[Lustre-discuss] MDS mount "cannot assign requested address"

[Mike Hanby]

Aha, my bad, I was specifying the MDS's own IPs as failover, removing those resolved the issue:

# tunefs.lustre --writeconf --erase-params --param="mdt.group_upcall=NONE" /dev/sdb

I need to do that on the failover, not the primary I guess.


-----Original Message-----
From: lustre-discuss-bounces at lists.lustre.org [mailto:lustre-discuss-bounces at lists.lustre.org] On Behalf Of Mike Hanby
Sent: Wednesday, February 02, 2011 11:31 AM
To: lustre-discuss at lists.lustre.org
Subject: Re: [Lustre-discuss] MDS mount "cannot assign requested address"

Here are the entries from syslog

Feb  2 11:14:21 lustre-mds-0-0 kernel: kjournald starting.  Commit interval 5 seconds
Feb  2 11:14:21 lustre-mds-0-0 kernel: LDISKFS FS on sdb, internal journal
Feb  2 11:14:21 lustre-mds-0-0 kernel: LDISKFS-fs: mounted filesystem with ordered data mode.
Feb  2 11:14:21 lustre-mds-0-0 kernel: kjournald starting.  Commit interval 5 seconds
Feb  2 11:14:21 lustre-mds-0-0 kernel: LDISKFS FS on sdb, internal journal
Feb  2 11:14:21 lustre-mds-0-0 kernel: LDISKFS-fs: mounted filesystem with ordered data mode.
Feb  2 11:14:21 lustre-mds-0-0 kernel: Lustre: MGS MGS started
Feb  2 11:14:21 lustre-mds-0-0 kernel: Lustre: MGC192.168.2.20 at o2ib: Reactivating import
Feb  2 11:14:21 lustre-mds-0-0 kernel: Lustre: Denying initial registration attempt from nid 192.168.2.20 at o2ib, specified as failover
Feb  2 11:14:21 lustre-mds-0-0 kernel: LustreError: 8493:0:(obd_mount.c:1097:server_start_targets()) Required registration failed for lustre-MDT0000: -99
Feb  2 11:14:21 lustre-mds-0-0 kernel: LustreError: 8493:0:(obd_mount.c:1655:server_fill_super()) Unable to start targets: -99
Feb  2 11:14:21 lustre-mds-0-0 kernel: LustreError: 8493:0:(obd_mount.c:1438:server_put_super()) no obd lustre-MDT0000
Feb  2 11:14:21 lustre-mds-0-0 kernel: LustreError: 8493:0:(obd_mount.c:147:server_deregister_mount()) lustre-MDT0000 not registered
Feb  2 11:14:22 lustre-mds-0-0 kernel: Lustre: MGS has stopped.
Feb  2 11:14:22 lustre-mds-0-0 kernel: Lustre: server umount lustre-MDT0000 complete
Feb  2 11:14:22 lustre-mds-0-0 kernel: LustreError: 8493:0:(obd_mount.c:2050:lustre_fill_super()) Unable to mount  (-99)

-----Original Message-----
From: lustre-discuss-bounces at lists.lustre.org [mailto:lustre-discuss-bounces at lists.lustre.org] On Behalf Of Mike Hanby
Sent: Wednesday, February 02, 2011 11:06 AM
To: lustre-discuss at lists.lustre.org
Subject: [Lustre-discuss] MDS mount "cannot assign requested address"

Howdy,

We recently changed our ip scheme for our cluster. I disconnected all of the clients, ost's, and unmounted the combined mdt/mgs.

changed the ip addresses for eth0, ib0 and ib1 on the MDS, and changed the LNET parameters in modprobe.ddn to reflect the new ip's.

Next I ran the tunefs.lustre to make the changes (see below for the full syntax and output).

Now, when I attempt to mount the mdt/mgs I get:

# mount /mnt/mdt-mtpt 
mount.lustre: mount /dev/sdb at /mnt/mdt-mtpt failed: Cannot assign requested address

I've tried a reboot, thinking maybe a port might be in use, still the same problem. I've searched the docs and found the error message in the source code, but no useful tech docs. Have you encountered this or do you have any suggestions?

Here's the tunefs command I ran to change the nids to our new IP scheme:

# hostname
lustre-mds-0-0

# lctl list_nids
192.168.2.20 at o2ib
192.168.3.20 at o2ib1
172.20.0.20 at tcp

# tunefs.lustre --writeconf --erase-params --param="failover.node=192.168.2.20 at o2ib" --param="failover.node=192.168.3.20 at o2ib1" --param="failover.node=172.20.0.20 at tcp" --param="mdt.group_upcall=NONE" /dev/sdb
checking for existing Lustre data: found CONFIGS/mountdata
Reading CONFIGS/mountdata

   Read previous values:
Target:     lustre-MDT0000
Index:      0
Lustre FS:  lustre
Mount type: ldiskfs
Flags:      0x545
              (MDT MGS update writeconf )
Persistent mount opts: errors=remount-ro,iopen_nopriv,user_xattr
Parameters: failover.node=192.168.2.20 at o2ib failover.node=192.168.3.20 at o2ib1 failover.node=172.20.0.20 at tcp mdt.group_upcall=NONE

# ifconfig
eth0      Link encap:Ethernet  HWaddr 84:2B:xxxx  
          inet addr:172.20.0.20  Bcast:172.20.255.255  Mask:255.255.0.0

ib0       Link encap:InfiniBand  HWaddr 80:00:xxxxx
          inet addr:192.168.2.20  Bcast:192.168.2.255  Mask:255.255.255.0

ib1       Link encap:InfiniBand  HWaddr 80:00:xxxx
          inet addr:192.168.3.20  Bcast:192.168.3.255  Mask:255.255.255.0

=================================
Mike Hanby
mhanby at uab.edu
UAB School of Engineering
Information Systems Specialist II
IT HPCS / Research Computing


_______________________________________________
Lustre-discuss mailing list
Lustre-discuss at lists.lustre.org
http://lists.lustre.org/mailman/listinfo/lustre-discuss
_______________________________________________
Lustre-discuss mailing list
Lustre-discuss at lists.lustre.org
http://lists.lustre.org/mailman/listinfo/lustre-discuss