Hello,

This is an interesting topic and would like to know a solution to this
problem. Does that mean we should never use Dell storage as ceph storage
device?  I have similar setup with Dell 4 iscsi LUNs attached to openstack
controller and compute node in active-active situation.

As they were in active active 1 selected first 2 luns as osd on node1 and
last 2 as osd on node 2.

Is it ok to have this configuration specially when and node will be down or
considering live migration.

Regards
Gaurav Goyal
On 10-Jul-2016 9:02 pm, "Christian Balzer" <ch...@gol.com> wrote:

>
> Hello,
>
> On Sun, 10 Jul 2016 12:46:39 +0000 (UTC) William Josefsson wrote:
>
> > Hi everyone,
> >
> > I have problem with swapping drive and partition names on reboot. My
> > Ceph is Hammer on CentOS7, Dell R730 6xSSD (2xSSD OS RAID1 PERC,
> > 4xSSD=Journal drives), 18x1.8T SAS for OSDs.
> >
> > Whenever I reboot, drives randomly seem to change names. This is
> > extremely dangerous and frustrating when I've initially setup CEPH with
> > ceph-deploy, zap, prepare and activate. It has happened that I've
> > accidentally erased wrong disk too when e.g. /dev/sdX had
> > become /dev/sdY.
> >
> This isn't a Ceph specific question per se and you could probably keep
> things from moving around by enforcing module loads in a particular order.
>
> But that of course still wouldn't help if something else changed or a
> drive totally failed.
>
> So in the context of Ceph, it doesn't (shouldn't) care if the OSD (HDD)
> changes names, especially since you did set it up with ceph-deploy.
>
> And to avoid the journals getting jumbled up, do what everybody does
> (outside of Ceph as well), use /dev/disk/by-id or uuid.
>
> Like:
> ---
> # ls -la /var/lib/ceph/osd/ceph-28/
>
>  journal -> /dev/disk/by-id/wwn-0x55cd2e404b73d569-part3
> ---
>
> Christian
> > Please see an output below of how this drive swapping below appears SDC
> > is shifted, indexes and drive names got shuffled. Ceph OSDs didn't come
> > up properly.
> >
> > Please advice on how to get this corrected, with no more drive name
> > shuffling. Can this be due to the PERC HW raid? thx will
> >
> >
> >
> > POST REBOOT 2 (expected outcome.. with sda,sdb,sdc,sdd as journal. sdw
> > is a perc raid1)
> >
> >
> > [cephnode3][INFO  ] Running command: sudo /usr/sbin/ceph-disk list
> > [cephnode3][DEBUG ] /dev/sda :
> > [cephnode3][DEBUG ]  /dev/sda1 other,
> > ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG ]  /dev/sda2
> > other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG
> > ]  /dev/sda3 other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7
> > [cephnode3][DEBUG ]  /dev/sda4 other,
> > ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG ]  /dev/sda5
> > other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG
> > ] /dev/sdb : [cephnode3][DEBUG ]  /dev/sdb1 other,
> > ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG ]  /dev/sdb2
> > other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG
> > ]  /dev/sdb3 other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7
> > [cephnode3][DEBUG ]  /dev/sdb4 other,
> > ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG ]  /dev/sdb5
> > other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG
> > ] /dev/sdc : [cephnode3][DEBUG ]  /dev/sdc1 other,
> > ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG ]  /dev/sdc2
> > other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG
> > ]  /dev/sdc3 other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7
> > [cephnode3][DEBUG ]  /dev/sdc4 other,
> > ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG ]  /dev/sdc5
> > other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG
> > ] /dev/sdd : [cephnode3][DEBUG ]  /dev/sdd1 other,
> > ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG ]  /dev/sdd2
> > other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG
> > ]  /dev/sdd3 other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7
> > [cephnode3][DEBUG ]  /dev/sdd4 other,
> > ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG ]  /dev/sdd5
> > other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG
> > ] /dev/sde : [cephnode3][DEBUG ]  /dev/sde1 ceph data, active, cluster
> > ceph, osd.0 [cephnode3][DEBUG ] /dev/sdf : [cephnode3][DEBUG
> > ]  /dev/sdf1 ceph data, active, cluster ceph, osd.1 [cephnode3][DEBUG
> > ] /dev/sdg : [cephnode3][DEBUG ]  /dev/sdg1 ceph data, active, cluster
> > ceph, osd.2 [cephnode3][DEBUG ] /dev/sdh : [cephnode3][DEBUG
> > ]  /dev/sdh1 ceph data, active, cluster ceph, osd.3 [cephnode3][DEBUG
> > ] /dev/sdi : [cephnode3][DEBUG ]  /dev/sdi1 ceph data, active, cluster
> > ceph, osd.4 [cephnode3][DEBUG ] /dev/sdj : [cephnode3][DEBUG
> > ]  /dev/sdj1 ceph data, active, cluster ceph, osd.5 [cephnode3][DEBUG
> > ] /dev/sdk : [cephnode3][DEBUG ]  /dev/sdk1 ceph data, active, cluster
> > ceph, osd.6 [cephnode3][DEBUG ] /dev/sdl : [cephnode3][DEBUG
> > ]  /dev/sdl1 ceph data, active, cluster ceph, osd.7 [cephnode3][DEBUG
> > ] /dev/sdm : [cephnode3][DEBUG ]  /dev/sdm1 other, xfs
> > [cephnode3][DEBUG ] /dev/sdn :
> > [cephnode3][DEBUG ]  /dev/sdn1 ceph data, active, cluster ceph, osd.9
> > [cephnode3][DEBUG ] /dev/sdo :
> > [cephnode3][DEBUG ]  /dev/sdo1 ceph data, active, cluster ceph, osd.10
> > [cephnode3][DEBUG ] /dev/sdp :
> > [cephnode3][DEBUG ]  /dev/sdp1 ceph data, active, cluster ceph, osd.11
> > [cephnode3][DEBUG ] /dev/sdq :
> > [cephnode3][DEBUG ]  /dev/sdq1 ceph data, active, cluster ceph, osd.12
> > [cephnode3][DEBUG ] /dev/sdr :
> > [cephnode3][DEBUG ]  /dev/sdr1 ceph data, active, cluster ceph, osd.13
> > [cephnode3][DEBUG ] /dev/sds :
> > [cephnode3][DEBUG ]  /dev/sds1 ceph data, active, cluster ceph, osd.14
> > [cephnode3][DEBUG ] /dev/sdt :
> > [cephnode3][DEBUG ]  /dev/sdt1 ceph data, active, cluster ceph, osd.15
> > [cephnode3][DEBUG ] /dev/sdu :
> > [cephnode3][DEBUG ]  /dev/sdu1 ceph data, active, cluster ceph, osd.16
> > [cephnode3][DEBUG ] /dev/sdv :
> > [cephnode3][DEBUG ]  /dev/sdv1 ceph data, active, cluster ceph, osd.17
> > [cephnode3][DEBUG ] /dev/sdw :
> > [cephnode3][DEBUG ]  /dev/sdw1 other, xfs, mounted on /
> > [cephnode3][DEBUG ]  /dev/sdw2 swap, swap
> >
> >
> > POST REBOOT 1:
> >
> >
> > [cephnode3][DEBUG ] /dev/sda :
> > [cephnode3][DEBUG ]  /dev/sda1 other,
> > ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG ]  /dev/sda2
> > other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG
> > ]  /dev/sda3 other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7
> > [cephnode3][DEBUG ]  /dev/sda4 other,
> > ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG ]  /dev/sda5
> > other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG
> > ] /dev/sdb : [cephnode3][DEBUG ]  /dev/sdb1 other,
> > ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG ]  /dev/sdb2
> > other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG
> > ]  /dev/sdb3 other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7
> > [cephnode3][DEBUG ]  /dev/sdb4 other,
> > ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG ]  /dev/sdb5
> > other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG
