[GE users] qstat -j -xml output information

Wilfried Gänsheimer wilfried at gaensheimer.de
Tue May 6 18:11:04 BST 2008


    [ The following text is in the "ISO-8859-1" character set. ]
    [ Your display is set for the "ISO-8859-10" character set.  ]
    [ Some special characters may be displayed incorrectly. ]

Hallo Michael,

some weeks ago I  have  sent some  questions regarding the XML output of 
GE6.1u4
It's not exactly a deep analysis, but maybe you could add that as 
requirements or issues to your
overview ...

Here is a cut&paste of the relevant part:

+++++snip+++++

I have tried 6.1u4 and all the info is back again (priority, predecessor,
project, execution time). Thanks  :-) 

One thing puzzled me, though: Priority in the xml format is the actual
priority value + 1024.

One other remark: is there a special reason, why the xml output for
"qstat -f -xml" does not show the submission time (nor a execution time
when -a was used) for running jobs.

Knowing the submissions time and the start time at one glance helps
when issues with job order needs to be clarified
(users complaining about long waiting time while other jobs started later
are executing).

The other way round "qstat -f -j N" does report submission (and execution
time if applicable), but not the start time. Getting that together
is a little bit tedious (w/o -xml)

BTW., the submission time format differs from qstat -f -xml:
      <JB_submission_time>2008-04-09T18:40:41</JB_submission_time>
vs. qstat -f -xml -j
      <JB_submission_time>1207819256</JB_submission_time>

Last but not least I do not find info about the current status of a job
like "wq" or "hqw" or "r" (i would be interested in getting a hint about
hold state) in the xml output. e.g. the output of "qstat -xml -f -j N"
does not at all change when a hold gets set or released (apart from
JB_version being incremented). I would expect to get an indication about a
set hold.

Just discovered the job state is added when the "qstat -xml -f -r" flags
are used. "qstat -xml -f -r -j N" does not report anything about a
hold, though.

Is there already a issue filed about that? Should I?

+++++ snip +++++

Thanks, bye
Wilfried

Michael Pospisil - Sun Microsystems - Prague Czech Republic wrote:
> Hello Chris (and others who replied),
> thanks for the reply and the useful suggestions. I still have one 
> important question based on your requirements though, and I think it 
> is important to clarify this before I continue.
> In your response, you wrote that "Extra data is fine but making  
> "qstat" and "qstat -xml" report the same data should be a priority."
> Does this "Extra data" mean that "qstat -xml" can have more 
> information that the "qstat" output, or do you require that all the 
> output in "qstat -xml" is shown in "qstat" and vice versa?
> Thanks for the clarification,
>
> Michael
>
> Chris Dagdigian wrote:
>
>> Hi Michael,
>>
>> Happy to hear you are working on this!
>>
>> I only have one absolutely solid requirement:
>>
>> Any data reported via "qstat -j" *MUST* also be reported when the "- 
>> xml" switch is used, inconsistency is unacceptable and breaks with 
>> the  initial promises made when 6.0 came out that "-xml" was a 
>> feature that  would simply alter the formatting of the information 
>> being sent back  from the qmaster.
>>
>> XML output is a huge win for everyone and a massively useful output  
>> format option. I'm happy to see extra/additional data reported when 
>> "- xml" is used but I'm very unhappy when I see things via "qstat" 
>> that  are not reported when the same command is run with the 
>> additional xml  output switch appended.
>>
>> So my main request is for consistency. Extra data is fine but making  
>> "qstat" and "qstat -xml" report the same data should be a priority.
>>
>> Regards,
>> Chris
>>
>>
>>
>> On Apr 30, 2008, at 7:42 AM, Michael Pospisil - Sun Microsystems -  
>> Prague Czech Republic wrote:
>>
>>> Hello all,
>>> In an attempt to rectify the state of the qstat -j -xml output, I 
>>> am  creating a list of all information that is deemed as 
>>> "important"  output. In the 61u4 output, the following fields are 
>>> found in qstat - xml:
>>>
>>> JB_job_number
>>> JB_exec_file
>>> JB_submission_time
>>> JB_owner
>>> JB_uid
>>> JB_group
>>> JB_gid
>>> JB_account
>>> JB_merge_stderr
>>> JB_mail_list, MR_user, MR_host
>>> JB_notify
>>> JB_job_name
>>> JB_jobshare
>>> JB_shell_list, PN_path, PN_host, PN_file_host, PN_file_staging
>>> JB_env_list, __SGE_PREFIX__O_HOME, __SGE_PREFIX__O_LOGNAME,  
>>> __SGE_PREFIX__O_PATH, __SGE_PREFIX__O_SHELL, __SGE_PREFIX__O_TZ,  
>>> __SGE_PREFIX__O_MAIL, __SGE_PREFIX__O_HOST, __SGE_PREFIX__O_WORKDIR
>>>
>>> JB_script_file
>>> JB_ja_tasks, JAT_status, JAT_task_number
>>> JB_deadline
>>> JB_execution_time
>>> JB_checkpoint_attr
>>> JB_checkpoint_interval
>>> JB_reserve
>>> JB_mail_options
>>> JB_priority
>>> JB_restart
>>> JB_verify
>>> JB_script_size
>>> JB_verify_suitable_queues
>>> JB_soft_wallclock_gmt
>>> JB_hard_wallclock_gmt
>>> JB_override_tickets
>>> JB_version
>>>
>>> I have already received a request to add back in JAT_start_time as  
>>> well as the queue name through either JAT_master_queue or JG_qname.
>>>
>>> Now I would like to open up the discuss to you, the users, to find  
>>> out what is still missing or what you would find useful in the 
>>> qstat  -j -xml output. Please however note, what we are discussing 
>>> here is  only the output for the qstat -j -xml command and will not 
>>> affect  the current output of qstat -j.
>>> It would be most interesting to hear from users who are actually  
>>> using the qstat -j -xml output and find out which fields are indeed  
>>> useful and what information needs to be added back to the data that  
>>> was filtered out of the -xml output in 61u3.
>>>
>>> Feel free to make any suggestions that you think will better the  
>>> results the of qstat -j -xml output.
>>> Thanks in advance for any input, and hopefully we can finally come  
>>> to a solution that suits all users.
>>> Best regards,
>>>
>>> Michael
>>>
>>>
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
>>> For additional commands, e-mail: users-help at gridengine.sunsource.net
>>
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
>> For additional commands, e-mail: users-help at gridengine.sunsource.net
>>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
> For additional commands, e-mail: users-help at gridengine.sunsource.net
>


---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net




More information about the gridengine-users mailing list