[Swift-user] multisite passive workers

[Allan Espinosa]

Ahh...

I'm just getting one URL service .  Is it because I have the same url
for each site?

With this, for each entry, should I specify a port manually like
url="https://communicado.ci.uchicago.edu:1984" ,
url="https://communicado.ci.uchicago.edu:1985" ... ?

swift-r3649 cog-r2890

RunID: osg
Progress:
Progress:  Initializing:1  Selecting site:817
Progress:  Selecting site:1273  Initializing site shared directory:27
Stage in:1
Passive queue processor initialized. Callback URI is http://128.135.125.17:50003
Progress:  Selecting site:1100  Initializing site shared directory:24
Stage in:9  Submitted:168
Progress:  Selecting site:1058  Initializing site shared directory:23
Stage in:4  Submitted:216
Progress:  Selecting site:1058  Initializing site shared directory:23
Stage in:4  Submitted:216
Progress:  Selecting site:1041  Initializing site shared directory:23
Stage in:21  Submitted:216


site entries;
  <pool handle="UJ-OSG">
    <execution provider="coaster" url="communicado.ci.uchicago.edu"
        jobmanager="local:local" />

    <profile namespace="globus" key="workerManager">passive</profile>

    <profile namespace="karajan" key="initialScore">20.0</profile>
    <profile namespace="karajan" key="jobThrottle">2.7</profile>

    <profile namespace="globus" key="lowOverallocation">36</profile>

    <gridftp  url="gsiftp://osg-ce.grid.uj.ac.za"/>
    <workdirectory>/nfs/data/osg_store/engage-scec/swift_scratch</workdirectory>
  </pool>

  <pool handle="UMissHEP">
    <execution provider="coaster" url="communicado.ci.uchicago.edu"
        jobmanager="local:local" />

    <profile namespace="globus" key="workerManager">passive</profile>

    <profile namespace="karajan" key="initialScore">20.0</profile>
    <profile namespace="karajan" key="jobThrottle">0.44</profile>

    <profile namespace="globus" key="lowOverallocation">36</profile>

    <gridftp  url="gsiftp://umiss001.hep.olemiss.edu"/>
    <workdirectory>/osgremote/osg_data/engage-scec/swift_scratch</workdirectory>
  </pool>

  <pool handle="USCMS-FNAL-WC1">
    <execution provider="coaster" url="communicado.ci.uchicago.edu"
        jobmanager="local:local" />

    <profile namespace="globus" key="workerManager">passive</profile>

    <profile namespace="karajan" key="initialScore">20.0</profile>
    <profile namespace="karajan" key="jobThrottle">68.7</profile>

    <profile namespace="globus" key="lowOverallocation">36</profile>

    <gridftp  url="gsiftp://cmsosgce3.fnal.gov"/>
    <workdirectory>/uscms_grid/data/engage-scec/swift_scratch</workdirectory>
  </pool>

  <pool handle="UTA_SWT2">
    <execution provider="coaster" url="communicado.ci.uchicago.edu"
        jobmanager="local:local" />

    <profile namespace="globus" key="workerManager">passive</profile>

    <profile namespace="karajan" key="initialScore">20.0</profile>
    <profile namespace="karajan" key="jobThrottle">9.46</profile>

    <profile namespace="globus" key="lowOverallocation">36</profile>

    <gridftp  url="gsiftp://gk04.swt2.uta.edu"/>
    <workdirectory>/cluster/grid/data/engage-scec/swift_scratch</workdirectory>
  </pool>

  <pool handle="WQCG-Harvard-OSG">
    <execution provider="coaster" url="communicado.ci.uchicago.edu"
        jobmanager="local:local" />

    <profile namespace="globus" key="workerManager">passive</profile>

    <profile namespace="karajan" key="initialScore">20.0</profile>
    <profile namespace="karajan" key="jobThrottle">0.22</profile>

    <profile namespace="globus" key="lowOverallocation">36</profile>

    <gridftp  url="gsiftp://tuscany.med.harvard.edu"/>
    <workdirectory>/osg/storage/data/engage-scec/swift_scratch</workdirectory>
  </pool>


2010/10/4 Mihael Hategan <hategan at mcs.anl.gov>:
> On Mon, 2010-10-04 at 13:21 -0500, Allan Espinosa wrote:
>> Hi,
>>
>> I noticed that there is only one callback URI when I have multiple
>> sites with a passive worker configuration.  How does the coaster
>> service know which site the registering worker belong to?
>
> You should get one URL for each service, unless there is some failure in
> starting a service.
>
>>
>> My test jobs maybe failing because the site swift initializes for a
>> particular worker maybe the wrong one.
>>
>> thanks,
>> -Allan