Re: [HTCondor-devel] information from condor

[Todd Tannenbaum <tannenba@xxxxxxxxxxx>]

On 12/21/2015 7:33 AM, åå wrote:
> Hi,
>
> I'd like to add some scheduling strategies on HTCondor, so I have to get
> the information about job queue, node status and so on. Now I don't know
> where to get these informations. Are there APIs or do I have to go to
> the source code?
>
> Regards & Thanks,
>
> Kathy
In HTCondor, everything (nodes, jobs, daemons) is represented with a 
data structure known as a "ClassAd" -- see http://is.gd/RLOmjK
To find out about node status and aggregate job queue information, you 
will want to send a query to the condor_collector, which is a database 
of ClassAds including ClassAds on the nodes (so-called machine or startd 
classads).  You can do this with the "condor_status" command-line tool, 
which when used with "-format" or the "-autoformat" command-line options 
is very easy to ingest its output into your code.  If you really do not 
want to use the condor_status tool, you can also query the 
condor_collector via a Python API or via a Web Services (soap) API. 
Details on both the Python and Web Serives API are in the manual in the 
chapter titled "Application Programming Interfaces".   The Python API is 
more scalable and widely used these days than the Web Services/SOAP 
interface, so I would suggest going with Python if possible.
See http://is.gd/1WQouF

To find out detailed job queue information (i.e. if you need to iterate 
through specific job records), you will need to query the condor_schedd 
daemons, which contain a database of ClassAds representing individual 
jobs.  You can do this with the "condor_q" command-line tool, and just 
like condor_status, it has "-format" and "-autoformat".  Or again you 
can use either the Python API or Web Service/SOAP  interface if you are 
adverse to command-line tools.
Hope this helps
Todd