Weekly Meeting

Name: Weekly Meeting
Start: 2007-06-27T14:00:00+02:00
End: 2007-06-27T15:00:00+02:00
Location: RAL

Wednesday 27 Jun 2007, 14:00 → 15:00 Europe/Zurich

CR02 (RAL)

Steve Fisher (RAL)

Description

<html> <head> <title>Minutes.html</title> </head> <body>

Alastair should assist Dave K with the configuration
APEL Core etc are fine - but the APEL publisher is failing with some ETICS feature. Alastair will follow up.
Building and testing new code
Testing is underway; quite a lot of failures some of which are caused by the removal of support for the deprecated calls.
Registry able to forward to old registry
There will be a delay in seeing new producers as it will wait for the consumer to poll the registry. Will probably need to make it register the consumer otherwise short-lived producers may not be seen.
Pursue weekly failures of RGMA17
Tagged a fix made by Kush on the leaking file descriptors. Will also look at trying the reconnect option. Tried the reconnect but the flexi-archiver script could not cope with the extra "=" sign.
Investigate the configuration parameters for MySQL and for Tomcat
Tried on RGMA17 but it does not work because the parameters are for the wrong JVM
RGMA_HOME value is built into the RPM. Should move to CATALINA_OPTS
GIN patch #1093
In configuration. The EMT did not think this important enough to certify at the moment.
Patch for #26344 , #26345 and CLI
This remains at relatively low priority until we are happy with the registry
#26683 spec files will not work under SuSE
Eamon has a fix as the spec file rules are more strict - Alastiair will propose what to do with the "big"

#26527 rgma-glue-archiver unstable since update 29
This is seen as the problem with the registry locking. There are also a coupel of calls from the registrar that are still trying to clean up old resouces and there is the locking which is too coarse. These registry fixes will be tried - and if that fails ask major sites to turn off job wrapper publishing
Update the inspector to work with new and old consumers
Currently it assumes separate war files. Will probably need to determine version of a service before addressing it further. I
Test that new streaming really works - given that we have no new producers
Tests may exist but need to check. Started work on new PP.
Send DN and VOMS attributes in an open ended way to the prodcuer in the start call. The spec will need updating to match later.
Need to find out how to extract the info from the VOMS extensions - should ask Stuart. However it may not be needed - wait to hear from Stuart

Write schema system tests
There is a problem with the authz rules as it uses a list<string> bit in the DB; it may make sense to change the DB to make authz rules a table. Ming has some concerns about time stamps at start up - do we need to replicate the master table?
#19645 The Sub exception org.edg.info.ReadTimedOutException is an inappropriate type

Convert tests to python
Basic new framework exists - need to see what is going wrong with remote calls and how to handle the dummies

Logging document with Ales
Good agreement with Ales - but there may be some problems with the security demands. Some more discussions with the notion of event type. Ales now has "the pen"
New SD spec
Review REMIND bugs
Work has started
Add new new call to list available VDBs to spec
Add to the spec that both client and mon box names should be a stringified numeric IP addresses
Add a getTableTimestamp(vdb, tableName) for the schema in design and specification
All schema operations that change state return time stamp rather than OK - the spec needs updating

</body> </html>

The agenda of this meeting is empty