I don't think the AggregatorData section requires anything to be there
in this case. I've added Laura to the CC list in case I'm missing
something obvious, or if the WebMDS reported error is indicating
something else.
WebMDS doesn't consider it an error; it's just reporting that the Index
service is configured to poll for that data but doesn't actually have
the data. This could mean that the registration happened recently
enough that the Index server hasn't polled yet or that, for whatever
reason, the Index server tried to poll for the data but didn't get any.
If the Index server tried to poll but got an error, that error should be
reflected in the Index server's logs (i.e., the log files belonging to
the Globus container in which the Index server was running). You can
also try to make the same kind of request that the Index server would
make when polling; in this case, it would be:
wsrf-get-property -a -z none -e epr.xml
"{http://mds.globus.org/glue/ce/1.1}GLUECE";
Where epr.xml is a file into which you've copied the EPR from the
MemberServiceEPR associated with this service group entry (you've
included about half of that in your raw xml output, so I assume it would
be:
<MemberServiceEPR>
<ns8:Address
xmlns:ns8="http://schemas.xmlsoap.org/ws/2004/03/addressing";>https://foo.ari.uni-heidelberg.de:8443/wsrf/services/ManagedJobFactoryService</ns8:Address>
<ns9:ReferenceProperties
xmlns:ns9="http://schemas.xmlsoap.org/ws/2004/03/addressing";>
<ns1:ResourceID
xmlns:ns1="http://www.globus.org/namespaces/2004/10/gram/job";>Multi</ns1:ResourceID>
</ns9:ReferenceProperties>
<ns10:ReferenceParameters
xmlns:ns10="http://schemas.xmlsoap.org/ws/2004/03/addressing"/>
</MemberServiceEPR>
-- Laura
P.S. Do you have some local way of resolving the name
foo.ari.uni-heidelberg.de? A DNS lookup of that name doesn't work for me.
On 08/06/2007 08:15 AM, Neill Miller wrote:
On Mon, Aug 06, 2007 at 04:37:33PM +0200, Thomas Bruesemeister wrote:
In the Information column of the Servicegroup overview
the following entry appears for the Multi Job GRAM :
Aggregator entry with no content from
https://foo.ari.uni-heidelberg.de:8443/wsrf/services/ManagedJobFactoryService
And in fact it contains no aggregator data:
==================================
<ns1:ResourceID
xmlns:ns1="http://www.globus.org/namespaces/2004/10/gram/job";>Multi</ns1:ResourceID>
</ns9:ReferenceProperties>
<ns10:ReferenceParameters
xmlns:ns10="http://schemas.xmlsoap.org/ws/2004/03/addressing"/>
</ns1:MemberServiceEPR>
<ns1:Content xsi:type="ns11:AggregatorContent">
<ns11:AggregatorConfig>
<ns11:GetResourcePropertyPollType>
<!-- Specifies that the index should refresh information
every 60000 milliseconds (once per minute) -->
<ns11:PollIntervalMillis>60000</ns11:PollIntervalMillis>
<!-- specifies the resource property that should be
aggregated, which in this case is the GLUE cluster
and scheduler information RP -->
<ns11:ResourcePropertyName>glue:GLUECE</ns11:ResourcePropertyName>
</ns11:GetResourcePropertyPollType>
</ns11:AggregatorConfig>
<ns11:AggregatorData/>
...
=======================================
Is this intended behaviour for the Multi GRAM type?
I don't think the AggregatorData section requires anything to be there
in this case. I've added Laura to the CC list in case I'm missing
something obvious, or if the WebMDS reported error is indicating
something else.
I used the mds-gluerp-configure configure tool which adds a
GangliaElementProducer for every MJFS.
At least GT <= 4.0.3 also published Ganglia information for every MJFS
when configuring it via the gluerp.xml. But I agree with you here it should
only be published for the DefaultIndexService.
Can you post your generated configuration? If you have multiple
instances of reporting Ganglia data in the config, it will be reported
multiple times. You should be able to get the scheduling data without
repeating that data. I just need to be sure of what's going on by
looking at the config though.
thanks,
-Neill.
