A simple solution is to install the schedd on a remote, stable
server (you can use the same computer as the collector/negotiator but there are
security considerations involved) and submit to that machine by one of the
following options:
1. shh/remote into this box, submit the job, remote back in to
get the results
Simple, effective, security issues involved, would involve
transferring the required executable/submit files manually to the submit
machine
2. use the -remote submission argument but submit from your
machine.
Even simpler than 1 but does have some foibles and is ill
suited to tasks where many jobs are part of a single submission since the input
files and executable are duplicated for every job.
You must use the condor_transfer_data command to retrieve the
results.
3. Write your own custom front end which runs on the scheduling
server and deals with submitting the jobs and getting you back the data one the
original machine is reconnected.
Much more up front effort, gives you a platform configured for
your needs though.
In all these cases your scheduling machine must have sufficient
space available to it to handle all the data involved and it becomes a single
point of failure unless you have several such machines. It also becomes a choke
point, but this is likely to make a difference only if you have > 100 execute
nodes and/or your data requirements for transfer in/out of the execute nodes on
start/finish are high.
From: condor-users-bounces@xxxxxxxxxxx
[mailto:condor-users-bounces@xxxxxxxxxxx] On Behalf Of Tao.3.Chen@xxxxxxxxxxxxxxxxxxxxxxxxxxx
Sent: 05 August 2009 08:03
To: Condor-Users Mail List
Subject: [Condor-users] Antwort: Re: can jobs be executed while the
condor submitter isturned off ?
Hi, Alex,
Thank you so much!
I also checked the condor workshop presentation, in it the job
startup steps are as follows:
1. Startd sends collector ClassAd describing itself. (The Schedd does as
well, but it has nothing interesting to say yet.)
2. The user calls condor_submit to submit a job. The job is handed off to
the schedd and condor_submit returns.
3. The schedd alerts the collector that it now has a job waiting.
4. The negotiator asks the collector for a list machines able to run jobs and
schedd queues with waiting jobs.
5. The negotiator contacts the schedd to learn about the waiting job.
6. The negotiator matches the waiting job with the waiting machine.
7. The negotiator alerts the schedd and the startd that there is a match.
8. The schedd contacts the startd to claim the match.
9. The schedd starts a shadow to monitor the job.
10. The startd starts a starter to start the job.
11. The starter and the shadow contact each other.
11. The starter starts the job.
12. If the job is using the Condor syscall library (typically through being
condor_compiled), it will contact the shadow to access necessary files.
hmm....... So I think you are right, the whole condor system
architecture is designed as the above, it is impossible to change. But I am not
sure if there is any special design which is adapted to the submitter off?Also,
for my work now ,I need this kind of system which would allow the
submitter unplug or shutdown, and the jobs work well, any ideas?
Good
Luck!
Tao
|
"Alas,
Alex [FEDI]" <aalas@xxxxxxxxxxxxx>
Gesendet von:
condor-users-bounces@xxxxxxxxxxx
08/04/2009
09:12 PM
|
Bitte antworten an
Condor-Users Mail List <condor-users@xxxxxxxxxxx>
|
|
|
An
|
"Condor-Users
Mail List" <condor-users@xxxxxxxxxxx>
|
|
Kopie
|
|
|
Thema
|
Re:
[Condor-users] can jobs be executed while the condor submitter
isturned off ?
|
|
No,
In my experience the submitter needs to be online until the last jobs’
execution is completed. That is using the default configuration after
installing condor in the execute node. Every time I tried either to shutdown or
unplug the computer after submitting a set of jobs, the jobs go immediately
into the hold state. The reason why I leave it to some more experienced condor
user.
I
hope this statement at helps a little in your quest of answers,
Respectfully,
Alex.
From:
condor-users-bounces@xxxxxxxxxxx [mailto:condor-users-bounces@xxxxxxxxxxx] On
Behalf Of Tao.3.Chen@xxxxxxxxxxxxxxxxxxxxxxxxxxx
Sent: Tuesday, August 04, 2009 5:06 AM
To: condor-users@xxxxxxxxxxx
Subject: [Condor-users] can jobs be executed while the condor submitter
isturned off ?
Hi,
I have read the manual and the some presentations
carefully, found that during the execution of job, the schedd daemon should
watch the progress of executing machine. And for the jobs in the queue, the
schedd daemon will maintain the persistent queue, contact the available
machines and match the job with executing machines.
does that mean if I submit several jobs in the
queue, the submitter should keep online and contact with the central manager
and the executing machine?
Is that possible after submitting the jobs, the
submitter could be shut down, and the central manager takes the response for
the whole jobs? Or are there any other kinds of software could achieve
this target?
Any ideas would be appreciated! Thanks in advance!
Tao _______________________________________________
Condor-users mailing list
To unsubscribe, send a message to condor-users-request@xxxxxxxxxxx with a
subject: Unsubscribe
You can also unsubscribe by visiting
https://lists.cs.wisc.edu/mailman/listinfo/condor-users
The archives can be found at:
https://lists.cs.wisc.edu/archive/condor-users/
![[Computer Systems Lab]](http://www.cs.wisc.edu/pics/csl_logo.gif){kind=logo}