Basically, we want to make the NOOP nodes less "special-casey". Right now DAGMan writes "fake" events to the log file for NOOP nodes, and, in general we want to move away from DAGMan writing events. It should also simplify the DAGMan code if we eventually can remove the "don't actually submit" option.
Kent From: HTCondor-users <htcondor-users-bounces@xxxxxxxxxxx> on behalf of Scott Koranda <skoranda@xxxxxxxxx>
Sent: Thursday, February 18, 2016 11:53 AM To: HTCondor-Users Mail List Subject: Re: [HTCondor-users] proposed change to DAG NOOP nodes On Thu, Feb 18, 2016 at 11:50 AM, R. Kent Wenger
<wenger@xxxxxxxxxxx> wrote:
Currently, nodes that are specified as NOOPs in a DAG file are not actually submitted to HTCondor -- DAGMan itself "marks" the nodes as successful immediately. We want to move away from this, and actually submit such jobs to HTCondor (but still have them be NOOPs). My apologies if this has already been discussed and I missed it, but why do you want to submit such jobs to HTCondor? To what advantage?
Thanks,
Scott K
|