On Mon, 14 Mar 2016 23:38:24 -0700 Blade Doyle wrote: > On Mon, Mar 14, 2016 at 3:48 PM, Christian Balzer <chibi@xxxxxxx> wrote: > > > > > Hello, > > > > On Mon, 14 Mar 2016 09:16:13 -0700 Blade Doyle wrote: > > > > > Hi Ceph Community, > > > > > > I am trying to use "ceph -w" output to monitor my ceph cluster. The > > > basic setup is: > > > > > > A python script runs ceph -w and processes each line of output. It > > > finds the data it wants and reports it to InfluxDB. I view the data > > > using Grafana, and Ceph Dashboard. > > > > > > > A much richer and more precise source of information would be the > > various performance counters and using collectd to feed them into > > graphite and friends. > > http://lists.ceph.com/pipermail/ceph-users-ceph.com/2014-May/039953.html > > > > I'm using the DWM one, YMMV. > > > > Thanks much for your reply, Christian. > > Ugh. Ok, then it looks like the key info here is to get the data from > the osd/mon sockets. Forgive me for not digging too deep yet, but it > looks like I would do something like: > > ceph --admin-daemon /var/run/ceph/ceph-osd.4.asok perf dump > Correct. > * which of that data is read/write bytes? More than one choice, the obvious ones are "counter-osd_op_out_bytes" and "counter-osd_op_in_bytes". This is why collectd with graphite is so much fun, you just click and drool until the data (graph) makes sense. > * Is that data for the entire cluster, or just that osd? (would I need > to read data from each individual osd sock in the cluster?) > Indeed the later, the mons don't keep track of this. Christian > Thanks, > Blade. -- Christian Balzer Network/Systems Engineer chibi@xxxxxxx Global OnLine Japan/Rakuten Communications http://www.gol.com/ _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
 |