[lustre-discuss] tune2fs being blocked by MMP

[Dilger, Andreas]

On Aug 11, 2016, at 03:59, Gibbins, Faye <Faye.Gibbins at cirrus.com> wrote:
> 
> So we have three lustre nodes ATM, say A, B and C. The MSG+OSS and two additional ones which are OSS's too.
> 
> In this case we have an LVM called /dev/mapper/audiodb--1--5-audiodb_0 which is one part of many ext4 file systems that make up the whole audiodb file system under Lustre.
> 
> I've checked all three machines and confirmed that only host A mounts that LVM, however all three can see the LVM as we are multi-pathing the fibre to the backend storage.
> 
> On host A I can run "tune2fs -l /dev/mapper/audiodb--1--5-audiodb_0" just fine.

That's because this command is opening the filesystem read-only and does not trigger the MMP check.

> But when I try to run "tune2fs -e panic /dev/mapper/audiodb--1--5-audiodb_0" this happens:
> 
> Snip---
> 0 A:~# tune2fs -e panic /dev/mapper/audiodb--1--5-audiodb_0
> tune2fs 1.42.13.wc4 (28-Nov-2015)

> tune2fs: MMP: device currently active while trying to open /dev/mapper/audiodb--1--5-audiodb_0
> MMP error info: last update: Thu Aug 11 10:55:48 2016
> node: A device: dm-20
> 0 A:~#
> Snip---

This is what MMP is intended to avoid - multiple tools modifying the filesystem at the same time.  In this case, the tune2fs command is just setting a bit in the superblock, so in theory it should be OK to do on a running filesystem.

However, since the kernel code does not refresh the s_mount_opt flag from disk after it is first read at mount (see test_opt() macro), there is no benefit to changing this field while the filesystem is mounted, so you may as well wait until the next time the filesystem is remounted.

Cheers, Andreas

> Can anyone help with this?
> 
> Yours
> Faye
> 
> PS Hostnames changed.
> 
> -----Original Message-----
> From: COUVEE, PHILIPPE [mailto:philippe.couvee at atos.net] 
> Sent: 04 August 2016 14:56
> To: Gibbins, Faye <Faye.Gibbins at cirrus.com>; Mohr Jr, Richard Frank (Rick Mohr) <rmohr at utk.edu>
> Cc: lustre-discuss at lists.lustre.org
> Subject: RE: [lustre-discuss] tune2fs being blocked by MMP
> 
> Hi,
> If I'm not wrong, MMP stands for Multiple Mount Protection, and is activated to prevent mounting a device that is already mounted on another node in an HA cell.
> In your case, are you trying to issue the tunefs command on the "backup" node for that device ? What if you try running tunefs on the OSS where the device is currently mounted ?
> Philippe
> 
> -----Original Message-----
> From: lustre-discuss [mailto:lustre-discuss-bounces at lists.lustre.org] On Behalf Of Gibbins, Faye
> Sent: Thursday, August 04, 2016 3:40 PM
> To: Mohr Jr, Richard Frank (Rick Mohr)
> Cc: lustre-discuss at lists.lustre.org
> Subject: Re: [lustre-discuss] tune2fs being blocked by MMP
> 
> Hi,
> 
> Yes it is mounted. But that's not always a problem. We have a test lustre cluster where it's mounted and the tune2fs works fine. But it fails in production.
> 
> Production have failover turned on for the OSTs. Something absent on that particular test cluster.
> 
> Is it possible there's something in lustre that prohibits tune2fs working only when failover is on for an OST?
> 
> Faye
> 
> -----Original Message-----
> From: Mohr Jr, Richard Frank (Rick Mohr) [mailto:rmohr at utk.edu]
> Sent: 02 August 2016 17:38
> To: Gibbins, Faye <Faye.Gibbins at cirrus.com>
> Cc: lustre-discuss at lists.lustre.org
> Subject: Re: [lustre-discuss] tune2fs being blocked by MMP
> 
> 
>> On Aug 2, 2016, at 10:38 AM, Gibbins, Faye <Faye.Gibbins at cirrus.com> wrote:
>> 
>> 
>> tune2fs: MMP: device currently active while trying to open 
>> /dev/mapper/scratch--1--5-scratch_3
>> 
>> MMP error info: last update: Tue Aug  2 15:34:09 2016
>> 
>> node: edi-vf-1-5.ad.cirrus.com device: dm-19
>> 
>> 0 edi-vf-1-5:~#
> 
> Is the device currently mounted?  If so, that would explain the error.
> 
> --
> Rick Mohr
> Senior HPC System Administrator
> National Institute for Computational Sciences https://urldefense.proofpoint.com/v2/url?u=http-3A__www.nics.tennessee.edu&d=DQIFAg&c=O3LcjD-V2Iepl5V0N1424A&r=zYKOt5zq8DxD76qBilrgQKxtITGPmVmWHHVOs6JpkMQ&m=HHqIHO6NJdOqJbinMrvMmk8S9oklNJRSgMNfnTOJLVk&s=9Ntf1l5dP_4JpRf-8OjiWKFFzwoJ5BP7VC-uNgSLF3k&e=
> 
> This message and any attachments may contain privileged and confidential information that is intended solely for the person(s) to whom it is addressed. If you are not an intended recipient you must not: read; copy; distribute; discuss; take any action in or make any reliance upon the contents of this message; nor open or read any attachment. If you have received this message in error, please notify us as soon as possible on the following telephone number and destroy this message including any attachments. Thank you. Cirrus Logic International (UK) Ltd and Cirrus Logic International Semiconductor Ltd are companies registered in Scotland, with registered numbers SC089839 and SC495735 respectively. Our registered office is at Westfield House, 26 Westfield Road, Edinburgh, EH11 2QB, UK. Tel: +44 (0)131 272 7000. cirrus.com _______________________________________________
> lustre-discuss mailing list
> lustre-discuss at lists.lustre.org
> https://urldefense.proofpoint.com/v2/url?u=http-3A__lists.lustre.org_listinfo.cgi_lustre-2Ddiscuss-2Dlustre.org&d=DQIFAg&c=O3LcjD-V2Iepl5V0N1424A&r=zYKOt5zq8DxD76qBilrgQKxtITGPmVmWHHVOs6JpkMQ&m=fWmW7_jkiSUZRhKkznsqAA6VMAPgHz6MaGwyaVQK83o&s=0Rgf5dFYQXm0fZA8rKzfRItZ__7GPWwsVDUNaY0-ayo&e= 
> _______________________________________________
> lustre-discuss mailing list
> lustre-discuss at lists.lustre.org
> http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org