Re: [HTCondor-users] Python bindings questions

[Jaime Frey <jfrey@xxxxxxxxxxx>]

> On Feb 14, 2019, at 8:25 AM, Stephen Jones <sjones@xxxxxxxxxxxxxxxx> wrote:
> 
> Hi,
> 
> We use HTCondor-CE at our tier 2 grid site.
> 
> I think the BDII software that comes with HTCondor-CE interacts with HTCondor-CE via "python bindings", perhaps?
> 
> Here's the issue. We have HTCondor CE and a condor batch system on the same computer.
> 
> The BDII (which I think is part of the CE) gets its config from the batch system, not the CE.
> 
> So to fix this, I have to put the config in the "wrong" place... (and then I need to change a puppet module to support his kludge.)
> 
> I don't like that much. The python bindings _should_ read the config from the right place. But perhaps they don't have a way to know?
> 
> Note that on sites (e.g. CERN) that  put the BDII on a different computer to the batch system, this, perhaps, doesn't occur? I don't know.
> 
> Anyway, it's something to watch out for.... can it be fixed? Or am I doing something wrong?

The BDII package is code contributed by CERN.
Our understanding is that BDII wants information about the local cluster that the HTCondor-CE doesnât have. The python script in question queries that clusterâs central manager, locates all of the schedds fronted by an HTCondor-CE (CERNâs pools have multiple such schedds), and pulls the information it wants.

 - Jaime