[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Condor-users] execute node choice and disk parameter



Answered my own question.


On Fri, 2010-03-12 at 11:40 -0800, kristian kvilekval wrote:
> I've set up a small condor test of a head node with two slave execute
> nodes.
> 
> Unfortunately I can't seem to get anything scheduled on the slaves
> as they are not chosen due to TARGET.Disk > 1500 parameter. See log
> below.
> 
> These execute nodes are booted remotely and use a NFS root,
> however they do have access to a shared NFS filesystem, and 
> each has a 500G scratch disk.

EXECUTE in condor_config is used to determine the file system 
where the space parameter is taken from.  Since this originally 
pointed to the ramdisk (stateless provisioning), the result was 0.

Pointed this to the local scratch space and it worked.

[snipped]

> Is it possible to change how condor calculates available disk space 
> on the slave execute nodes?