[Linux-cluster] gfs/nfs trouble

Jürgen Knödlseder jurgen.knodlseder at cesr.fr
Thu Jan 22 13:54:09 UTC 2009


I confirm that the problem persists with cluster 2.03.11 release ...

By the way: replacing the gfs filesystem by gfs2 makes the problem go  
away.
Yet as one reads so often: gfs2 is not production ready, so I would  
prefer to
get this working with gfs :-)

Jürgen

Le 21 janv. 09 à 23:10, Jürgen Knödlseder a écrit :

> I did some more testing concerning the gfs export problem over nfs.  
> Here the
> strace results. First for a small file where the copy was done  
> successfully, then
> for a larger file where the problems start. Note that the strace  
> results do
> vary for an issue of identical commands, indicating some  
> arbitrariness in the
> behaviour.
>
> Success:
> open("test.xml", O_RDONLY|O_LARGEFILE)  = 3
> fstat64(3, {st_mode=S_IFREG|0644, st_size=1858, ...}) = 0
> open("test2.xml", O_WRONLY|O_TRUNC|O_LARGEFILE) = 4
> fstat64(4, {st_mode=S_IFREG|0644, st_size=0, ...}) = 0
> fstat64(3, {st_mode=S_IFREG|0644, st_size=1858, ...}) = 0
> read(3, "<?xml version=\"1.0\" ?><source_li"..., 4096) = 1858
> write(4, "<?xml version=\"1.0\" ?><source_li"..., 1858) = 1858
> read(3, "", 4096)                       = 0
> close(4)                                = 0
> close(3)                                = 0
> exit_group(0)                           = ?
>
> Failure:
> open("test.png", O_RDONLY|O_LARGEFILE)  = 3
> fstat64(3, {st_mode=S_IFREG|0644, st_size=17318, ...}) = 0
> open("test2.png", O_WRONLY|O_TRUNC|O_LARGEFILE) = 4
> fstat64(4, {st_mode=S_IFREG|0644, st_size=0, ...}) = 0
> fstat64(3, {st_mode=S_IFREG|0644, st_size=17318, ...}) = 0
> read(3, "\211PNG\r\n\32\n\0\0\0\rIHDR\0\0\3 \0\0\2X\10\6\0\0\0 
> \232"..., 4096) = 4096
> write(4, "\211PNG\r\n\32\n\0\0\0\rIHDR\0\0\3 \0\0\2X\10\6\0\0\0 
> \232"..., 4096) = 4096
> read(3, "\222$)\243\364\367\303\366\355a\305\243\266\26*+ 
> \335bW"..., 4096) = 4096
> write(4, "\222$)\243\364\367\303\366\355a\305\243\266\26*+ 
> \335bW"..., 4096) = 4096
> read(3, "\244\24\364\302\v/\360\375\357\177\377\312\327N\262 
> \334"..., 4096) = 4096
> write(4, "\244\24\364\302\v/\360\375\357\177\377\312\327N\262 
> \334"..., 4096) = 4096
> read(3, "D\222$I\267\316\21,)I\215\217CM\r\24\26\302\322\245p 
> \356"..., 4096) = 4096
> write(4, "D\222$I\267\316\21,)I\215\217CM\r\24\26\302\322\245p 
> \356"..., 4096) = 4096
> read(3, "\"\361\361\361\272\343\216;t\340\300\201p\227\202\10 
> \365"..., 4096) = 934
> write(4, "\"\361\361\361\272\343\216;t\340\300\201p\227\202\10 
> \365"..., 934) = -1 EINVAL (Invalid argument)
>
> Another failure:
> open("test.png", O_RDONLY|O_LARGEFILE)  = 3
> fstat64(3, {st_mode=S_IFREG|0644, st_size=17318, ...}) = 0
> open("test2.png", O_WRONLY|O_CREAT|O_LARGEFILE, 0100644) = 4
> fstat64(4, {st_mode=S_IFREG|0644, st_size=0, ...}) = 0
> fstat64(3, {st_mode=S_IFREG|0644, st_size=17318, ...}) = 0
> read(3, "\211PNG\r\n\32\n\0\0\0\rIHDR\0\0\3 \0\0\2X\10\6\0\0\0 
> \232"..., 4096) = 4096
> write(4, "\211PNG\r\n\32\n\0\0\0\rIHDR\0\0\3 \0\0\2X\10\6\0\0\0 
> \232"..., 4096) = 4096
> read(3, "\222$)\243\364\367\303\366\355a\305\243\266\26*+ 
> \335bW"..., 4096) = 4096
> write(4, "\222$)\243\364\367\303\366\355a\305\243\266\26*+ 
> \335bW"..., 4096) = 4096
> read(3, "\244\24\364\302\v/\360\375\357\177\377\312\327N\262 
> \334"..., 4096) = 4096
> write(4, "\244\24\364\302\v/\360\375\357\177\377\312\327N\262 
> \334"..., 4096) = 4096
> read(3, "D\222$I\267\316\21,)I\215\217CM\r\24\26\302\322\245p 
> \356"..., 4096) = 4096
> write(4, "D\222$I\267\316\21,)I\215\217CM\r\24\26\302\322\245p 
> \356"..., 4096) = 4096
> read(3, "\"\361\361\361\272\343\216;t\340\300\201p\227\202\10 
> \365"..., 4096) = 934
> write(4, "\"\361\361\361\272\343\216;t\340\300\201p\227\202\10 
> \365"..., 934) = 934
> read(3, "", 4096)                       = 0
> close(4)                                = -1 EINVAL (Invalid argument)
>
> And yet another one:
> open("/home2/knodlseder/test.png", O_RDONLY|O_LARGEFILE) = 3
