CentOS

I found the following in dmesg and I am wondering what I can do to fix it:

EXT4-fs (snumbd1d): group descriptors corrupted

Will I need to take the server offline in order to recover the file system or is it possible to do it without taking the server offline?

Code: Select all

EXT4-fs (snumbd1d): ext4_check_descriptors: Checksum for group 128 failed (22619!=896)
EXT4-fs (snumbd1d): group descriptors corrupted!
snumbd_open_blk(blkid,65537): Disable access (515,210331,210331)...
snumbdctl_release(mms,223881): OK s=ffff88001f1ea000
snumbdctl_open(mms,223881): OK s=ffff88001f1ea000
session_init(mms,223881): OK. kdev=fc:1, len=16351232 sect=512 s=ffff88001f1ea000 pgrp=(210331).
snumbd_open_blk(blkid,65545): Disable access (515,210331,210331)...
snumbd_open_blk(udisks-part-id,65546): Disable access (515,210331,210331)...
snumbdctl_release(mms,223881): OK s=ffff88001f1ea000

You can only run fsck against an unmounted filesystem so you will need to be able to do that. If this is your root filesystem or anything used by essential services that are required to boot up then you will need to reboot from alternate boot media (the install iso image has a 'rescue' option off the troubleshooting menu) and run fsck from there. If this is a filesystem that you can umount while the server is up then you can do that and then run fsck against it and remount it.

Is there a way to determine which partition is having issues from the snippet provided?

Nothing there that I can see helps. There's a mention of "kdev=fc:1" but I can't see anything on any of the systems that I maintain that has any entries in /dev using a major node of 252 (0xfc) though LVM seems to use 253.

CentOS

CentOS 7 : Group descriptors corrupted

CentOS 7 : Group descriptors corrupted

Re: CentOS 7 : Group descriptors corrupted

Re: CentOS 7 : Group descriptors corrupted

Re: CentOS 7 : Group descriptors corrupted