[Gluster-users] deleted files make bricks full ?

Lakshmipathi Ganapathi lakshmipathi at gluster.com
Wed Aug 17 07:16:12 UTC 2011


Hi -
Are you sure all NFS clients are unmounted ?  For example, if you have two nfs mounts (on machine client1 and client2) .
If you delete a file at client1 and umount client1. You will see the  deleted files  (at /proc) until you perform an umount at client2.


.
..
...
Cheers,
Lakshmipathi.G
FOSS Programmer.


________________________________________
From: gluster-users-bounces at gluster.org [gluster-users-bounces at gluster.org] on behalf of Tomoaki Sato [tsato at valinux.co.jp]
Sent: Monday, August 15, 2011 9:24 PM
To: gluster-users at gluster.org
Subject: Re: [Gluster-users] deleted files make bricks full ?

Hi,

I've reported a trouble with 3.1.5-1 as mentioned below.
I can see the same issue with 3.1.6-1 too.
Why glusterfsd keep open files deleted by NFS clients?

Best,

(2011/08/02 8:43), Tomoaki Sato wrote:
> a simple way to reproduce the issue:
> 1) NFS mount and create 'foo' and umount.
> 2) NFS mount and delete 'foo' and umount.
> 3) replete 1) 2) till ENOSPC.
>
> command logs are following:
> [root at vhead-010 ~]# rpm -qa | grep gluster
> glusterfs-fuse-3.1.5-1
> glusterfs-core-3.1.5-1
> [root at vhead-010 ~]# cat /etc/issue
> CentOS release 5.6 (Final)
> Kernel \r on an \m
>
> [root at vhead-010 ~]# for i in small-{1..4}-4-private; do ssh $i df /mnt/brick; do
> ne
> Filesystem 1K-blocks Used Available Use% Mounted on
> /dev/mapper/00002cde:00002cdf:8
> 103212320 192256 103020064 1% /mnt/brick
> Filesystem 1K-blocks Used Available Use% Mounted on
> /dev/mapper/00002ceb:00002cec:8
> 103212320 192256 103020064 1% /mnt/brick
> Filesystem 1K-blocks Used Available Use% Mounted on
> /dev/mapper/00002cf8:00002cf9:8
> 103212320 192256 103020064 1% /mnt/brick
> Filesystem 1K-blocks Used Available Use% Mounted on
> /dev/mapper/00002d05:00002d06:8
> 103212320 192256 103020064 1% /mnt/brick
> [root at vhead-010 ~]# mount small:/small /mnt
> [root at vhead-010 ~]# ls /mnt
> [root at vhead-010 ~]# dd if=/dev/zero of=/mnt/foo bs=1M count=1024
> 1024+0 records in
> 1024+0 records out
> 1073741824 bytes (1.1 GB) copied, 17.8419 seconds, 60.2 MB/s
> [root at vhead-010 ~]# ls -l /mnt/foo
> -rw-r--r-- 1 root root 1073741824 Aug 2 08:14 /mnt/foo
> [root at vhead-010 ~]# umount /mnt
> [root at vhead-010 ~]# for i in small-{1..4}-4-private; do ssh $i df /mnt/brick; do
> ne
> Filesystem 1K-blocks Used Available Use% Mounted on
> /dev/mapper/00002cde:00002cdf:8
> 103212320 1241864 101970456 2% /mnt/brick
> Filesystem 1K-blocks Used Available Use% Mounted on
> /dev/mapper/00002ceb:00002cec:8
> 103212320 192256 103020064 1% /mnt/brick
> Filesystem 1K-blocks Used Available Use% Mounted on
> /dev/mapper/00002cf8:00002cf9:8
> 103212320 192256 103020064 1% /mnt/brick
> Filesystem 1K-blocks Used Available Use% Mounted on
> /dev/mapper/00002d05:00002d06:8
> 103212320 192256 103020064 1% /mnt/brick
> [root at vhead-010 ~]# mount small:/small /mnt
