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

Re: [HTCondor-users] HTCondor: job outputs not cleaned up?

On 08/02/2016 14:55, Thomas Hartmann wrote:
Hi all,

nodes in our test cluster are filling up with left-overs from old jobs.
We are wondering, if we have missed something in the configuration but
would expect, that normally the jobs would be cleaned up?

An example node has left hardly any space
/dev/sda6 71G 65G 2.2G 97% /var
wn13-test > find /var/jobs/ -mtime +7 | wc -l

with the majority being old files from previous jobs.

/var/jobs/ isn't a directory that HTCondor writes to (at least in its out-of-the-box configuration). A normal system uses /var/lib/condor/execute and /var/lib/condor/spool, and is usually pretty good about tidying up after itself.

What sort of directory structure are you seeing under /var/jobs/ ?

Also, are you looking here at the node where the jobs are submitted, or one of the nodes where jobs are executed?