Hello, On Mon, 19 Dec 2016 12:31:44 -0700 Ken Dreyer wrote: > On Tue, Dec 13, 2016 at 4:42 AM, Francois Lafont > <francois.lafont.1978@xxxxxxxxx> wrote: > > So, now with 10.2.5 version, in my process, OSD daemons are stopped, > > then automatically restarted by the upgrade and then stopped again > > by the reboot. This is not an optimal process of course. ;) > > We do not intend for anything in the packaging to restart the daemons. > I think you're fixating wrongly on the "restart" bit. What Francois wrote and what most Ceph admins do is to manually stop demons before starting the upgrade process. And what he didn't want nor expected was for Ceph to start these demons up after/during the upgrade. I think this might be related to the new feature of Ceph to restart demons if they failed. Something that's nice enough, but ought to be configurable for people who do NOT want to trawl log files to see if something had a hiccup. The only time OSD or MON demons died on me was during brutal tests on my vastly underpowered (especially RAM) alpha-test cluster. Christian > The last time I looked into this issue, it behaved correctly (dpkg did > not restart the daemons during the apt-get process - the PID files > were the same before and after the upgrade). > > Did you dig further to find out what is restarting them? Are you using > any configuration management system (ansible, chef, puppet) to do your > package upgrades? > > - Ken > _______________________________________________ > ceph-users mailing list > ceph-users@xxxxxxxxxxxxxx > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com > -- 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
 |