Roger Sersted
2014-Jan-10 02:43 UTC
Tunefs.lustre changed number in target name to "=00000"
Lustre 2.4.2 on CentOS 6.4 This was an attempted upgrade from Lustre 1.8.9. When Trying to mount the MDS, I ran into bug LU-4460. That bug prevents mounting the MDS if you have multiple NIDs listed for the failover parameter. After some research, I ran tunefs.lustre --erase-params /dev/sds2 I was able to mount it. I unmounted it and ran tunefs.lustre -write-conf /dev/sds2 Then proceeded to run tunefs.lustre -write-conf on each OST on its respective OSS. No Lustre MDT or OSTs were mounted. I then noticed (far too late) that this was happening: [root@myoss08 ~]# tunefs.lustre -write-conf /dev/mapper/mpathh checking for existing Lustre data: found Reading CONFIGS/mountdata Read previous values: Target: lustrefc-OST000b <<<< good value >>>> Index: 11 Lustre FS: lustrefc Mount type: ldiskfs Flags: 0x402 (OST ) Persistent mount opts: errors=remount-ro,extents,mballoc Parameters: mgsnode=172.17.1.112@o2ib,172.16.1.112@tcp mgsnode=172.17.1.113@o2ib,172.17.1.113@tcp failover.node=172.17.1.107@o2ib,172.16.1.107@tcp Permanent disk data: Target: lustrefc=OST0000 <<<< BAD value >>>> Index: 0 Lustre FS: lustrefc Mount type: ldiskfs Flags: 0x502 (OST writeconf ) Persistent mount opts: errors=remount-ro,extents,mballoc Parameters: mgsnode=172.17.1.112@o2ib,172.16.1.112@tcp mgsnode=172.17.1.113@o2ib,172.17.1.113@tcp failover.node=172.17.1.107@o2ib,172.16.1.107@tcp Writing CONFIGS/mountdata How do I possibly fix this? I know the correct OST index values. Thanks, Roger S.