Re: [MTT users] mtt fails, error: identical key already exists

2014-01-03 Thread Ralph Castain

On Jan 3, 2014, at 8:15 AM, Muhammad Wahaj Sethi <se...@hlrs.de> wrote:

> Thanks for the quick reply.
> 
> Right now I am making use of option no 1. 
> 
> As per my understanding, mtt runs tests only if new versions/updates are 
> available. Info about last run is stored in the scratch dir.

No, it is stored in the mtt-versions directory for just that reason. You should 
have an entry like the following in your .ini file:

#==
# MPI get phase
#==

[MPI get: ompi-nightly-trunk]
mpi_details = Open MPI

module = OMPI_Snapshot
ompi_snapshot_url = http://www.open-mpi.org/nightly/trunk
ompi_snapshot_version_file = 
/home/common/mtt-versions/trunk("MTT_VERSION_FILE_SUFFIX").txt

#--

# Get the 1.7 nightly

[MPI get: ompi-nightly-v1.7]
mpi_details = Open MPI

module = OMPI_Snapshot
ompi_snapshot_url = http://www.open-mpi.org/nightly/v1.7
ompi_snapshot_version_file = 
/home/common/mtt-versions/v1.7("MTT_VERSION_FILE_SUFFIX").txt

#--

Note the "snapshot_version_file" entry where the version gets stored


> 
> By removing/changing workspace dir this info will be lost. And next mtt 
> execution may submit old results again.
> 
> - Original Message -
> From: "Ralph Castain" <r...@open-mpi.org>
> To: "MTT Users" <mtt-us...@open-mpi.org>
> Sent: Friday, January 3, 2014 4:11:18 PM
> Subject: Re: [MTT users] mtt fails, error: identical key already exists
> 
> You have two options:
> 
> 1. After each mtt execution, run a "cleanup" script that simply does "rm -rf 
> "
> 
> 2. give the mtt scratch location a different name for each execution. 
> However, be careful as you will fill your disk this way - so perhaps run a 
> cleanup script every N times that whacks the oldest location
> 
> Ralph
> 
> On Jan 3, 2014, at 4:42 AM, Muhammad Wahaj Sethi <se...@hlrs.de> wrote:
> 
>> Hello!
>>   I am running mtt daily using a cron job. After every successful execution, 
>> next mtt execution fails with the following error message.
>> 
>> *** ERROR: An identical key already exists in memory when MTT tried to read 
>> the file 
>> /lustre/ws1/ws/hpcmtt-mtt-0/mtt-scratch/sources/mpi_sources-ompi-nightly-trunk.1.9a1r30043.dump
>>  (key=1.9a1r30042).  This should not happen.  It likely indicates that 
>> multiple MTT clients are incorrectly operating in the same scratch tree. at 
>> /zhome/academic/HLRS/hlrs/hpcmtt/mtt-trunk/lib/MTT/Messages.pm line 131.
>> 
>> I have checked mtt successful run log file and was not able to notice 
>> something unusual. 
>> 
>> Kindly, let me know if addition info is required.
>> 
>> Any help will be greatly appreciated.
>> 
>> regards,
>> Wahaj
>> ___
>> mtt-users mailing list
>> mtt-us...@open-mpi.org
>> http://www.open-mpi.org/mailman/listinfo.cgi/mtt-users
> 
> ___
> mtt-users mailing list
> mtt-us...@open-mpi.org
> http://www.open-mpi.org/mailman/listinfo.cgi/mtt-users
> ___
> mtt-users mailing list
> mtt-us...@open-mpi.org
> http://www.open-mpi.org/mailman/listinfo.cgi/mtt-users



Re: [MTT users] mtt fails, error: identical key already exists

2014-01-03 Thread Muhammad Wahaj Sethi
Thanks for the quick reply.

Right now I am making use of option no 1. 

As per my understanding, mtt runs tests only if new versions/updates are 
available. Info about last run is stored in the scratch dir.

By removing/changing workspace dir this info will be lost. And next mtt 
execution may submit old results again.

- Original Message -
From: "Ralph Castain" <r...@open-mpi.org>
To: "MTT Users" <mtt-us...@open-mpi.org>
Sent: Friday, January 3, 2014 4:11:18 PM
Subject: Re: [MTT users] mtt fails, error: identical key already exists

You have two options:

1. After each mtt execution, run a "cleanup" script that simply does "rm -rf 
"

2. give the mtt scratch location a different name for each execution. However, 
be careful as you will fill your disk this way - so perhaps run a cleanup 
script every N times that whacks the oldest location

Ralph

On Jan 3, 2014, at 4:42 AM, Muhammad Wahaj Sethi <se...@hlrs.de> wrote:

> Hello!
>I am running mtt daily using a cron job. After every successful execution, 
> next mtt execution fails with the following error message.
> 
> *** ERROR: An identical key already exists in memory when MTT tried to read 
> the file 
> /lustre/ws1/ws/hpcmtt-mtt-0/mtt-scratch/sources/mpi_sources-ompi-nightly-trunk.1.9a1r30043.dump
>  (key=1.9a1r30042).  This should not happen.  It likely indicates that 
> multiple MTT clients are incorrectly operating in the same scratch tree. at 
> /zhome/academic/HLRS/hlrs/hpcmtt/mtt-trunk/lib/MTT/Messages.pm line 131.
> 
> I have checked mtt successful run log file and was not able to notice 
> something unusual. 
> 
> Kindly, let me know if addition info is required.
> 
> Any help will be greatly appreciated.
> 
> regards,
> Wahaj
> ___
> mtt-users mailing list
> mtt-us...@open-mpi.org
> http://www.open-mpi.org/mailman/listinfo.cgi/mtt-users

___
mtt-users mailing list
mtt-us...@open-mpi.org
http://www.open-mpi.org/mailman/listinfo.cgi/mtt-users