On Tue, 23 Nov 2010, Jim Schutt wrote: > Hi Sage, > > On Tue, 2010-11-23 at 00:11 -0700, Sage Weil wrote: > > Hi Jim, > > > > On Fri, 19 Nov 2010, Jim Schutt wrote: > > > I've just created a brand-new filesystem using current unstable > > > branch. > > > > > > ceph -w shows me this after I start it up and it settles down,: > > > > > > 2010-11-19 13:07:39.279045 pg v247: 3432 pgs: 3432 active; 54 KB data, 98200 KB used, 3032 GB / 3032 GB avail; 95/108 degraded (87.963%) > > > 2010-11-19 13:07:39.532174 pg v248: 3432 pgs: 3432 active; 54 KB data, 98232 KB used, 3032 GB / 3032 GB avail; 95/108 degraded (87.963%) > > > 2010-11-19 13:07:41.123789 pg v249: 3432 pgs: 3432 active; 54 KB data, 98416 KB used, 3032 GB / 3032 GB avail; 95/108 degraded (87.963%) > > > > There were some issues in unstable that were preventing the recovery from > > completing. They should be sorted out in the current git. > > Thanks for taking a look. > > FWIW, as of c327c6a2064f I can still reproduce this. > My recipe is: build a filesystem with 7 monitor instances, > 7 mds instances; 13 osd instances. Start all the mon > instances with a pdsh; start all the mds instances with a pdsh; > start the osd instances one-by-one, with a few seconds between > starting instances. Okay, I just fixed a number of issues and have this working on 24 nodes. Just pushed it all to the unstable branch. Let us know if you see any remaining osd recovery problems. Thanks! sage > > Let me know if there's anything else I can do. > > -- Jim > > > > > Thanks! > > sage > > > > > > > > > > That output seems to come from PGMap::print_summary(). > > > If so, it seems to be telling me I have 108 objects, > > > of which 95 are degraded. > > > > > > If so, why would I have any degraded objects on > > > a brand-new file system? All my osds are up/in; > > > shouldn't any degraded objects have been recovered? > > > > > > Note that I haven't even mounted it anywhere yet. > > > > > > Also, the above result is after starting > > > each of my 13 osds one at a time, waiting for > > > the PGs for each osd to go active before > > > starting up the next osd. > > > > > > If I start up all the cosds a newly created file system > > > roughly simultaneously, using pdsh, I get 7/108 objects > > > degraded. > > > > > > What am I missing? > > > > > > How can I learn what objects are degraded? > > > > > > Thanks -- Jim > > > > > > > > > > > > -- > > > To unsubscribe from this list: send the line "unsubscribe ceph-devel" in > > > the body of a message to majordomo@xxxxxxxxxxxxxxx > > > More majordomo info at http://vger.kernel.org/majordomo-info.html > > > > > > > > -- > > To unsubscribe from this list: send the line "unsubscribe ceph-devel" in > > the body of a message to majordomo@xxxxxxxxxxxxxxx > > More majordomo info at http://vger.kernel.org/majordomo-info.html > > > > > -- > To unsubscribe from this list: send the line "unsubscribe ceph-devel" in > the body of a message to majordomo@xxxxxxxxxxxxxxx > More majordomo info at http://vger.kernel.org/majordomo-info.html > > -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