> [root at vhead-010 ~]# rm -f /mnt/foo
> [root at vhead-010 ~]# ls /mnt
> [root at vhead-010 ~]# umount /mnt
> [root at vhead-010 ~]# for i in small-{1..4}-4-private; do ssh $i df /mnt/brick; do
> ne
> Filesystem 1K-blocks Used Available Use% Mounted on
> /dev/mapper/00002cde:00002cdf:8
> 103212320 1241864 101970456 2% /mnt/brick
> Filesystem 1K-blocks Used Available Use% Mounted on
> /dev/mapper/00002ceb:00002cec:8
> 103212320 192256 103020064 1% /mnt/brick
> Filesystem 1K-blocks Used Available Use% Mounted on
> /dev/mapper/00002cf8:00002cf9:8
> 103212320 192256 103020064 1% /mnt/brick
> Filesystem 1K-blocks Used Available Use% Mounted on
> /dev/mapper/00002d05:00002d06:8
> 103212320 192256 103020064 1% /mnt/brick
> [root at vhead-010 ~]# ssh small-1-4-private
> [root at localhost ~]# du /mnt/brick
> 16 /mnt/brick/lost+found
> 24 /mnt/brick
> [root at localhost ~]# ps ax | grep glusterfsd | grep -v grep
> 7246 ? Ssl 0:03 /opt/glusterfs/3.1.5/sbin/glusterfsd --xlator-option
> small-server.listen-port=24009 -s localhost --volfile-id small.small-1-4-private
> .mnt-brick -p /etc/glusterd/vols/small/run/small-1-4-private-mnt-brick.pid --bri
> ck-name /mnt/brick --brick-port 24009 -l /var/log/glusterfs/bricks/mnt-brick.log
>
> [root at localhost ~]# ls -l /proc/7246/fd
> total 0
> lrwx------ 1 root root 64 Aug 2 08:18 0 -> /dev/null
> lrwx------ 1 root root 64 Aug 2 08:18 1 -> /dev/null
> lrwx------ 1 root root 64 Aug 2 08:18 10 -> socket:[153304]
> lrwx------ 1 root root 64 Aug 2 08:18 11 -> socket:[153306]
> lrwx------ 1 root root 64 Aug 2 08:18 12 -> socket:[153388]
> lrwx------ 1 root root 64 Aug 2 08:18 13 -> /mnt/brick/foo (deleted) <====
> lrwx------ 1 root root 64 Aug 2 08:18 2 -> /dev/null
> lr-x------ 1 root root 64 Aug 2 08:18 3 -> eventpoll:[153252]
> l-wx------ 1 root root 64 Aug 2 08:18 4 -> /var/log/glusterfs/bricks/mnt-brick.
> log
> lrwx------ 1 root root 64 Aug 2 08:18 5 -> /etc/glusterd/vols/small/run/small-1
> -4-private-mnt-brick.pid
> lrwx------ 1 root root 64 Aug 2 08:18 6 -> socket:[153257]
> lrwx------ 1 root root 64 Aug 2 08:18 7 -> socket:[153301]
> lrwx------ 1 root root 64 Aug 2 08:18 8 -> /tmp/tmpfpuXk7N (deleted)
> lrwx------ 1 root root 64 Aug 2 08:18 9 -> socket:[153297]
> [root at localhost ~]# exit
> [root at vhead-010 ~]# mount small:/small /mnt
> [root at vhead-010 ~]# dd if=/dev/zero of=/mnt/foo bs=1M count=1024
> 1024+0 records in
> 1024+0 records out
> 1073741824 bytes (1.1 GB) copied, 21.4717 seconds, 50.0 MB/s
> [root at vhead-010 ~]# ls -l /mnt/foo
> -rw-r--r-- 1 root root 1073741824 Aug 2 08:19 /mnt/foo
> [root at vhead-010 ~]# umount /mnt
> [root at vhead-010 ~]# for i in small-{1..4}-4-private; do ssh $i df /mnt/brick; do
> ne
> Filesystem 1K-blocks Used Available Use% Mounted on
> /dev/mapper/00002cde:00002cdf:8
> 103212320 2291472 100920848 3% /mnt/brick
> Filesystem 1K-blocks Used Available Use% Mounted on
> /dev/mapper/00002ceb:00002cec:8
> 103212320 192256 103020064 1% /mnt/brick
> Filesystem 1K-blocks Used Available Use% Mounted on
