I just compiled and upgraded OpenPBS from 22.05.11 to 23.06.06 on Rocky 9. Everything works except hooks, and in an odd way.
The first hook to run gets a the full complement of pbs.event() as input to the hook, and runs successfully:
pbs.event().job.Submit_arguments=<jsdl-hpcpa:Argument>-N</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>fluent</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>-P</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>cfd</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>-A</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>5d8c20e5ad650497d7d318bd3ed23dc2975b70bced2e388f8c76da2fbdc61a70</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>-q</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>cfd1</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>-l</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>select=6:ncpus=28:mpiprocs=28</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>-l</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>place=excl</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>combustor_12m.jou.sh</jsdl-hpcpa:Argument>
pbs.event().job.Job_Name=fluent
pbs.event().job.project=cfd
pbs.event().job.Account_Name=5d8c20e5ad650497d7d318bd3ed23dc2975b70bced2e388f8c76da2fbdc61a70
pbs.event().job.Resource_List[select]=6:ncpus=28:mpiprocs=28
pbs.event().job.Resource_List[place]=excl
pbs.event().job.Checkpoint=u
pbs.event().job.Hold_Types=n
pbs.event().job.Join_Path=n
pbs.event().job.Keep_Files=n
pbs.event().job.Mail_Points=a
pbs.event().job.Priority=0
pbs.event().job.Rerunable=TRUE
pbs.event().job.Join_Path=oe
pbs.event().job.Keep_Files=oe
pbs.event().job.Mail_Points=abe
pbs.event().job.Variable_List=PBS_O_HOME=/home/johnson1,PBS_O_LANG=en_US.UTF-8,PBS_O_LOGNAME=johnson1,PBS_O_PATH=/usw/ansys_inc/2024R1/v241/fluent/bin:/usr/share/Modules/bin:/usr/lib64/ccache:/usr/local/bin:/usr/bin:/usr/local/sbin:/usr/sbin:/opt/pbs/bin:.:/usw/launch:/opt/local/bin,PBS_O_MAIL=/var/spool/mail/johnson1,PBS_O_SHELL=/bin/bash,PBS_O_WORKDIR=/home/johnson1/BENCH/FLUENT/combustor_12m,PBS_O_SYSTEM=Linux
pbs.event().job.queue=cfd1
pbs.get_local_nodename()=7vd0111x.mydomain.net
pbs.event().type=queuejob
pbs.event().hook_name=rejectInteractive
pbs.event().hook_type=site
pbs.event().requestor=johnson1
pbs.event().requestor_host=7vw0101x
pbs.event().user=pbsadmin
pbs.event().alarm=30
The second hook gets only a partial set of pbs.event() input, and since the hook uses one of the unsent values (pbs.event().hook_name in this case), it errors out, and no further hooks are run:
pbs.event().job.Job_Name=fluent
pbs.event().job.Account_Name=5d8c20e5ad650497d7d318bd3ed23dc2975b70bced2e388f8c76da2fbdc61a70
pbs.event().job.Checkpoint=u
pbs.event().job.Hold_Types=n
pbs.event().job.Join_Path=oe
pbs.event().job.Keep_Files=oe
pbs.event().job.Mail_Points=abe
pbs.event().job.Priority=0
pbs.event().job.Rerunable=1
pbs.event().job.Resource_List[select]=6:ncpus=28:mpiprocs=28
pbs.event().job.Resource_List[place]=excl
pbs.event().job.Variable_List=PBS_O_HOME=/home/johnson1,PBS_O_LANG=en_US.UTF-8,PBS_O_LOGNAME=johnson1,PBS_O_PATH=/usw/ansys_inc/2024R1/v241/fluent/bin:/usr/share/Modules/bin:/usr/lib64/ccache:/usr/local/bin:/usr/bin:/usr/local/sbin:/usr/sbin:/opt/pbs/bin:.:/usw/launch:/opt/local/bin,PBS_O_MAIL=/var/spool/mail/johnson1,PBS_O_SHELL=/bin/bash,PBS_O_WORKDIR=/home/johnson1/BENCH/FLUENT/combustor_12m,PBS_O_SYSTEM=Linux
pbs.event().job.Submit_arguments=<jsdl-hpcpa:Argument>-N</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>fluent</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>-P</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>cfd</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>-A</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>5d8c20e5ad650497d7d318bd3ed23dc2975b70bced2e388f8c76da2fbdc61a70</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>-q</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>cfd1</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>-l</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>select=6:ncpus=28:mpiprocs=28</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>-l</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>place=excl</jsdl-hpcpa:Argument><jsdl-hpcpa:Argument>combustor_12m.jou.sh</jsdl-hpcpa:Argument>
pbs.event().job.project=cfd
It is missing at least these:
pbs.event().job.queue=cfd1
pbs.get_local_nodename()=7vd0111x.mydomain.net
pbs.event().type=queuejob
pbs.event().hook_name=rejectDisabled
pbs.event().hook_type=site
pbs.event().requestor=johnson1
pbs.event().requestor_host=7vw0101x
pbs.event().user=pbsadmin
pbs.event().alarm=30
Which produces the error:
08/02/2024 15:13:23;0002;Server@7vd0111x;Svr;_pbs_python_event_set_attrval;can't set event attribute hook_name = rejectDisabled: event is unset
08/02/2024 15:13:23;0100;Server@7vd0111x;Hook;rejectDisabled;Failed to set event 'hook_name'.
08/02/2024 15:13:23;0100;Server@7vd0111x;Hook;;queuejob event: accept req by default
Any idea why the first hook to run gets complete pbs.event() data but not the second?