[Lustre-discuss] Unable to mount OSTs

[Andrus, Brian Contractor]

I am focusing my effort on the simplest lustre filesystem consisting of a single OST.
I can mount the MDT/MGS and the OST with no apparent errors.

However, when a client is booting and trying to mount, I start getting a log entry on the MGS:
Dec 16 08:10:45 nas-0-0 kernel: LustreError: 5732:0:(ldlm_lib.c:1919:target_send_reply_msg()) @@@ processing error (-11)  req at ffff8103dfa61400 x1355301045403661/t0 o38-><?>@<?>:0/0 lens 368/0 e 0 to 0 dl 1292515945 ref 1 fl Interpret:/0/0 rc -11/0

The client loops giving LustreError: 11-0. The mds_connect operation failed with -11.

Brian Andrus

-----Original Message-----
From: Wang Yibin [mailto:wang.yibin at oracle.com] 
Sent: Thursday, December 16, 2010 6:38 AM
To: Andrus, Brian Contractor
Cc: lustre-discuss at lists.lustre.org
Subject: Re: [Lustre-discuss] Unable to mount OSTs

What does dmesg tell you after you mounted that particular OST?
You can find some clue there on the OSS and (maybe) MGS.

在 2010-12-16，下午10:30， Andrus, Brian Contractor 写道：

> Thank you for your assistance.
> I am almost there.
> I am able to get all the OSTs to mount, but when I do "lctl dl", one of them is not showing up for the work filesystem.
> 
> I can ping all OSSs from each other and once I have done that see them all in a peer_list. I have even tried mounting the OST on an OSS that already has another OST that is being recognized.
> 
> Brian Andrus
> 
> -----Original Message-----
> From: Wang Yibin [mailto:wang.yibin at oracle.com] 
> Sent: Thursday, December 16, 2010 12:29 AM
> To: Andrus, Brian Contractor
> Cc: Andreas Dilger; lustre-discuss at lists.lustre.org
> Subject: Re: [Lustre-discuss] Unable to mount OSTs
> 
> This error sounds like that there is some unrecognized parameter(s) on your OSTs.
> Please check your OST parameters. 
> You hit this issue mostly because we have some fix disallowing unrecognized parameters (22386).
> 
> 在 2010-12-16，下午3:56， Andrus, Brian Contractor 写道：
> 
>> Well e2fsck didn't help. Same error for those two (Function not
>> implemented).
>> I also see in the MGS/MDS log each time I try:
>> LustreError: 5974:0:(mgs_handler.c:487:mgs_handle_target_reg()) Failed
>> to write nscratch-OST0001 log (-38)
>> LustreError: 5973:0:(mgs_handler.c:487:mgs_handle_target_reg()) Failed
>> to write nscratch-OST0002 log (-38)
>> 
>> 
>> 
>> Brian Andrus
>> 
>> -----Original Message-----
>> From: lustre-discuss-bounces at lists.lustre.org
>> [mailto:lustre-discuss-bounces at lists.lustre.org] On Behalf Of Andrus,
>> Brian Contractor
>> Sent: Wednesday, December 15, 2010 11:48 PM
>> To: Andreas Dilger
>> Cc: lustre-discuss at lists.lustre.org
>> Subject: Re: [Lustre-discuss] Unable to mount OSTs
>> 
>> I actually did run writeconf on the MGS/MDS as well. 
>> I just did it again to test.
>> 
>> Followed the instructions in section 4.3.11 in the luster manual.
>> Mounting MDT goes fine.
>> Mounting the OSTs, the first just seems to be hanging. The other two
>> both gave:
>> [root at ost2 ‾]# mount -t lustre /dev/VG_lustre/scratch_ost2
>> /mnt/lustre/scratch/ost2/
>> mount.lustre: mount /dev/VG_lustre/scratch_ost2 at
>> /mnt/lustre/scratch/ost2 failed: Function not implemented
>> 
>> Currently running e2fsck on the two OSTs that didn't want to mount. I
>> have already run it, but doing so again, just in case.
>> 
>> Brian Andrus
>> 
>> -----Original Message-----
>> From: Andreas Dilger [mailto:andreas.dilger at oracle.com] 
>> Sent: Wednesday, December 15, 2010 10:55 PM
>> To: Andrus, Brian Contractor
>> Cc: lustre-discuss at lists.lustre.org
>> Subject: Re: [Lustre-discuss] Unable to mount OSTs
>> 
>> On 2010-12-15, at 22:50, Andrus, Brian Contractor wrote:
>>> I did change IPs of the NID for 2 or our lustre filesystems.
>>> Made appropriate changes to each ost. One thing I did that I probably
>> shouldn't have was doing a --writeconf on the OSTs. Now I see:
>>> 
>>> Flags:      0x142
>>>             (OST update writeconf )
>>> and I cannot mount them. I get:
>>> 
>>> mount.lustre: mount /dev/VG_lustre/scratch_ost1 at
>> /mnt/lustre/scratch/ost1 failed: Function not implemented
>>> however I do seem to be able to mount the OSTs that I didn't do
>> --writeconf on.
>> 
>> You can binary edit this flag in the "CONFIGS/mountdata" file on each
>> OST.
>> Alternately, you could follow the process in the manual, and run
>> writeconf on the MGS and MDS as well, and then the OSTs should be able
>> to register again.
>> 
>>> I see in syslog:
>>> LustreError: 6182:0:(obd_mount.c:1097:server_start_targets()) Required
>> registration failed for nscratch-OST0001: -38
>> 
>> Cheers, Andreas
>> --
>> Andreas Dilger
>> Lustre Technical Lead
>> Oracle Corporation Canada Inc.
>> 
>> _______________________________________________
>> 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
> 
> _______________________________________________
> Lustre-discuss mailing list
> Lustre-discuss at lists.lustre.org
> http://lists.lustre.org/mailman/listinfo/lustre-discuss