Mailing List ArchivesPublic Access |
![[Computer Systems Lab]](https://www-auth.cs.wisc.edu/pics/csl_logo.gif){kind=logo}
|
Mailing List ArchivesPublic Access |
|
Condor doesn't provide this ability. The best way to do this that I can think of is to wrap the daemons in a script that determines what pool they should join, sets CONDOR_HOST appropriately, then starts condor_master.-- JaimeOn Jul 7, 2005, at 3:04 PM, Jyotishman Pathak wrote:I agree with you here as well. But again, flocking is based on static configurations. If you have 2 pools, A & B, you configure their respective managers beforehand. What I am looking for is more of a dynamic situation -- as soon as the node (which is going to execute jobs) is alive, I want it to be a part of *some* condor pool based on some criteria.
- JyotiOn 7/7/05, Rajesh Rajamani < raj@xxxxxxxxxx> wrote:I'm tempted to answer yes, but it is possible to get a resource/schedd
accept/schedule jobs from/on a schedd/resource in another pool. You can
do this by enabling flocking between the two (or more) pools -
http://www.cs.wisc.edu/condor/manual/v6.7/5_2Connecting_Condor.html
Jyotishman Pathak wrote:
> I agree. So, is it not possible for a particular node to be a member for
> more than one pool or subscribe to more than one manager?
>
> - Jyoti.
>
> On 7/7/05, *Rajesh Rajamani* <raj@xxxxxxxxxx <mailto: raj@xxxxxxxxxx>> wrote:
>
> Condor supports only one Central Manager per pool.
>
> Jyotishman Pathak wrote:
> > Hi Rajesh,
> >
> > Thank you for your reply and the pointers. It sure helped. However, I
> > think in a grid infrastructure and assuming that a particular
> node will
> > report to many central managers, it is hard to envision *all* the
> values
> > for CONDOR_HOST beforehand. In other words, if a node comes alive, it
> > should be dynamically inducted into a condor pool based on some
> > heuristics (e.g., locality), as opposed to hardwiring the condor
> control
> > manager that it will report to.
> >
> > I am not sure if my point makes sense here. Please let me know
> your views.
> >
> > Regards,
> > Jyoti.
> >
> >
> >
> >
> > On 7/7/05, *Rajesh Rajamani* < raj@xxxxxxxxxx
> <mailto: raj@xxxxxxxxxx> <mailto: raj@xxxxxxxxxx
> <mailto:raj@xxxxxxxxxx >>> wrote:
> >
> > >
> > > Imagine I have a condor pool of 50 machines which are
> > monitored/managed
> > > by some "parent node". Based upon the status of these 50
> nodes,
> > the jobs
> > > are submitted to them.
> >
> > When you say "monitored/managed", do you mean the Condor
> system or a
> > third-party management software? I'm assuming it is the
> former. Check
> > out
> >
> http://www.cs.wisc.edu/condor/manual/v6.7/3_1Introduction.html#SECTION00411000000000000000
>
> >
> < http://www.cs.wisc.edu/condor/manual/v6.7/3_1Introduction.html#SECTION00411000000000000000>
> > for information about the Central Manager
> >
> > > Now, think of a new node (51st machine) that has been
> provisioned by
> > > some resource allocator. This node was not a part of the
> original
> > condor
> > > pool. In other words, originally the "parent node" is not
> aware
> > of this
> > > new node. My question is -- how do I make the parent node
> aware
> > of the
> > > fact that a new machine has arrived (and it needs to be
> included
> > in the
> > > pool and can be used for job execution, monitoring etc.) ?
> >
> > Resources report to the collector, which then works with the
> negotiator
> > to schedule jobs. All you need to do is use the proper value
> for
> > CONDOR_HOST in the config file of the new "node". See
> > http://12.22.53.199:6080/display/CONDOR/CONDOR_HOST for details.
> >
> >
> > --
> > Rajesh Rajamani
> > Senior Member of Technical Staff
> > Direct : +1.408.321.9000
> > Fax : +1.408.321.9030
> > Mobile : +1.650.218.9131
> > raj@xxxxxxxxxx <mailto:raj@xxxxxxxxxx> <mailto: raj@xxxxxxxxxx
> <mailto: raj@xxxxxxxxxx>>
> >
> >
> > Optena Corporation
> > 2860 Zanker Road, Suite 201
> > San Jose, CA 95134
> > www.optena.com <http://www.optena.com> < http://www.optena.com>
> >
> >
> > This electronic transmission (and any attached documents)
> contains
> > information from Optena Corporation and is for the sole use
> of the
> > individual or entity it is addressed to. If you receive this
> message in
> > error, please notify me and destroy the attached message (and
> all
> > attached documents) immediately.
> >
> >
>
> --
> Rajesh Rajamani
> Senior Member of Technical Staff
> Direct : +1.408.321.9000
> Fax : +1.408.321.9030
> Mobile : +1.650.218.9131
> raj@xxxxxxxxxx <mailto: raj@xxxxxxxxxx>
>
>
> Optena Corporation
> 2860 Zanker Road, Suite 201
> San Jose, CA 95134
> www.optena.com < http://www.optena.com>
>
>
> This electronic transmission (and any attached documents) contains
> information from Optena Corporation and is for the sole use of the
> individual or entity it is addressed to. If you receive this message in
> error, please notify me and destroy the attached message (and all
> attached documents) immediately.
>
>
--
Rajesh Rajamani
Senior Member of Technical Staff
Direct : +1.408.321.9000
Fax : +1.408.321.9030
Mobile : +1.650.218.9131
raj@xxxxxxxxxx
Optena Corporation
2860 Zanker Road, Suite 201
San Jose, CA 95134
www.optena.com
This electronic transmission (and any attached documents) contains
information from Optena Corporation and is for the sole use of the
individual or entity it is addressed to. If you receive this message in
error, please notify me and destroy the attached message (and all
attached documents) immediately._______________________________________________Condor-users mailing list+----------------------------------+---------------------------------+
| Jaime Frey | Public Split on Whether |
| jfrey@xxxxxxxxxxx | Bush Is a Divider |
| http://www.cs.wisc.edu/~jfrey/ | -- CNN Scrolling Banner |
+----------------------------------+---------------------------------+