how to debug problem with "failing unmounting"

[Valdis Kl=?utf-8?Q?=c4=93?=tnieks]

On Thu, 09 Jul 2020 14:22:00 +0200, Tomek The Messenger said:

> On the soc on which I work there is issue with unmouting some
> partitions/directories during /sbin/reboot, here is some example:
>
> [   OK   ]  Stopped target Local File System.
> Unmouting Temporary Directory (/tmp)...
> Unmouting /run/user/0...
> ...
> [   FAILED  ]  Failed unmouting Temporary Directory (/tmp).
> [   FAILED  ]  Failed unmouting  /run/user/0.
> ....
>
> Unfortunately like it is typical in kernel information about why
> unmouting failed doesn't appear. 

Those aren't kernel messages, they're messages issued by your init system.
(Based on OK/FAILED, you're probably using a sysvinit rather than systemd, and
the messages are coming from scripts in /etc/rc.d/rc6/ which is the 'reboot'
runlevel - rc6 has symlinks to the actual script.  The main shutdown/reboot
script basically does a 'for $script in /etc/rc.d/rc6; do $script stop; done'.

The most common reason for a filesystem to fail to unmount is that a process still
has a file open on the file system.

And for this, lsof is your friend.   Stick an '/bin/lsof /tmp' in the script at the appropriate
place to find out what's still got /tmp busy.

(Depending on the system, it may not be worth looking into - often /tmp
and /run are tmpfs systems that are going to evaporate *anyhow*, so there's
no actual danger of a filesystem or data getting corrupted due to a hard
dismount while the file system is active)
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 832 bytes
Desc: not available
URL: <http://lists.kernelnewbies.org/pipermail/kernelnewbies/attachments/20200709/1f8294ef/attachment.sig>