Mailing List Archives
Authenticated access
|
|
|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Condor-users] Out of servers - 0 jobs matched, 2 jobs idle, 1 jobs rejected
- Date: Mon, 18 Apr 2005 11:11:39 -0500
- From: Jaime Frey <jfrey@xxxxxxxxxxx>
- Subject: Re: [Condor-users] Out of servers - 0 jobs matched, 2 jobs idle, 1 jobs rejected
On Apr 18, 2005, at 9:47 AM, Garry Smith wrote:
I am trying to submit a C code to the condor vanilla universe, running
on a
cluster with NFS filesystem. If I change the universe to standard and
link
the C code to the condor libraries, the execution works successfully.
The failing submission gives:
tail -f SchedLog:
<snip>
4/18 15:37:27 Checking consistency running and runnable jobs
4/18 15:37:27 Tables are consistent
4/18 15:37:27 Out of servers - 0 jobs matched, 2 jobs idle, 1 jobs
rejected
Submit file:
########################
# Submit description file for hello program
########################
Executable = hello
Universe = vanilla
Output = $ENV(HOME)/hello.c.$(Cluster).$(Process)
Log = $ENV(HOME)/hello.c.$(Cluster).$(Process)
Queue 2
Have you run 'condor_q -analyze'? That should be your first step when
jobs won't run. If it indicates that most/all of your machines are
rejecting your job because of requirements, or vice versa, you should
check the requirements expressions of both the job and the machines.
By default on unix, condor assumes you want to transfer job files via a
shared filesystem, but that you don't have one. Have you configured
condor to know about NFS:
http://www.cs.wisc.edu/condor/manual/v6.7/
3_3Configuration.html#SECTION00435000000000000000
If you haven't, then the FileSystemDomain attribute of the job and
machine ads will be set to the hostname of the local machine and jobs
won't match with remote machines.
Another option is to tell condor to transfer job files:
http://www.cs.wisc.edu/condor/manual/v6.7/
2_5Submitting_Job.html#SECTION00354000000000000000
All of this filesystem stuff is irrelevant to standard universe, since
all the I/O calls get sent back to the submit machine to be executed.
+----------------------------------+---------------------------------+
| Jaime Frey | Public Split on Whether |
| jfrey@xxxxxxxxxxx | Bush Is a Divider |
| http://www.cs.wisc.edu/~jfrey/ | -- CNN Scrolling Banner |
+----------------------------------+---------------------------------+