docs/master/rados/operations/monitoring-osd-pg/#peering and http://ceph.com/docs/master/rados/configuration/mon-osd-interaction/ to learn how OSDs interact with each other and monitors. Regards, John
Semi-related, it would be nice (but not strictly critical) if we could distinguish between health warnings that prevented operation in a single-osd configuration. It would make osd performance testing a bit cleaner.
Mark
On Mon, May 6, 2013 at 8:04 AM, Guido Winkelmann <guido-ceph@xxxxxxxxxxxxxxxxx <mailto:guido-ceph@xxxxxxxxxxxxxxxxx>> wrote: Am Montag, 6. Mai 2013, 16:59:12 schrieb Wido den Hollander: > On 05/06/2013 04:51 PM, Guido Winkelmann wrote: > > Am Montag, 6. Mai 2013, 16:41:43 schrieb Wido den Hollander: > >> On 05/06/2013 04:15 PM, Guido Winkelmann wrote: > >>> Am Montag, 6. Mai 2013, 16:05:31 schrieb Wido den Hollander: > >>>> On 05/06/2013 04:00 PM, Guido Winkelmann wrote: > >>>>> Hi, > >>>>> > >>>>> How do I run a 1-node cluster with no replication? > >>>>> > >>>>> I'm trying to run a small 1-node cluster on my local workstation and > >>>>> another on my notebook for experimentation/development purposes, but > >>>>> since I only have on OSD, I'm always getting HEALTH_WARN as the > >>>>> cluster > >>>>> status from ceph -s. Can I somehow tell ceph to just not bother with > >>>>> replication for this cluster? > >>>> > >>>> Have you set min_size to 1 for all the pools? > >>> > >>> You mean in the crushmap? > >> > >> No, it's pool setting. > >> > >> See: http://ceph.com/docs/master/rados/operations/pools/#set-pool-values > > > > Hm, I set that to 1 now, and nothing changed: > Have you also set "size" to 1? Meaning no replication. > > Both size and min_size should be set to 1. I set size to 1 now, too. ceph -s no longer reports degraded pgs now, but I still get a HEALTH_WARN: $ ceph -s health HEALTH_WARN 384 pgs stuck unclean _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx <mailto:ceph-users@xxxxxxxxxxxxxx> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com -- John Wilkins Senior Technical Writer Intank john.wilkins@xxxxxxxxxxx <mailto:john.wilkins@xxxxxxxxxxx> (415) 425-9599 http://inktank.com _______________________________________________ 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
