perfsonar development work

["Michael Bischoff" <>]

I'm pretty sure that was traced back to a 'magically slow bullet'(as it was 
called) with the
gls when sending a pretty empty discovery message, where as sending a xquery 
message with the
same scematics/payload returned much, much quicker. (possibly indicating a 
bug) Jason said he
was going to investigate.

This was all with the ps-gls as there was no java/mdm variant available. It's 
also because of
this that we couldn't 100% verify that it wasn't due to the LS client api.

About other problems, there is still the issue with retrieving measurement 
metadata - for
some services it's necessary to know on beforehand what is being measured by 
the MA in order
to formulate a perfsonar message. (For example an MA monitors a couple of 
routers and the
request should contain some or all these. Now this data is distributed 
through the LS (as
measurement metadata) but is not easily retrievable right now through the 
api. Some work was
started on this but required more research).

There are a couple of other things but those have implications LS 
architecture wide.

kind regards,

Michael

> I'm moving this discussion to perfsonar-dev list as it involves our
> partners from I2 too.
>
> Krzysztof Jêdrzejewski pisze:
>
>> Hi,
>>
>>
>> It's my first mail here, so I'll introduce myself a little. I'm
>> Krzysztof Jedrzejewski from PSNC and I'm developing Lookup Service API.
>> On the mailing list has been mentioned the problem that may be related
>> to LS API ( 
>> https://mail.internet2.edu/wws/arc/perfsonar-dev/2009-02/msg00124.html
>> https://mail.internet2.edu/wws/arc/perfsonar-dev/2009-02/msg00129.html
>> ). Can anyone say if that problem was solved? If wasn't can anyone
>> describe this problem more precisely? Please inform me also about other 
>> problems with LS API
>> (via bugzilla or
>> e-mail).
>
>
> --
> Szymon Trocha
>
>
> Poznan Supercomputing & Netw. Center ::: NETWORK OPERATION CENTER
> Tel. +48 618582022 ::: http://noc.man.poznan.pl
>
>
>