Are you running ntpd? If so you may need to stop, run ntpdate, and restart ntpd. Sometimes if the clock is too far out of sync ntp won't update the time.
On May 5, 2013, at 8:52 AM, Varun Chandramouli <[email protected]> wrote: > Hi All, > > I have a cluster of 4 nodes with 1 mds, 3 mons and 4 osds. Whenever I do ceph > health or ceph -s, it shows a health warning saying clock skew detected in 2 > of the 3 mons. When I run a mapreduce application on the cluster, one of the > monitors crashes (the one in which the skew is not detected) soon after the > application is started. Sometimes the application completes, sometimes, it > fails. I would like to know what this warning means. Is it responsible for > the failing of the application. If yes, how to remove the warning? > > Here is my ceph.conf: > > [global] > auth client required = none > auth cluster required = none > auth service required = none > > [osd] > osd journal data = 1000 > filestore xattr use omap = true > > [mon.a] > host = lnx147-73 > mon addr = 10.72.147.73:6789 > > [mon.b] > host = lnx148-20 > mon addr = 10.72.148.20:6789 > > [mon.c] > host = lnx-148-27 > mon addr = 10.72.148.27:6789 > > [mds.a] > host = lnx147-73 > > [osd.0] > host = lnx147-73 > > [osd.1] > host = lnx148-20 > > [osd.2] > host = lnx-148-27 > > [osd.3] > host = ln148-28 > > I can mail the mon logs and the output of ceph -w for the duration of the > application. > > Regards > Varun > _______________________________________________ > ceph-users mailing list > [email protected] > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
_______________________________________________ ceph-users mailing list [email protected] http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
