Re: [Condor-users] Linux schedd on a windows pool

[Eric Abel <Eric.Abel@xxxxxxxxxx>]

Tim,

 

Thank you for the response. 

For point 1.), I am trying to integrate this into an existing pool which already has a running windows CM, and I believe I have all of the correct ALLOW settings, since it’s been running for about 9 months now.   

In point 2.) you brought up some points on which I would like a little clarification.  What do you mean “be certain that ID’s are consistent across your domain”?  What is the “slot_user” is this specific to a linux CM?

3.) I have a working credd host, which has been handling credentials on my pool.  Are the daemon level authorizations different for jobs submitted from a linux scheduler than a windows one?

 

Again, thanks for your response.

 

Eric

 

From: condor-users-bounces@xxxxxxxxxxx [mailto:condor-users-bounces@xxxxxxxxxxx] On Behalf Of Tim St Clair
Sent: Friday, March 02, 2012 10:41 AM
To: Condor-Users Mail List
Subject: Re: [Condor-users] Linux schedd on a windows pool

 

Eric - 

 

When working in a mixed env there are some tricks to keep in mind: 

 

1.) Make sure all nodes are still pointing to a consistent CM (Usually Linux is preferred for scalability), and have the correct ALLOW settings.  

2.) When doing an initial setup(for feasibilty) it's sometimes easier to use CLAIM_TO_BE and have windows use the normal slot_user account.  However you need to be certain that your ID's are consistent across your domain (namely for remote submission).  

3.) If you want to use run_as_owner on your windows machines then a CREDD will need to be installed on a windows box and daemon level authorizations need to be correct.

     

Hope this helps,

Tim

---

From: "Eric Abel" <Eric.Abel@xxxxxxxxxx>
To: "Condor-Users Mail List" <condor-users@xxxxxxxxxxx>
Sent: Friday, March 2, 2012 12:07:46 PM
Subject: [Condor-users] Linux schedd on a windows pool

Fellow condor users,

 

Following advice I received from condor developers regarding a maximum number of submitted jobs for my windows schedd, I decided to install condor on a linux box, and set that up as a scheduler.  I was able to get the latest version (7.6.6) installed, and can see the pool with condor_status, but strangely, the linux slots are not present even though startd is running (as is schedd) :

 

condor    3816     1  0 07:48 ?        00:00:01 /usr/sbin/condor_master -pidfile /var/lib/condor/log/condor.pid

condor    3817  3816  0 07:48 ?        00:00:00 condor_startd -f

condor    3818  3816  0 07:48 ?        00:00:00 condor_schedd -f

root      3820  3818  0 07:48 ?        00:00:00 condor_procd -A /var/lib/condor/log/procd_pipe.SCHEDD -R 10000000 -S 60 -C 115

 

When I try to do condor_store_cred –c add, I get

 

Operation failed.

    Make sure you have CONFIG access to the target Master.

 

And when I try to submit a job (not surprisingly) it doesn’t work, and it gives the following message:

 

Submitting job(s)

ERROR: Failed to connect to local queue manager

AUTHENTICATE:1003:Failed to authenticate with any method

 

my config.local (which I pulled from a windows schedd) reads as follows:

 

CREDD_CACHE_LOCALLY = True

ALLOW_CONFIG = *.varian.com,$(IP_ADDRESS),$(CONDOR_HOST),Administrator@*

ALLOW_WRITE = $(FULL_HOSTNAME),$(IP_ADDRESS),*.varian.com

LOCAL_CONFIG_FILE = $(LOCAL_DIR)/condor_config.local

STARTER_ALLOW_RUNAS_OWNER = True

SEC_CLIENT_AUTHENTICATION_METHODS = NTSSPI, PASSWORD

SEC_CONFIG_NEGOTIATION = REQUIRED

SEC_CONFIG_AUTHENTICATION = REQUIRED

SEC_CONFIG_ENCRYPTION = REQUIRED

SEC_CONFIG_INTEGRITY = REQUIRED

 

Any suggestions on this would be greatly appreciated.

 

Thanks,

 

Eric

 

_______________________________________________
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/