Re: Testing Performance

[Dan Mick <dan.mick@xxxxxxxxxxx>]

No.  Shell loops work though:

for o in $(rados -p data ls); do rados -p data rm $o; done

or you can remove and recreate the pool.

On 03/01/2013 04:43 AM, femi anjorin wrote:
> I think this command is good
>
> # rados -p data rm <obj-name> ...                remove object(s)
>
> but is it possible to
>
> #rados -p data rm "all objects"  ????
>
> just like rm -f
>
>
>
> Regards,
>
> Femi.
>
>
>
>
> On Fri, Mar 1, 2013 at 12:09 PM, femi anjorin <femi.anjorin@xxxxxxxxx> wrote:
> > Hi ,
> >
> > When i Started test on ceph cluster today:
> >
> >
> > #ceph -w
> >
> > 2013-03-01 08:57:43.990529 mon.0 [INF] pgmap v7601: 6400 pgs: 6400
> > active+clean;6849 MB data, 997 GB used, 674 TB / 675 TB avail;
> > 5261KB/s wr, 5op/s
> > 2013-03-01 08:57:44.895664 mon.0 [INF] pgmap v7602: 6400 pgs: 6400
> > active+clean;6859 MB data, 997 GB used, 674 TB / 675 TB avail; 0B/s
> > rd, 8257KB/s wr, 267op/s
> > 2013-03-01 08:57:47.374271 mon.0 [INF] pgmap v7603: 6400 pgs: 6400
> > active+clean;6901 MB data, 997 GB used, 674 TB / 675 TB avail; 0B/s
> > rd, 36514KB/s wr, 2099op
> >                                                                /s
> >
> > 2013-03-01 08:57:49.801775 mon.0 [INF] pgmap v7604: 6400 pgs: 6400
> > active+clean; 6913 MB data, 997 GB used, 674 TB / 675 TB avail;0B/s
> > rd, 5575KB/s wr, 430op/s
> > 2013-03-01 08:57:51.833305 mon.0 [INF] pgmap v7605: 6400 pgs: 6400
> > active+clean; 6933 MB data, 998 GB used, 674 TB / 675 TB avail;0B/s
> > rd, 8741KB/s wr, 172op/s
> > 2013-03-01 08:57:53.170430 mon.0 [INF] pgmap v7606: 6400 pgs: 6400
> > active+clean; 6943 MB data, 998 GB used, 674 TB / 675 TB avail;
> > 7056KB/s wr, 7op/s
> > 2013-03-01 08:57:54.914054 mon.0 [INF] pgmap v7607: 6400 pgs: 6400
> > active+clean; 6945 MB data, 998 GB used, 674 TB / 675 TB avail;
> > 1532KB/s wr, 2op/s
> > 2013-03-01 08:57:55.769889 mon.0 [INF] pgmap v7608: 6400 pgs: 6400
> > active+clean; 6967 MB data, 998 GB used, 674 TB / 675 TB avail;
> > 12922KB/s wr, 12op/s
> > 2013-03-01 08:57:57.393463 mon.0 [INF] pgmap v7609: 6400 pgs: 6400
> > active+clean; 7001 MB data, 998 GB used, 674 TB / 675 TB avail;
> > 29706KB/s wr, 83op/s
> > 2013-03-01 08:57:58.490561 mon.0 [INF] pgmap v7610: 6400 pgs: 6400
> > active+clean; 7006 MB data, 998 GB used, 674 TB / 675 TB avail;
> > 3866KB/s wr, 3op/s
> > 2013-03-01 08:57:59.754231 mon.0 [INF] pgmap v7611: 6400 pgs: 6400
> > active+clean; 7009 MB data, 998 GB used, 674 TB / 675 TB avail;
> > 2607KB/s wr, 2op/s
> > 2013-03-01 08:58:00.826096 mon.0 [INF] pgmap v7612: 6400 pgs: 6400
> > active+clean; 7023 MB data, 998 GB used, 674 TB / 675 TB avail;
> > 12965KB/s wr, 89op/s
> >
> > NOTICE: data started with 6913 MB
> >
> > BUT after a series of write,read and delete operations
> >
> > #ceph -w
> >
> > 2013-03-01 11:52:29.568074 mon.0 [INF] pgmap v13814: 6400 pgs: 6400
> > active+clean; 91856 MB data, 1248 GB used, 674 TB / 675 TB avail
> > 2013-03-01 11:52:30.762315 mon.0 [INF] pgmap v13815: 6400 pgs: 6400
> > active+clean; 91856 MB data, 1248 GB used, 674 TB / 675 TB avail
> > 2013-03-01 11:52:31.907384 mon.0 [INF] pgmap v13816: 6400 pgs: 6400
> > active+clean; 91856 MB data, 1248 GB used, 674 TB / 675 TB avail
> > 2013-03-01 11:52:33.076477 mon.0 [INF] pgmap v13817: 6400 pgs: 6400
> > active+clean; 91856 MB data, 1248 GB used, 674 TB / 675 TB avail
> > 2013-03-01 11:52:34.258712 mon.0 [INF] pgmap v13818: 6400 pgs: 6400
> > active+clean; 91856 MB data, 1248 GB used, 674 TB / 675 TB avail
> > 2013-03-01 11:52:36.003697 mon.0 [INF] pgmap v13819: 6400 pgs: 6400
> > active+clean; 91856 MB data, 1248 GB used, 674 TB / 675 TB avail
> >
> >
> >
> > Question:
> > data has increased to 91856 MB
> > Although i have done a delete operation, after all write+read operations.
> > Is there any rados utility command to verify what exactly these 91856
> > MB datas are? and what command is good to clean up the cluster without
> > re-initializing the disk all the time with mkcephfs command.????
> >
> > My intention is to ensure a clean cluster for each of the test cases
> > since i notice performance of the ceph cluster dropped after a few
> > write,read and delete operations using same system conditions but
> > different load clients.
> >
> >
> > Regards,
> > Femi.
> _______________________________________________
> 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