Sounds like they aren't handling the transition very well when trying to 
calculate old OSDs which might have held the PG. Are you trying to salvage the 
data that was in it, or can you throw it away?  
Can you post the backtrace they're producing?
-Greg
Software Engineer #42 @ http://inktank.com | http://ceph.com


On Wednesday, April 10, 2013 at 3:59 AM, Witalij Poljatchek wrote:

> Hello,
>  
> need help to solve segfault on all osd in my test cluster.
>  
>  
> Setup ceph from scratch.
> service ceph -a start
>  
> ceph -w
> health HEALTH_OK
> monmap e1: 3 mons at 
> {1=10.200.20.1:6789/0,2=10.200.20.2:6789/0,3=10.200.20.3:6789/0}, election 
> epoch 6, quorum 0,1,2 1,2,3
> osdmap e5: 4 osds: 4 up, 4 in
> pgmap v305: 960 pgs: 960 active+clean; 0 bytes data, 40147 MB used, 26667 GB 
> / 26706 GB avail
> mdsmap e1: 0/0/1 up
>  
>  
> if i set replica size to 0 "i know this make no sense"  
> ceph osd pool set rbd size 0
> and then back to 2
> ceph osd pool set rbd size 2
>  
> then i see that on all OSDs the process ceph-osd crash with segfault
>  
> If i stop MONs daemons then i can start OSDs but if i start MONs back then 
> die all OSDs again.
>  
>  
>  
> How i cann repair this behavior ?
>  
>  
>  
>  
>  
> My setup
> Nothing specials:
>  
> Centos 6.3
>  
> Kernel: 3.8.3-1.el6.elrepo.x86_64
>  
> ceph-fuse-0.56.4-0.el6.x86_64
> ceph-test-0.56.4-0.el6.x86_64
> libcephfs1-0.56.4-0.el6.x86_64
> ceph-0.56.4-0.el6.x86_64
> ceph-release-1-0.el6.noarch
>  
> cat /etc/ceph/ceph.conf  
>  
> [global]
> auth cluster required = none
> auth service required = none
> auth client required = none
> keyring = /etc/ceph/$name.keyring
> [mon]
> [mds]
> [osd]
> osd journal size = 10000
> [mon.1]
> host = ceph-mon1
> mon addr = 10.200.20.1:6789
> [mon.2]
> host = ceph-mon2
> mon addr = 10.200.20.2:6789
> [mon.3]
> host = ceph-mon3
> mon addr = 10.200.20.3:6789
>  
> [osd.0]
> host = ceph-osd1
> [osd.1]
> host = ceph-osd2
> [osd.2]
> host = ceph-osd3
> [osd.3]
> host = ceph-osd4
>  
> [mds.a]
> host = ceph-mds1
> [mds.b]
> host = ceph-mds2
> [mds.c]
> host = ceph-mds3
>  
> Thanks much.
> -- AIXIT GmbH - Witalij Poljatchek (T) +49 69 203 4709-13 - (F) +49 69 203 
> 470 979 [email protected]  - http://www.aixit.com AIXIT GmbH Strahlenbergerstr. 
> 14 63067 Offenbach am Main (T) +49 69 203 470 913 Amtsgericht Offenbach, HRB 
> 43953 Geschäftsführer: Friedhelm Heyer, Holger Grauer  



_______________________________________________
ceph-users mailing list
[email protected]
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com

Reply via email to