> fstat64(3, {st_mode=S_IFREG|0644, st_size=17318, ...}) = 0
> open("/home2/knodlseder/test2.png", O_WRONLY|O_TRUNC|O_LARGEFILE) = 4
> fstat64(4, {st_mode=S_IFREG|0644, st_size=0, ...}) = 0
> fstat64(3, {st_mode=S_IFREG|0644, st_size=17318, ...}) = 0
> read(3, "\211PNG\r\n\32\n\0\0\0\rIHDR\0\0\3 \0\0\2X\10\6\0\0\0 
> \232"..., 4096) = 4096
> write(4, "\211PNG\r\n\32\n\0\0\0\rIHDR\0\0\3 \0\0\2X\10\6\0\0\0 
> \232"..., 4096) = 4096
> read(3, "\222$)\243\364\367\303\366\355a\305\243\266\26*+ 
> \335bW"..., 4096) = 4096
> write(4, "\222$)\243\364\367\303\366\355a\305\243\266\26*+ 
> \335bW"..., 4096) = 4096
> read(3, "\244\24\364\302\v/\360\375\357\177\377\312\327N\262 
> \334"..., 4096) = 4096
> write(4, "\244\24\364\302\v/\360\375\357\177\377\312\327N\262 
> \334"..., 4096) = 4096
> read(3, "D\222$I\267\316\21,)I\215\217CM\r\24\26\302\322\245p 
> \356"..., 4096) = 4096
> write(4, "D\222$I\267\316\21,)I\215\217CM\r\24\26\302\322\245p 
> \356"..., 4096) = -1 EINVAL (Invalid argument)
>
> Jürgen
>
> Le 19 janv. 09 à 09:46, Jürgen Knödlseder a écrit :
>
>> Did it. Here the results:
>>
>> lstat64("/home2/knodlseder/test2.png", {st_mode=S_IFREG|0644,  
>> st_size=0, ...}) = 0
>> stat64("/home2/knodlseder/test2.png", {st_mode=S_IFREG|0644,  
>> st_size=0, ...}) = 0
>> stat64("/home2/knodlseder/test.png", {st_mode=S_IFREG|0644,  
>> st_size=17318, ...}) = 0
>> stat64("/home2/knodlseder/test2.png", {st_mode=S_IFREG|0644,  
>> st_size=0, ...}) = 0
>> open("/home2/knodlseder/test.png", O_RDONLY|O_LARGEFILE) = 3
>> fstat64(3, {st_mode=S_IFREG|0644, st_size=17318, ...}) = 0
>> open("/home2/knodlseder/test2.png", O_WRONLY|O_TRUNC|O_LARGEFILE) = 4
>> fstat64(4, {st_mode=S_IFREG|0644, st_size=0, ...}) = 0
>> fstat64(3, {st_mode=S_IFREG|0644, st_size=17318, ...}) = 0
>> read(3, "\211PNG\r\n\32\n\0\0\0\rIHDR\0\0\3 \0\0\2X\10\6\0\0\0 
>> \232"..., 32768) = 17318
>> write(4, "\211PNG\r\n\32\n\0\0\0\rIHDR\0\0\3 \0\0\2X\10\6\0\0\0 
>> \232"..., 17318) = 17318
>> read(3, "", 32768)                      = 0
>> close(4)                                = -1 EINVAL (Invalid  
>> argument)
>>
>> Any clues ?
>>
>> Jürgen
>>
>> Le 19 janv. 09 à 03:29, jumanjiman at gmail.com a écrit :
>>
>>> Try running 'strace -o /tmp/errs cp /path/to(test2.png  /path/to/ 
>>> destination/' and then inpect the temp file.
>>>
>>> -paul
>>>
>>> Sent via BlackBerry by AT&T
>>>
>>> -----Original Message-----
>>> From: "Jürgen Knödlseder" <jurgen.knodlseder at cesr.fr>
>>>
>>> Date: Sun, 18 Jan 2009 22:29:48
>>> To: linux clustering<linux-cluster at redhat.com>
>>> Subject: [Linux-cluster] gfs/nfs trouble
>>>
>>>
>>> Hi all,
>>>
>>> I have trouble exporting a GFS over NFS. Here the situation:
>>>
>>> I have 2 PE1950 with a MD3000 SAN attached to build a failover
>>> cluster. Both machines run
>>> kernel 2.6.27.7 and cluster-2.03.10. I configured several GFS on the
>>> storage device that I would
>>> like to export via NFS to a rather old PE2800 which runs kernel
>>> 2.4.26 (I have to stick to this
>>> kernel on this machine since it runs openmosix).
>>>
>>> When I copy on the PE2800 a small file (1858 Bytes) everything works
>>> fine. When I try to copy
>>> a slightly larger file (e.g. 17318 Bytes) I get the error
>>>
>>> cp: closing `test2.png': Invalid argument
>>>
>>> and I have an empty file 'test2.png' on the disk ...
>>>
>>> I've seen a similar bug on bugzilla (https://bugzilla.redhat.com/
>>> show_bug.cgi?id=432544)
>>> yet this should be fixed in my configuration ...
>>>
>>> Can somebody help?
>>>
>>> Jürgen
>>>
>>> --
>>> Linux-cluster mailing list
>>> Linux-cluster at redhat.com
>>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>>
>>> --
>>> Linux-cluster mailing list
>>> Linux-cluster at redhat.com
>>> https://www.redhat.com/mailman/listinfo/linux-cluster
>>
>>
>> --
>> Linux-cluster mailing list
>> Linux-cluster at redhat.com
>> https://www.redhat.com/mailman/listinfo/linux-cluster
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090122/96e44e2a/attachment.htm>


More information about the Linux-cluster mailing list