The HPCBP contains a BasicFilter element as an extension to the GetFactoryAttributes operation request. This allows the requesting entity to select either the list of activities and/or the list of contained resources. There is no additional filtering operations. For large clusters or large job queues the returned XML document will become very large. To ensure that the returned document just contains information that the client is interested in, it is proposed to specify an extension that will filter the returned results. Some search options could include (within the HPCBP activities can be equated to jobs): * Return activities just relating to the currently authorized user * Return activities just relating to a specified user(s) * Return activities in a specified state * Return activities with a specified job id range [assumes there is an underlying numerical job id - not mandated in any specification - but fairly common] * Return activities active within a specified time/date range. WRT to reducing the size of the resource section of the document... I would propose exploiting the homogeneous nature of many clusters by eliminating duplicate elements. So a cluster containing 16 identical nodes and a head node would return a document (psudeo form) <nodes count="16"> <resource details> Compute Node information </resource details> </nodes count> <nodes count="1"> <resource details> Head node information </resource details> </nodes count> Instead of returning an XML document with 17 resource details elements. In addition the returned resource details could be filtered by 'used' and 'available' if they were running a job or were available for running a job. If there are any further uses that come to mind please respond to this. Once we have some consensus I'll draft some XML for discussion on a later call. Steven