> > ] /dev/sdc : [cephnode3][DEBUG ]  /dev/sdc1 ceph data, active, cluster
> > ceph, osd.13 [cephnode3][DEBUG ] /dev/sdd : [cephnode3][DEBUG
> > ]  /dev/sdd1 other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7
> > [cephnode3][DEBUG ]  /dev/sdd2 other,
> > ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG ]  /dev/sdd3
> > other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG
> > ]  /dev/sdd4 other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7
> > [cephnode3][DEBUG ]  /dev/sdd5 other,
> > ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG ] /dev/sde :
> > [cephnode3][DEBUG ]  /dev/sde1 other,
> > ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG ]  /dev/sde2
> > other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG
> > ]  /dev/sde3 other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7
> > [cephnode3][DEBUG ]  /dev/sde4 other,
> > ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG ]  /dev/sde5
> > other, ebd0a0a2-b9e5-4433-87c0-68b6b72699c7 [cephnode3][DEBUG
> > ] /dev/sdf : [cephnode3][DEBUG ]  /dev/sdf1 ceph data, active, cluster
> > ceph, osd.0 [cephnode3][DEBUG ] /dev/sdg : [cephnode3][DEBUG
> > ]  /dev/sdg1 ceph data, active, cluster ceph, osd.14 [cephnode3][DEBUG
> > ] /dev/sdh : [cephnode3][DEBUG ]  /dev/sdh1 ceph data, active, cluster
> > ceph, osd.1 [cephnode3][DEBUG ] /dev/sdi : [cephnode3][DEBUG
> > ]  /dev/sdi1 ceph data, active, cluster ceph, osd.2 [cephnode3][DEBUG
> > ] /dev/sdj : [cephnode3][DEBUG ]  /dev/sdj1 ceph data, active, cluster
> > ceph, osd.3 [cephnode3][DEBUG ] /dev/sdk : [cephnode3][DEBUG
> > ]  /dev/sdk1 ceph data, active, cluster ceph, osd.4 [cephnode3][DEBUG
> > ] /dev/sdl : [cephnode3][DEBUG ]  /dev/sdl1 ceph data, active, cluster
> > ceph, osd.15 [cephnode3][DEBUG ] /dev/sdm : [cephnode3][DEBUG
> > ]  /dev/sdm1 ceph data, active, cluster ceph, osd.5 [cephnode3][DEBUG
> > ] /dev/sdn : [cephnode3][DEBUG ]  /dev/sdn1 ceph data, active, cluster
> > ceph, osd.6 [cephnode3][DEBUG ] /dev/sdo :
> > [cephnode3][DEBUG ]  /dev/sdo1 ceph data, active, cluster ceph, osd.7
> > [cephnode3][DEBUG ] /dev/sdp :
> > [cephnode3][DEBUG ]  /dev/sdp1 other, xfs
> > [cephnode3][DEBUG ] /dev/sdq :
> > [cephnode3][DEBUG ]  /dev/sdq1 ceph data, active, cluster ceph, osd.9
> > [cephnode3][DEBUG ] /dev/sdr :
> > [cephnode3][DEBUG ]  /dev/sdr1 ceph data, active, cluster ceph, osd.16
> > [cephnode3][DEBUG ] /dev/sds :
> > [cephnode3][DEBUG ]  /dev/sds1 ceph data, active, cluster ceph, osd.10
> > [cephnode3][DEBUG ] /dev/sdt :
> > [cephnode3][DEBUG ]  /dev/sdt1 ceph data, active, cluster ceph, osd.11
> > [cephnode3][DEBUG ] /dev/sdu :
> > [cephnode3][DEBUG ]  /dev/sdu1 ceph data, active, cluster ceph, osd.12
> > [cephnode3][DEBUG ] /dev/sdv :
> > [cephnode3][DEBUG ]  /dev/sdv1 ceph data, active, cluster ceph, osd.17
> > [cephnode3][DEBUG ] /dev/sdw :
> > [cephnode3][DEBUG ]  /dev/sdw1 other, xfs, mounted on /
> > [cephnode3][DEBUG ]  /dev/sdw2 swap, swap
> > _______________________________________________
> > ceph-users mailing list
> > ceph-users@lists.ceph.com
> > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
> >
>
>
> --
> Christian Balzer        Network/Systems Engineer
> ch...@gol.com           Global OnLine Japan/Rakuten Communications
> http://www.gol.com/
> _______________________________________________
> ceph-users mailing list
> ceph-users@lists.ceph.com
> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
>
_______________________________________________
ceph-users mailing list
ceph-users@lists.ceph.com
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com

Reply via email to