Hello, On Tue, 19 Jul 2016 08:10:27 +0800 Pei Feng Lin wrote: > Dear Cephers: > > I have two questions that needs advice. > > 1) If there is a OSD disk failure (for example, pulling disk out), how long > does the osd daemon detect the disk failure? and how long does the ceph > cluster mark this osd daemon down? > Can't test this now, but I'm pretty sure nothing will be detected until the daemon actually has to access the disk. So in a totally idle, quiescent cluster that could be until the next scrub or deep-scrub comes along... As for marking the daemon out, that should be happening quickly, as the affected daemon is going to kill itself when encountering this scenario. > Is there any config option to allow the ceph cluster to detect the status > of osd disk? > Dont't think so. And in normal situations (access to all OSDs every few seconds or so), not really needed either. > 2) If there is a journal disk failure, how long does the ceph cluster > detect the journal disk failure? > > Is there any config option to allow the ceph cluster to detect the status > of journal disk? > Same as above. Christian -- 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
 |