> /dev/mapper/00002cf8:00002cf9:8
> 103212320 192256 103020064 1% /mnt/brick
> Filesystem 1K-blocks Used Available Use% Mounted on
> /dev/mapper/00002d05:00002d06:8
> 103212320 192256 103020064 1% /mnt/brick
> [root at vhead-010 ~]# mount small:/small /mnt
> [root at vhead-010 ~]# rm -f /mnt/foo
> [root at vhead-010 ~]# ls /mnt
> [root at vhead-010 ~]# umount /mnt
> [root at vhead-010 ~]# for i in small-{1..4}-4-private; do ssh $i df /mnt/brick; do
> ne
> Filesystem 1K-blocks Used Available Use% Mounted on
> /dev/mapper/00002cde:00002cdf:8
> 103212320 2291472 100920848 3% /mnt/brick
> Filesystem 1K-blocks Used Available Use% Mounted on
> /dev/mapper/00002ceb:00002cec:8
> 103212320 192256 103020064 1% /mnt/brick
> Filesystem 1K-blocks Used Available Use% Mounted on
> /dev/mapper/00002cf8:00002cf9:8
> 103212320 192256 103020064 1% /mnt/brick
> Filesystem 1K-blocks Used Available Use% Mounted on
> /dev/mapper/00002d05:00002d06:8
> 103212320 192256 103020064 1% /mnt/brick
> [root at vhead-010 ~]# ssh small-1-4-private ls -l /proc/7246/fd
> total 0
> lrwx------ 1 root root 64 Aug 2 08:18 0 -> /dev/null
> lrwx------ 1 root root 64 Aug 2 08:18 1 -> /dev/null
> lrwx------ 1 root root 64 Aug 2 08:18 10 -> socket:[153304]
> lrwx------ 1 root root 64 Aug 2 08:18 11 -> socket:[153306]
> lrwx------ 1 root root 64 Aug 2 08:18 12 -> socket:[153388]
> lrwx------ 1 root root 64 Aug 2 08:18 13 -> /mnt/brick/foo (deleted) <====
> lrwx------ 1 root root 64 Aug 2 08:21 14 -> /mnt/brick/foo (deleted) <====
> lrwx------ 1 root root 64 Aug 2 08:18 2 -> /dev/null
> lr-x------ 1 root root 64 Aug 2 08:18 3 -> eventpoll:[153252]
> l-wx------ 1 root root 64 Aug 2 08:18 4 -> /var/log/glusterfs/bricks/mnt-brick.
> log
> lrwx------ 1 root root 64 Aug 2 08:18 5 -> /etc/glusterd/vols/small/run/small-1
> -4-private-mnt-brick.pid
> lrwx------ 1 root root 64 Aug 2 08:18 6 -> socket:[153257]
> lrwx------ 1 root root 64 Aug 2 08:18 7 -> socket:[153301]
> lrwx------ 1 root root 64 Aug 2 08:18 8 -> /tmp/tmpfpuXk7N (deleted)
> lrwx------ 1 root root 64 Aug 2 08:18 9 -> socket:[153297]
>
> Tomo Sato
>
> (2011/08/02 7:14), Tomoaki Sato wrote:
>> Hi,
>>
>> My simple test program, which repeat create-write-read-delete 64 of 1GB
>> files on 100GB x 4 bricks from 4 NFS clients, fails due to ENOSPC.
>> I found some glusterfsd have many file descriptors of same name, deleted
>> files and these files fill the bricks.
>> Is this know issue?
>>
>> Best,
>>
>> Tomo Sato
>>
>>
>>
>>
>> _______________________________________________
>> Gluster-users mailing list
>> Gluster-users at gluster.org
>> http://gluster.org/cgi-bin/mailman/listinfo/gluster-users
>
> _______________________________________________
> Gluster-users mailing list
> Gluster-users at gluster.org
> http://gluster.org/cgi-bin/mailman/listinfo/gluster-users

_______________________________________________
Gluster-users mailing list
Gluster-users at gluster.org
http://gluster.org/cgi-bin/mailman/listinfo/gluster-users



More information about the Gluster-users mailing list