Re: [Condor-users] Condor training

[Poornima Pochana <ppreddy@xxxxxxx>]

Hi John,

We, at UAB are trying to set up a generic condor environment as a pilot project. My experience with Condor so far has been in trying to:

1. migrate sge job as a condor workflow

2. set up a personal condor environment

Following U of Wisconsin's recommendation, I started by setting up a Personal Condor instance. This was very easy with the condor yum repos on a centos 6 vm. Just did a 'yum install condor' and 'condor start'. Did not even had to modify the config files. All my workflow migration and tests were performed on this personal condor. 

As a next step, I tried to set up a Personal Condor pool (with 2 linux vms - debian, centos). During this process, we got to understand that 3 things were critical to have a running condor pool:

1. ALLOW_WRITE and ALLOW_READ settings in condor_config.local

  - need to explicitly mention the host names/ip address of machines in the pool

2. A proper functioning DNS and /etc/hosts config 

  - should have the ip address of the machine in /etc/hosts, just having 127.0.0.1 won't work

3. Firewall, should know which ports to open  

   -  we had success resulting from turning off the firewall in the private environment, i.e., on the 2 vms.

   - we are currently faced with firewall problems, or more precisely NAT problems, in our cross-campus test pool.  We are exploring flocking as a potential solution.

     * the above firewall problem is not the firewall on the condor host (collector) but in trying to schedule jobs to compute nodes that are behind a firewall

On the nitty-gritty side, things I found extremely helpful in trying to migrate the sge job

*  condor_submit manual

* to query class ads and to trouble shoot job errors- condor_status -l, condor_q -analyze

* file transfer mechanism of condor 

* specify executables for more than one platform with the $$(OpSys)) class-ad attribute (Thanks to Brooklin Gore from UW).

* macro expansion at submit time - to send a platform-specific file at submission time - 

Thanks,

Poornima.

p.s. http://projects.uabgrid.uab.edu/condor/wiki/UnderstandingCondor

On Dec 13, 2011, at 9:18 AM, Todd Tannenbaum wrote:

On 12/13/2011 8:37 AM, Ben Cotton wrote:

On Mon, Dec 12, 2011 at 11:57 AM, John Krabacher

<jkrabacher3@xxxxxxxxxxxx>  wrote:

Does anyone know of any training available for Condor?

Matthew Farrellee's blog[0] has a lot of good introductory posts. You

can also check out the presentations that the Condor team does at

Condor Week every year[1]

[0] http://spinningmatt.wordpress.com

[1] http://research.cs.wisc.edu/condor/CondorWeek2011/tuesday_condor.html

Good luck!

BC

Re the Condor Week tutorials - If you like to have a talking head/audio to go along with the tutorial slides, a couple of years ago some of the basic introductory tutorials were recorded.  See
 http://research.cs.wisc.edu/condor/tutorials/videos/

Once you get up and going, there are a log of how-to recipes on the Condor wiki at
 https://condor-wiki.cs.wisc.edu/index.cgi/wiki?p=HowToAdminRecipes

regards,
Todd


--
Todd Tannenbaum                       University of Wisconsin-Madison
Center for High Throughput Computing  Department of Computer Sciences
tannenba@xxxxxxxxxxx                  1210 W. Dayton St. Rm #4257
Phone: (608) 263-7132                 Madison, WI 53706-1685
_______________________________________________
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/

Attachment: smime.p7s
Description: S/MIME cryptographic signature