Hi Mike,

I finally found the solution! I´ve just downgraded the globus toolkit
version from 5.2.4 to 5.2.3. I´ve already made many tests with the
parameter -r to guarantee that everything is ok.

I suggest to Globus team to check the version 5.2.4 if there is any problem
or just some packages that I have installed maybe were with some mistake.

Thanks for all help.


On Tue, Oct 8, 2013 at 5:29 PM, Fabio Moreira <[email protected]>wrote:

> Hi Mike,
>
> I've redone the tests with log_level set to all and I redirected the
> log_single to /var/log/globus-gridftp.log. With these parameters I've got
> another error in each server.
>
> From server1 I got the following error:
> [25634] Tue Oct  8 15:36:41 2013 :: Forcefully terminating process.  SEND
> stalled after 1 updates.
>
> From server3 I got the following error:
> [5028] Tue Oct  8 15:24:11 2013 :: Starting to transfer "/lustre/150G".
> [5028] Tue Oct  8 15:36:41 2013 :: Failure attempting to transfer
> "/lustre/150G".
> [5028] Tue Oct  8 15:36:41 2013 :: Transfer failure:
> callback failed.
> globus_xio: System error in writev: Connection reset by peer
> globus_xio: A system call failed: Connection reset by peer
>
> [5028] Tue Oct  8 15:36:41 2013 :: force_close:
> Handle not in the proper state
>
> I hope that these errors can help somehow.
>
> Thanks in advance,
>
> Fabio MS
>
> On Tue, Sep 24, 2013 at 10:12 PM, Fabio Moreira <[email protected]>wrote:
>
>> Hi Mike,
>>
>> I have run multiple tests. But I forgot to mention a difference between
>> the instances. I have started the instance from port 2811 with xinetd
>> without any difference from the installation. However, I have started the
>> instance to listen from port 6969 as the follow way:
>>
>> SERVER1:
>> /usr/sbin/globus-gridftp-server -p 7777 -dn -aa -S
>> /usr/sbin/globus-gridftp-server -p 6969 -r server1:7777,server3:7777 -aa
>> -S
>>
>> SERVER2:
>> /usr/sbin/globus-gridftp-server -p 7777 -dn -aa -S
>> /usr/sbin/globus-gridftp-server -p 6969 -r server2:7777,server4:7777 -aa
>> -S
>>
>> After that, I have made a test putting the line *-r
>> server1:7777,server3:7777 -aa -S* in */etc/xinetd.d/gridftp* as a *
>> server_args* and the problem also happened with the port 2811.
>>
>> I hope you can help me.
>>
>> Best Regards,
>>
>> Fabio de Souza
>>
>>
>>
>> On Tue, Sep 24, 2013 at 6:01 PM, Michael Link <[email protected]> wrote:
>>
>>> Transfers using 2811 always succeed and transfers to 6969 always fail?
>>> That is very strange if they have identical configurations.  Have you run
>>> multiple tests of each?  It is possible that there is an intermittent error
>>> that just seems to affect one instance.
>>>
>>> Mike
>>>
>>>
>>> On 9/24/2013 3:24 PM, Fabio Moreira wrote:
>>>
>>>> Hi Mike,
>>>>
>>>> There is no difference between the servers and both servers are running
>>>> an instance on 2811 and 6969.
>>>>
>>>> Thanks,
>>>>
>>>> On Tue, Sep 24, 2013 at 5:09 PM, Michael Link <[email protected]
>>>> <mailto:[email protected]>> wrote:
>>>>
>>>>     But each server is running an instance on 2811 and 6969.  Is there a
>>>>     difference in the version or configuration of those besides the port
>>>>     number?
>>>>
>>>>     Mike
>>>>
>>>>
>>>>
>>>>     On 9/24/2013 8:38 AM, Fabio Moreira wrote:
>>>>
>>>>         Hi Mike,
>>>>
>>>>         Server 1 has 6 cores with 96GB running CentOS 6.3 and Server2
>>>> has 8
>>>>         cores with 128GB running CentOS 6.4 and both with Globus 5.2.4.
>>>>         Besides
>>>>         that, there are no difference between them.
>>>>
>>>>
>>>>         On Mon, Sep 23, 2013 at 2:48 PM, Michael Link <
>>>> [email protected]
>>>>         <mailto:[email protected]>
>>>>         <mailto:[email protected] <mailto:[email protected]>>> wrote:
>>>>
>>>>              What is different about the two servers in terms of
>>>> software
>>>>              versions, configuration, etc?
>>>>
>>>>
>>>>              Mike
>>>>
>>>>
>>>>              On 9/20/2013 1:06 PM, Fábio Moreira wrote:
>>>>
>>>>                  Hi,
>>>>
>>>>                  I´m facing a big problem while transferring files
>>>>         bigger than 100GB.
>>>>                  When I transfer the way bellow, a message of
>>>>         end-of-file appears
>>>>                  in the
>>>>                  middle of the transfer. It happens with whatever file
>>>>         bigger
>>>>                  than 100GB
>>>>                  and never in the same point of the file.
>>>>
>>>>                  *globus-url-copy -vb gsiftp://server1:6969/lustre/_**
>>>> ___150G
>>>>
>>>>
>>>>
>>>>                  gsiftp://server2:6969/lustre/
>>>>                  Source: gsiftp://server1:6969/lustre/
>>>>                  Dest:   gsiftp://server2:6969/lustre/
>>>>                      150G
>>>>
>>>>                      76896796672 bytes        98.44 MB/sec avg
>>>>  94.90
>>>>                  MB/sec inst
>>>>                  error: an end-of-file was reached
>>>>                  globus_xio: An end of file occurred*
>>>>
>>>>
>>>>                  However, when I transfer as the follow way, the
>>>> transfer
>>>>                  finishes correctly:
>>>>                  *
>>>>                  globus-url-copy -vb gsiftp://server1/lustre/150G
>>>>                  gsiftp://server2/lustre/
>>>>                  Source: gsiftp://***server*1/lustre/
>>>>                  Dest:   gsiftp://***server*2/lustre/
>>>>                      150G
>>>>
>>>>                     161061273600 bytes        97.30 MB/sec avg
>>>>  96.11
>>>>                  MB/sec inst*
>>>>
>>>>
>>>>                  Has anyone ever faced this kind of trouble?
>>>>
>>>>                  Thanks in advance!
>>>>
>>>>                  Best Regards,
>>>>
>>>>


-- 
Fábio MS

Reply via email to