For your information, we found the bug. It only impacts robinhood running on Lustre 2.7 clients, so your reinstallation to 2.5 should be OK.
The fix is under review and validation and will be available in the next few days. Regards, Thomas On 06/16/15 16:26, Frederik Ferner wrote: > On 16/06/15 15:10, DOREAU Henri wrote: >> Hello, >> >> all or part of the output of lfs changelog would be helpful too. > > [bnh65367@cs04r-sc-serv-92 ~]$ sudo lfs changelog lustre03-MDT0000 | head > 3435135938 01CREAT 10:40:34.318984739 2015.06.15 0x0 > t=[0x20001021c:0x27:0x0] p=[0x20000f248:0x8:0x0] test.h5 > 3435135939 01CREAT 10:44:40.706579875 2015.06.15 0x0 > t=[0x200010204:0x9d:0x0] p=[0x20000db0c:0x50c7:0x0] > 20150615_test10001_00000.cbf.tmp > 3435135940 14SATTR 10:44:40.763581198 2015.06.15 0x24 > t=[0x200010204:0x9d:0x0] > 3435135941 08RENME 10:44:40.773581405 2015.06.15 0x0 t=[0:0x0:0x0] > p=[0x20000db0c:0x50c7:0x0] 20150615_test10001_00000.cbf > s=[0x200010204:0x9d:0x0] sp=[0x20000db0c:0x50c7:0x0] > 20150615_test10001_00000.cbf.tmp > 3435135942 01CREAT 10:44:41.89588579 2015.06.15 0x0 > t=[0x200010204:0x9e:0x0] p=[0x20000db0c:0x50c7:0x0] > 20150615_test10001_00001.cbf.tmp > 3435135943 14SATTR 10:44:41.160590193 2015.06.15 0x24 > t=[0x200010204:0x9e:0x0] > 3435135944 01CREAT 10:44:41.689602213 2015.06.15 0x0 > t=[0x200010204:0x9f:0x0] p=[0x20000db0c:0x50c7:0x0] > 20150615_test10001_00002.cbf.tmp > 3435135945 01CREAT 10:44:42.248614976 2015.06.15 0x0 > t=[0x200010204:0xa0:0x0] p=[0x20000db0c:0x50c7:0x0] > 20150615_test10001_00003.cbf.tmp > 3435135946 01CREAT 10:44:42.649623971 2015.06.15 0x0 > t=[0x200010204:0xa1:0x0] p=[0x20000db0c:0x50c7:0x0] > 20150615_test10001_00004.cbf.tmp > 3435135947 08RENME 10:44:42.726625753 2015.06.15 0x0 t=[0:0x0:0x0] > p=[0x20000db0c:0x50c7:0x0] 20150615_test10001_00001.cbf > s=[0x200010204:0x9e:0x0] sp=[0x20000db0c:0x50c7:0x0] > 20150615_test10001_00001.cbf.tmp > [bnh65367@cs04r-sc-serv-92 ~]$ > > > the whole changelog is quite big by now as robinhood has been down for > about a day... > >> Do you have in your logs a line like: >> "warning: llapi_changelog_start() called w/o CHANGELOG_FLAG_JOBID"? > > Not on the machine running robinhood, robinhood logs, and not on the MDS. > > Let me know if there is anything else I can provide. > > At the moment I'm tempted to revert that one machine to a 2.5 client > (and install a robinhood version for lustre 2.5) but I'm not sure if > this will remove my ability to reproduce the issue, so I'm currently > still holding back. > > Kind regards, > Frederik > ------------------------------------------------------------------------------ _______________________________________________ robinhood-support mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/robinhood-support
