On Sat, Nov 8, 2014 at 3:24 PM, Loic Dachary <loic@xxxxxxxxxxx> wrote: > > > On 09/11/2014 00:03, Gregory Farnum wrote: >> It's all about the disk accesses. What's the slow part when you dump historic and in-progress ops? > > This is what I see on g1 (6% iowait) Yeah, you're going to need to do some data collation (at least in your head). If it's consistent that one node has way more ops and a higher iowit than everybody else, it sounds to me like you've found your answer. If it doesn't, look at the historic ops and see if there are any patterns. I confess I don't recall what the reported status is when an op is waiting for promotes to occur; you'll probably want to check that out too and see how long that stage is taking. <snip> > Also when I ceph -w I see a new pgmap is created every second which is also not a good sign. That's normal, unless you've adjusted your config options to avoid it. -Greg _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
 |