Re: Dual Role Nodes

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



I've been running a cluster like this for several years serving VM block devices and CephFS with no difficulty.  As previously mentioned it's a question of resources and the corresponding expectations.  I use a set of original HP Microservers each with 16GB RAM which is more than enough, although it certainly won't break any speed records.

One aspect to consider is the CRUSH map.  Due to the small number of nodes having one go down takes a large portion of the OSDs out.  Depending on the layout this could generate a lot of rebalancing load.  To avoid this I use three replicas but require them to be on separate nodes.  Thus one node down only causes degraded pgs, but with a pool min_size of 2 things are still functional.

I wouldn't build a large production cluster in this manner, but it does work and the cluster can grow out of it later.

Some process information for reference:
USER       PID %CPU %MEM    VSZ   RSS TTY      STAT START   TIME COMMAND
ceph     13006  1.7  0.5 486588 89180 ?        Ssl  Apr29 179:39 /usr/bin/ceph-mon
ceph     13532 13.9  2.9 1645784 478672 ?      Ssl  Apr29 1398:26 /usr/bin/ceph-osd
ceph     13607 21.5  3.7 1785448 616292 ?      Ssl  Apr29 2155:32 /usr/bin/ceph-osd
ceph     14074  0.1  1.4 631988 239140 ?       Ssl  Apr29  13:55 /usr/bin/ceph-mds

Steve Stock


On Fri, May 06, 2016 at 11:07:51AM +0000, Dacre, Alex wrote:
> That's good to know. My plan is to have three nodes initially; each with every role.
> 
> Thanks,
> Alex
> 
> -----Original Message-----
> From: ceph-users [mailto:ceph-users-bounces@xxxxxxxxxxxxxx] On Behalf Of Oliver Dzombic
> Sent: 06 May 2016 12:06
> To: ceph-users@xxxxxxxxxxxxxx
> Subject: Re:  Dual Role Nodes
> 
> Hi Alex,
> 
> thats no problem.
> 
> It might get harder to troubleshoot problems, because the serverressourced will be used for everything. Also you will have on the same node logs for every different service.
> 
> But in general, its no problem technically to have all ( Mon+Osd+MDS ) on one node.
> 
> --
> Mit freundlichen Gruessen / Best regards
> 
> Oliver Dzombic
> IP-Interactive
> 
> mailto:info@xxxxxxxxxxxxxxxxx
> 
> Anschrift:
> 
> IP Interactive UG ( haftungsbeschraenkt ) Zum Sonnenberg 1-3
> 63571 Gelnhausen
> 
> HRB 93402 beim Amtsgericht Hanau
> Geschäftsführung: Oliver Dzombic
> 
> Steuer Nr.: 35 236 3622 1
> UST ID: DE274086107
> 
> 
> Am 06.05.2016 um 13:00 schrieb Dacre, Alex:
> > Hi,
> > 
> >  
> > 
> > I'm looking at standing up a small Ceph cluster, but I'm curious as to 
> > what the general opinion is around having dual role nodes. E.g. each 
> > node is both an OSD and MON.
> > 
> >  
> > 
> > Regards,
> > 
> > Alex
> > 
> > 
> > 
> > _______________________________________________
> > ceph-users mailing list
> > ceph-users@xxxxxxxxxxxxxx
> > https://urldefense.proofpoint.com/v2/url?u=http-3A__lists.ceph.com_lis
> > tinfo.cgi_ceph-2Dusers-2Dceph.com&d=DQIF-g&c=O3LcjD-V2Iepl5V0N1424A&r=
> > MW8D_KLIjbkgPV4I3ExMPW-6rQA8xlUsAXjzeSmCE80&m=Zrz-dViIT_3QUcQUXdQGZEsH
> > av9rT228pj_fFI0a8oQ&s=NZ4hiYs315UX7MIGhTyabHpGqrFXEju-WBZjIVFeGUU&e=
> > 
> _______________________________________________
> ceph-users mailing list
> ceph-users@xxxxxxxxxxxxxx
> https://urldefense.proofpoint.com/v2/url?u=http-3A__lists.ceph.com_listinfo.cgi_ceph-2Dusers-2Dceph.com&d=DQIF-g&c=O3LcjD-V2Iepl5V0N1424A&r=MW8D_KLIjbkgPV4I3ExMPW-6rQA8xlUsAXjzeSmCE80&m=Zrz-dViIT_3QUcQUXdQGZEsHav9rT228pj_fFI0a8oQ&s=NZ4hiYs315UX7MIGhTyabHpGqrFXEju-WBZjIVFeGUU&e= 
> _______________________________________________
> 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




[Index of Archives]     [Information on CEPH]     [Linux Filesystem Development]     [Ceph Development]     [Ceph Large]     [Ceph Dev]     [Linux USB Development]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [xfs]


  Powered by Linux