Minor Description
While attempting to update the llogs on grove to pick up the missing failover configuration I performed the following procedure.
1) Stop all the OSTs
2) Restart all the OSTs with the 'writeconf' option to regenerate the MDT and client llogs. The correct failover nids were already stored in zfs properties on the data set so they just needed to be registered with the MGS.
mount -t lustre -o writeconf grove641/ost0 /mnt/lustre/local/ls1-OST0281
Initially this appeared to work however the client log ended up with duplicate indexes resulting in the client no longer being able to mount. It may be because for some reason the full UUID for the server wasn't passed during registration.
— client log —
#40 (128)lov_modify_tgts add 0:lstest-clilov 1:lstest-OST0281_UUID 2:641 3:1
#807 (112)lov_modify_tgts add 0:lstest-clilov 1:_UUID 2:641 3:1
— mount errors —
LustreError: 11-0: MGC172.20.5.2@o2ib500: Communicating with 172.20.5.2@o2ib500, operation llog_origin_handle_create failed with -2
LustreError: 14218:0:(mgc_request.c:250:do_config_log_add()) failed processing sptlrpc log: -2
LustreError: 14230:0:(lov_obd.c:548:lov_add_target()) UUID lstest-OST0281_UUID already assigned at LOV target index 641
LustreError: 14230:0:(obd_config.c:1408:class_config_llog_handler()) Err -17 on cfg command:
Lustre: cmd=cf00d 0:lstest-clilov 1:_UUID 2:641 3:1
LustreError: 15c-8: MGC172.20.5.2@o2ib500: The configuration from log 'lstest-client' failed (-17). This may be the result of communication errors between this node and the MGS, a bad configuration, or other errors. See the syslog for more information.
LustreError: 14218:0:(llite_lib.c:975:ll_fill_super()) Unable to process log: -17
LustreError: 14101:0:(lov_obd.c:888:lov_cleanup()) lov tgt 641 not cleaned! deathrow=0, lovrc=1
LustreError: 14101:0:(lov_obd.c:888:lov_cleanup()) Skipped 127 previous similar messages