Immediate Logout

[Kevin_H_3]

I'm getting an almost immediate logout of an interactive job. (within less than a minute) It makes using the DevCloud impossible.

uXXXXX@login-2:~$ qsub -lnodes=1:gpu -lwalltime=02:00:00 -d. -I

qsub: waiting for job 490970.v-qsvr-1.aidevcloud to start

qsub: job 490970.v-qsvr-1.aidevcloud ready

########################################################################

#      Date:           Thu Feb 13 11:32:39 PST 2020

#    Job ID:           490970.v-qsvr-1.aidevcloud

#      User:           uXXXXX

# Resources:           neednodes=1:gpu,nodes=1:gpu,walltime=02:00:00

########################################################################

uXXXXX@s001-n141:~$ qstat

Job ID                    Name             User            Time Use S Queue

------------------------- ---------------- --------------- -------- - -----

490970.v-qsvr-1            STDIN            uXXXXX                 0 R batch          

uXXXXX@s001-n141:~$ qstat -l -f

Job Id: 490970.v-qsvr-1.aidevcloud

    Job_Name = STDIN

    Job_Owner = uXXXXX@login-2.aidevcloud

    job_state = R

    queue = batch

    server = v-qsvr-1.aidevcloud

    Checkpoint = u

    ctime = Thu Feb 13 11:32:37 2020

    Error_Path = /dev/pts/0

    exec_host = s001-n141/0

    Hold_Types = n

    interactive = True

    Join_Path = n

    Keep_Files = n

    Mail_Points = n

    mtime = Thu Feb 13 11:32:39 2020

    Output_Path = /dev/pts/0

    Priority = 0

    qtime = Thu Feb 13 11:32:37 2020

    Rerunable = False

    Resource_List.nodect = 1

    Resource_List.nodes = 1:gpu

    Resource_List.walltime = 02:00:00

    session_id = 9193

    Variable_List = PBS_O_QUEUE=batch,PBS_O_HOME=/home/uXXXXX,

PBS_O_LOGNAME=uXXXXX,

PBS_O_PATH=/glob/development-tools/versions/intel-parallel-studio/com

pilers_and_libraries_2019.3.199/linux/bin/intel64:/glob/development-to

ols/versions/intel-parallel-studio/compilers_and_libraries_2019.3.199/

linux/mpi/intel64/libfabric/bin:/glob/development-tools/versions/intel

-parallel-studio/compilers_and_libraries_2019.3.199/linux/mpi/intel64/

bin:/glob/development-tools/versions/intel-parallel-studio/debugger_20

19/gdb/intel64/bin:/glob/intel-python/python3/bin/:/glob/intel-python/

python2/bin/:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/

bin:/usr/games:/usr/local/games:/home/uXXXXX/.local/bin:/home/uXXXXX/b

in:/bin,PBS_O_MAIL=/var/mail/uXXXXX,PBS_O_SHELL=/bin/bash,

PBS_O_LANG=en_US,

PBS_O_SUBMIT_FILTER=/usr/local/sbin/torque_submitfilter,

PBS_O_INITDIR=/home/uXXXXX/.,PBS_O_WORKDIR=/home/uXXXXX,

PBS_O_HOST=login-2.aidevcloud,PBS_O_SERVER=v-qsvr-1.aidevcloud

    euser = uXXXXX

    egroup = uXXXXX

    queue_type = E

    etime = Thu Feb 13 11:32:37 2020

    submit_args = -lnodes=1:gpu -lwalltime=02:00:00 -d. -I

    start_time = Thu Feb 13 11:32:39 2020

    Walltime.Remaining = 7187

    start_count = 1

    fault_tolerant = False

    job_radix = 0

    submit_host = login-2.aidevcloud

[AthiraM_Intel]

Hi,

Thanks for reaching out to us.

Could you please try to login into gpu node using the below command and see if the issue still persists:

qsub -I -l nodes=1:gpu:ppn=2 -d.

You can also request a specific node using the below command

qsub -I -l nodes=s001-n208:gpu:ppn=2 -d.

The list of the nodes with GPU can be obtained by the next command:

pbsnodes | grep -B4 "gen9"

Hope this helps. If you face any further issue, please let us know.

[Kevin_H_3]

Thanks, specifying ppn=2 prevents the logout. Seems odd that ppn is required on an interactive login session. I'm not sure how to interpret that parameter under this condition.

kevin

[AthiraM_Intel]

Hi,

Glad to know that your issue got resolved.

qsub -I -l nodes=1:gpu:ppn=2 -d.  

In the above command, ppn=2 would request 2 processors on one node.

Please refer the below link for more details on job submission.

 http://docs.adaptivecomputing.com/torque/3-0-5/2.1jobsubmission.php 

[AthiraM_Intel]

Hi,

Could you please confirm whether the solution provided was helpful.

[Kevin_H_3]

Yes, setting ppn=2 prevents the logout and i can use the compute node normally.

kevin

[AthiraM_Intel]

Hi,

Thanks for your confirmation.

We are closing this thread. Please feel free to open a new thread if you have any further queries.