On Sat, Aug 19, 2017 at 1:39 AM, Mingliang LIU
<mingliang.liu@xxxxxxxxxxxxxx> wrote:
> Hi all,
>
> I have a quick question about the RBD kernel module - how to best collect
> the metrics or perf numbers? The command 'ceph -w' does print some useful
> event logs of cluster-wide while I'm interested in
> per-client/per-image/per-volume read/write bytes and latency etc.
>
> For the librbd, I found the admin socket is very handy. However, this seems
> not an option for krbd kernel module (correct me if I'm wrong).
>
> I found when loading one kernel module or mapping an image, there are some
> related system message readable via dmesg. But I did not find any logs about
> read/write.
>
> Any pointers?
rbd kernel module presents a standard block device, so you can use
anything you would use on any other block device: iostat, various sysfs
files, blktrace, etc. There is nothing admin socket like for krbd.
Thanks,
Ilya
_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
