[LU-2732] LDISKFS-fs (sdb1): couldn't mount RDWR because of unsupported optional features (100) Created: 31/Jan/13  Updated: 11/Jan/16  Resolved: 11/Jan/16

Status: Resolved
Project: Lustre
Component/s: None
Affects Version/s: Lustre 2.4.0
Fix Version/s: None

Type: Bug Priority: Major
Reporter: Sarah Liu Assignee: Hongchao Zhang
Resolution: Done Votes: 0
Labels: None
Environment:

before upgrade client and server: 2.1.4 RHEL6
after upgrade client and server: lustre-master build #1194 RHEL6


Severity: 3
Rank (Obsolete): 6636

 Description   

After upgrade the whole system from 2.1.4 to 2.4, then downgrade to 2.1.4 again, cannot mount MDS, got following error messages:

[root@fat-amd-1 ~]# mount -t lustre -o acl,user_xattr /dev/sdb1 /mnt/mds1
LDISKFS-fs warning (device sdb1): ldiskfs_fill_super: extents feature not enabled on this filesystem, use tune2fs.
LDISKFS-fs (sdb1): couldn't mount RDWR because of unsupported optional features (100)
LustreError: 8077:0:(obd_mount.c:1360:server_kernel_mount()) premount /dev/sdb1:0x0 ldiskfs failed: -22 Is the ldiskfs module available?
LustreError: 8077:0:(obd_mount.c:1700:server_fill_super()) Unable to mount device /dev/sdb1: -22
LustreError: 8077:0:(obd_mount.c:2198:lustre_fill_super()) Unable to mount  (-22)
mount.lustre: mount /dev/sdb1 at /mnt/mds1 failed: Invalid argument
This may have multiple causes.
Are the mount options correct?
Check the syslog for more info.


 Comments   
Comment by Niu Yawei (Inactive) [ 05/Feb/13 ]

The 0x100 is quota feature, you'd disable the ext4 quota feature by 'tune2fs -O ^quota' before downgrade.

Comment by Hongchao Zhang [ 07/Feb/13 ]

Hi,
During update&downgrade, are you using same linux distro? this error is about quota feature
in "ldiskfs/ldiskfs/ldiskfs.h"
...
#define LDISKFS_FEATURE_RO_COMPAT_QUOTA 0x0100
...

this feature isn't defined in some Linux Distro. e.g. it is defined in RHEL6 but is not in RHEL5.

Comment by Andreas Dilger [ 08/Feb/13 ]

Sarah, can you please put the test configuration (client/server distro versions and exact 2.4 version) into the "Environment" section of the ticket.

There was a patch landed on master (http://review.whamcloud.com/4968) about quieting quota messages, and I think that RO_COMPAT_QUOTA is no longer enabled by default for an upgrade. Niu, is that correct? If not, then the fix is to not enable RO_COMPAT_QUOTA on 2.4 by default, and if the filesystem already has quota enabled then it should refuse to mount in that case.

Comment by Niu Yawei (Inactive) [ 15/Feb/13 ]

Andreas, the quota is still enabled by default for 2.4 (see enable_default_ext4_features()), why we shouldn't enable it by default? for the performance problem or downgrading? For downgradning, I think user can just simply disable quota before downgranding lustre.

Comment by Johann Lombardi (Inactive) [ 16/Feb/13 ]

To be clear, upgraded filesystems don't have RO_COMPAT_QUOTA by default. The admin has to explicitly run tunefs.lustre --quota to enable it. Once this is done, the admin will indeed have to clear the flag with tune2fs if he wants to downgrade.
As for new filesystems formatted with 2.4, they have the quota feature enabled by default.

Comment by Andreas Dilger [ 16/Feb/13 ]

Sorry for confusion. I thought that quota was enabled by default when upgrading to 2.4, but if this has to be done manually by tunefs that is fine.

Comment by Andreas Dilger [ 23/Feb/13 ]

Is there any change needed for the manual in this case? If yes, please move this over to an LUDOC bug, otherwise it can just be closed.

Generated at Sat Feb 10 01:27:43 UTC 2024 using Jira 9.4.14#940014-sha1:734e6822bbf0d45eff9af51f82432957f73aa32c.