[Clas_offline] similar problem on batch farm

[Nikolai Pivnyuk]

Hi Nick,
all my executables in "fsgen-gsim-gpp-recsis" chain work fine ; the
results look as expected.
You can have a look through run_chain and run_output files in my directory
/volatile/clas/clasg10/pivnyuk/g10/qf_pipp/mc/shell.

Really the SINGULARITY OF YOURS is a great thing. I feel I have started to
respect all singularities
under the moon. Thank you so much for this valuable gift to all clas6 fans.

To send a great number of jobs to CEBAF batch farm I use for a long time a
couple of simple perl scripts - send_to_farm.pl and runmc.pl.
You can see them in /volatile/clas/clasg10/pivnyuk/g10/qf_pipp/mc/perl
directory.
There are still unsolved problem under the perl. Before proceed to run
user_ana I need to switch
CLAS_PARMS to /group/clas/parms . But it looks absolutely like the line(s)
(command(s)) in runmc.pl
------------------------------------------------
system ("unsetenv CLAS_PARMS");
system ("setenv CLAS_PARMS /group/clas/parms");
------------------------------------------------
result in nothing. The user_ana starts with wrong environment and crashes.
You can look through send_output file in .../perl directory. Currently I
have no idea what it might be the reason and how to fight it. May be
somebody met sometime similar situation ?

Once more : under the "shell" everything looks fine, under the "perl"
problem.

Anyway , thanks a lot one more time.

Nikolai




> Hi Nikolai,
>
> You have to load the singularity module both interactively and in batch
> scripts before you can use it.
>
> To do this either run this command or add it to your .cshrc/.bashrc file:
>
> module load singularity/3.4.0
>
> This will load the singularity command into your path so that you can use
> it.
>
> Nick
>
>> On Feb 19, 2020, at 9:07 AM, Nikolai Pivnyuk <pivnyuk at jlab.org> wrote:
>>
>> Hi Nick,
>>
>> I am a bit confused and upset. ifarm180i as well as the batch farm nodes
>> do not understand what does the command "singularity" mean. Could you
>> please
>> check that from your account.
>> ----------------------------------------------------------------
>> pivnyuk at ifarm1801.jlab.org> pwd
>> /u/home/pivnyuk/g10/mc/qf_2020/fsgen_gsim_gpp_recsis
>> pivnyuk at ifarm1801.jlab.org> date
>> Wed Feb 19 08:55:24 EST 2020
>> pivnyuk at ifarm1801.jlab.org> singularity exec
>> /work/clas/clase1/tylern/clas6.img bosdump
>> singularity: Command not found.
>> pivnyuk at ifarm1801.jlab.org>
>> -----------------------------------------------------------------------------------
>> It seems that ifarm180i have been rebooted a few hours ago.
>>
>> Regards,
>>
>> Nikolai
>>
>>
>>> Hi Nikolai,
>>>
>>> All of these programs you mention work with the singularity image as
>>> well,
>>> and in theory all the the clas6 executables should work inside of the
>>> image in this way. I have the code used to build everything and a bit
>>> of a
>>> readme about using it on the farm as a slurm submission available on my
>>> GitHub. (https://github.com/tylern4/clas6
>>> <https://github.com/tylern4/clas6>) You can think of the singularity
>>> image
>>> as a light weight virtual machine that is running centos6 and has all
>>> the
>>> clas6 software built inside of it.
>>>
>>> Or you can modify your job submission simply by appending "singularity
>>> exec /work/clas/clase1/tylern/clas6.img” before the executable you want
>>> to
>>> run with all the same arguments afterwards.
>>>
>>> For example:
>>> gsim_bat … => singularity exec /work/clas/clase1/tylern/clas6.img
>>> gsim_bat
>>> …
>>> gpp … =>  singularity exec /work/clas/clase1/tylern/clas6.img gpp …
>>> Etc.
>>>
>>> If you want I’m happy to help you get started using it on the farm.
>>>
>>> Nick
>>>
>>>
>>>> On Feb 18, 2020, at 10:52 AM, Nikolai Pivnyuk <pivnyuk at jlab.org>
>>>> wrote:
>>>>
>>>> Hi,
>>>>
>>>> it is not only the "bosdump" affected by centos77 !?
>>>> My executables in simulations also failed IN THE SIMILAR WAY as
>>>> bosdump
>>>> did.
>>>>
>>>> Below ARE the lines from my ".err" output file from farm batch (Auger)
>>>> job
>>>> with "OS: centos77" specification.
>>>>
>>>> -----------------------------------------------------------------------------------------------------------------------------------
>>>> gsim_bat: error while loading shared libraries: libXbae.so.4: cannot
>>>> open
>>>> shared object file: No such file or directory
>>>> gpp: error while loading shared libraries: libmysqlclient.so.20:
>>>> cannot
>>>> open shared object file: No such file or directory
>>>> mv: cannot stat ‘gpp.hbook’: No such file or directory
>>>> -----------------------------------------------------------------------------------------------------------------------------------
>>>> user_ana: error while loading shared libraries: libmysqlclient.so.20:
>>>> cannot open shared object file: No such file or directory
>>>> /home/claseg2/bin/LinuxRHEL7/user_ana: error while loading shared
>>>> libraries: libmysqlclient.so.20: cannot open shared object file: No
>>>> such
>>>> file or directory
>>>> -------------------------------------------------------------------------------------------------------------------------------------
>>>>
>>>> Hard times begin for the simple ordinary user ?
>>>> May be it might be reasonable to keep some nodes untouchable in 'old'
>>>> OS
>>>> conditions until 'new' start working perfectly ?
>>>>
>>>> Nikolai
>>>>
>>>> _______________________________________________
>>>> Clas_offline mailing list
>>>> Clas_offline at jlab.org
>>>> https://mailman.jlab.org/mailman/listinfo/clas_offline
>>>
>>>
>>
>>
>
>