linux-kernel - [Problem]NFS Server – Umount results in Device Busy.

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Date:	Thu, 10 May 2012 06:59:31 +0000 (GMT)
From:	AMIT SAHRAWAT <a.sahrawat@...sung.com>
To:	Trond.Myklebust@...app.com
Cc:	linux-nfs@...r.kernel.org, linux-kernel@...r.kernel.org,
	bfields@...ldses.org,
	"namjae.jeon@...sung.com" <namjae.jeon@...sung.com>
Subject: [Problem]NFS Server – Umount results in Device Busy.

Linux Kernel: 2.6.35.14

We are using NFS for data transfer from one target to another target.
On Server we are using 3 mount points – to be used by application. And they are mentioned in /etc/exports

When initially just NFS Server is started -We can mount and safely un-mount the NFS mount point.
Now, when NFS client is started and we mount at the client. Without doing any IO – if we do un-mount at the NFS Client –it un-mounts safely.
But – when tried to un-mount the device at the NFS server – it results in error “Device Busy”
“umount: can't umount /mnt: Device or resource busy”

Now , in order to umount the device -  we have to kill the process ‘nfsd’ – and then it un-mounts easily.

As per our understanding – it should not be an issue with the un-mount when it is not being used and also there should be no dependency to kill ‘nfsd’ to un-mount the device.

It leads to problem of killing ‘nfsd’ and restarting it – to make other mount points work without issue.

Is there any method to un-mount the device without getting this ‘BUSY’ error? Or can we kill only nfsd service which was using that mount point?
Please share your input on the solution to the above problem. Is this expected behavior?

Thanks & Regards,
Amit Sahrawat