On Wednesday 25 of May 2011 17:58:24 Sage Weil wrote: > On Wed, 25 May 2011, Michal Humpula wrote: > > Hi, > > > > I've just compiled the recent version of ceph v0.28.1. I wanted to give > > it a try on a setup of 8 machines (1 core Opterons). The design is 3 > > mons, 2 mds and 10 osds. Running latest vanilla kernel 2.6.29 with btrfs > > as a filestore. > > > > After initial setup, ceph -w is printing a lot of these > > > > 2011-05-25 12:19:38.720110 log 2011-05-25 12:19:37.671620 osd5 > > 192.168.0.23:6803/18299 151 : [WRN] map e3577 wrongly marked me down or > > wrong addr > > 2011-05-25 12:19:18.398004 log 2011-05-25 12:19:18.333978 mon0 > > 192.168.0.10:6789/0 3968 : [INF] osd6 192.168.0.24:6800/2830 failed (by > > osd7 192.168.0.24:6801/3064) > > > > I've tried to setup > > > > osd op threads = 1 > > > > and > > > > osd op threads = 0 > > > > but it doesn't seem to have any impact on a number of cosd daemon > > threads. Still seeing +20 of them and still seeing messages about > > degradation. > > > > Any hint, what to setup differently, would be appreciated. > > If you can reproduce this behavior with > > debug osd = 10 > debug ms = 10 > > in your [osd] section it should have enough information for us to identify > the problem. > > Thanks! > sage I've got a little problem to harvest the logs, but here they come. The nodes are PXE booting, so I had to store them directly on the data disks, because it was not possible to transfer them over NFS at this rate. Logs are accessible at http://81.91.83.58/ceph/. There is also a config which was used and few lines from "ceph -w". At the time of testing node4 was down. Michael -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
