2020-03-06 13:59:53

by Alok Jain

[permalink] [raw]
Subject: Ext4 errors in linux /var/log/messages

Hi Guys,

Thanks for providing the forum to check the issue, I already got
really helpful responses and need one more help wrt to EXT4 log
messages coming in /var/log/messages file

1) I have iSCSI device attached to my VM
2) Copied and extracted files in the attached device
3) Unmount and disconnected the device, post disconnect operation I
can see lots of EXT4 file system errors in /var/log/messages file,
attaching the same (err.txt) for review.

I am not sure what these log messages are correspond to as the device
was already disconnected at Feb 1:22:42, but the ext4 log messages
came after that, can you please review and share your findings.

Any help is appreciated

Thanks,
Alok


Attachments:
err.txt (7.68 kB)

2020-03-06 14:15:52

by Theodore Ts'o

[permalink] [raw]
Subject: Re: Ext4 errors in linux /var/log/messages

On Fri, Mar 06, 2020 at 07:29:22PM +0530, Alok Jain wrote:
> Hi Guys,
>
> Thanks for providing the forum to check the issue, I already got
> really helpful responses and need one more help wrt to EXT4 log
> messages coming in /var/log/messages file
>
> 1) I have iSCSI device attached to my VM
> 2) Copied and extracted files in the attached device
> 3) Unmount and disconnected the device, post disconnect operation I
> can see lots of EXT4 file system errors in /var/log/messages file,
> attaching the same (err.txt) for review.
>
> I am not sure what these log messages are correspond to as the device
> was already disconnected at Feb I1:22:42, but th*e ext4 log messages
> came after that, can you please review and share your findings.

There's nothing really deep here. Ext4 wasn't able to write to
/dev/sdd1. Assuming that /dev/sdd is your iSCSI device, it's probably
because the connection to the iSCSI device had gotten disrupted. This
is similar to what you might see if you were writing to a USB
thumbstick and the USB device is unceremoniously yanked out.

One possible scenario is thar while the umount was *started*, but it
had not completed writing out dirty blocks before the iSCSI device was
disconnected. If that's true, this would be a case of operator error.

- Ted