[LU-3947] Unable to start OSTs when specifying multiple networks in NIDs leads.

• Type: Bug
• Resolution: Not a Bug
• Priority: Minor
• Affects Version/s: Lustre 2.4.1
• Environment: CentOS 6.4, Lustre 2.4.1-rc2
• Severity: 3
• Assignee: WC Triage
• Reporter: Shuichi Ihara (Inactive)

Description

We have an issue which seems very similar to LU-3445, when we specify multiple nids for the same none in tunefs.lustre we cannot start the OST.

This is happening at a customer site and we haven't been able to reproduce it in-house. We believe the tunefs command used to be

tunefs.lustre --erase-params --mgsnode=192.168.200.0@o2ib10,192.168.200.0@o2ib200 --mgsnode=192.168.200.1@o2ib10,192.168.200.1@o2ib100 --servicenode=192.168.202.0@o2ib10,192.168.202.0@o2ib200 --servicenode=192.168.202.1@o2ib10,192.168.202.1@o2ib100 --param ost.quota_type=ug /dev/mapper/ost_gsfs_80

The error when mounting the device is: LDISKFS-fs (dm-36): Unrecognized mount option "10.8.202.0@tcp100" or missing value

Here is the information from the device itself.

[root@gsfs-oss-015 ~]# tunefs.lustre --dryrun /dev/mapper/ost_gsfs_91
checking for existing Lustre data: found
Reading CONFIGS/mountdata
Read previous values:
Target: gsfs-OST005b
Index: 91
Lustre FS: gsfs
Mount type: ldiskfs
Flags: 0x1102
(OST writeconf no_primnode )
Persistent mount opts: errors=remount-ro
Parameters: mgsnode=192.168.200.0@o2ib10,192.168.200.0@o2ib200 mgsnode=192.168.200.1@o2ib10,192.168.200.1@o2ib100 failover.node=192.168.202.15@o2ib10,192.168.202.15@o2ib100 failover.node=192.168.202.14@o2ib10,192.168.202.14@o2ib200 ost.quota_type=ug