[Gluster-devel] glusterfs crash with mainline-2.5 patch 260 with dbench

[Dale Dude]

Ran again with 261 (which only seems to fix perms?) with same results. 
Using ubuntu dapper (2.6.15-28-amd64-server) with fuse 2.6.5.

I only have one dbench test file so I have no choice.
-rw-r--r-- 1 root root 26214401 2005-11-19 21:19 
/usr/share/dbench/client.txt

root at teststorage:/volumes/t# dbench -t 10 10
dbench version 3.04 - Copyright Andrew Tridgell 1999-2004

Running for 10 seconds with load '/usr/share/dbench/client.txt' and 
minimum warmup 2 secs
10 clients started
  10       127    56.55 MB/sec  warmup   1 sec  
  10       494    57.12 MB/sec  execute   1 sec  
  10       608    47.18 MB/sec  execute   2 sec  
read failed on handle 10003
(612) open ./clients/client1/~dmtmp/PWRPNT/PCBENCHM.PPT failed for 
handle 10004 (Transport endpoint is not connected)
(613) ERROR: handle 10004 was not found
(612) open ./clients/client6/~dmtmp/PWRPNT/PCBENCHM.PPT failed for 
handle 10004 (Transport endpoint is not connected)
read failed on handle 10003
(613) ERROR: handle 10004 was not found
(612) open ./clients/client2/~dmtmp/PWRPNT/PCBENCHM.PPT failed for 
handle 10004 (Transport endpoint is not connected)
(613) ERROR: handle 10004 was not found
(616) open ./clients/client4/~dmtmp/PWRPNT/PCBENCHM.PPT failed for 
handle 10005 (Transport endpoint is not connected)
(617) ERROR: handle 10005 was not found
read failed on handle 10003
(612) open ./clients/client3/~dmtmp/PWRPNT/PCBENCHM.PPT failed for 
handle 10004 (Transport endpoint is not connected)
read failed on handle 10003
(613) ERROR: handle 10004 was not found
read failed on handle 10003
(612) open ./clients/client5/~dmtmp/PWRPNT/PCBENCHM.PPT failed for 
handle 10004 (Transport endpoint is not connected)
(613) ERROR: handle 10004 was not found
(612) open ./clients/client9/~dmtmp/PWRPNT/PCBENCHM.PPT failed for 
handle 10004 (Transport endpoint is not connected)
(613) ERROR: handle 10004 was not found
Child failed with status 1
(616) open ./clients/client0/~dmtmp/PWRPNT/PCBENCHM.PPT failed for 
handle 10005 (Transport endpoint is not connected)
(617) ERROR: handle 10005 was not found
read failed on handle 10004
(616) open ./clients/client7/~dmtmp/PWRPNT/PCBENCHM.PPT failed for 
handle 10005 (Transport endpoint is not connected)
read failed on handle 10004
root at newstorage1:/volumes/t# /bin/rm: /bin/rm: cannot lstat 
`./clients/client7'cannot lstat `./clients/client8': Transport endpoint 
is not connected: Transport endpoint is not connected




Dale Dude wrote:
> Harris,
>
> I only ran 'dbench -t 10 10' without the -c. When not using the -c it 
> uses all the test cases. Checking out 261 on this box and trying 
> again. Ran 261 on another box with 3 parallel rsync's but the mount 
> froze a few times and made rsync fail but mount became responsive 
> after the timeout.
>
> Dale
>
> Harris Landgarten wrote:
>> Dale,
>>
>> I have never used dbench before but I did the following:
>>
>> /mnt/glusterfs $ sudo dbench -t 10 10 -c 
>> /usr/share/dbench/client_oplocks.txt 10 clients started
>>    0     62477  6.35 MB/secc
>> Throughput 6.35386 MB/sec 10 procs
>>
>> This was with patch 261. How does this compare to want you did? I was 
>> trying to reproduce your crash.
>>
>> Harris
>>
>> ----- Original Message -----
>> From: "Dale Dude" <dale at oc3networks.com>
>> To: "gluster-devel" <gluster-devel at nongnu.org>
>> Sent: Monday, July 2, 2007 4:13:50 PM (GMT-0500) America/New_York
>> Subject: [Gluster-devel] glusterfs crash with mainline-2.5 patch 260 
>> with dbench
>>
>> Core dump when using 'dbench -t 10 10':
>>
>