[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: setup for MDC2





Dave Zimmerman wrote:
> 
> Arie Shoshani <shoshani@lbl.gov> wrote:
> >Dave,
> >
> >Three issues:
> >
> >1) We discussed the HPSS setup again, and my last conclusions still
> >holds. i.e.  0% in cache (or as closed to it as possible) before
> >emptying, and 10 minute before purging.
> >
> I have requested grace tsai to lower the cache level as far as possible.
> I believe this is 1%.  We will bring this point up again in the Friday
> morning meeting with John Riordan and Grace Tsai.

Also, the 10 minute setup must be made or the cache will not empty in 10
minutes.


> >2) To run more sophisticated and better controlled queries, we'll need
> >to know: a) which files are on which tapes; b) which conponents are on
> >which files.
> >
> At present I have listings of all GC files for each of the tapes that they
> occupy.  Unfortunately, due to glitches in the data transfer there is not a
> nice mapping between the files and the tapes.  That is, the files for a
> given event are likely to be distributed over more than 1 tape, but not in
> a manner which follows any particular rule.

Then, the only way to do experiments of single tape drive vs. multiple
tape drives will be to change the HPSS setup on how many drives are made
available.

 
> The mapping of file to components is done in the tag (by order) and the
> filename (by string) does this meet your criteria?

It will be helpful to know which files contain which component.  We can
better plan multi-component experiments.  Is that available?


> I would like to include the tape numbers in the tag for each event but am
> not sure that I will be able to provide that by Friday morning.  Certainly
> the file listing will be available.

Yes, we'll need this at a minimum.


> >3)  When exactly will the System available to us to start running
> >tests?  Should I contact somebody else to get an answer to that?
> >
> 
> I will speak to Yuri about what time the STAR reconstruction will stop on
> Friday morning.   Doug and I plan on meeting John and Grace Friday morning
> at 10:00.  At that meeting we will get more information about the
> configuration of the HPSS.

You mean the system will not be available till after your meeting with
John and Grace?  i.e. the hpss setup will only be done after you meet
with them?

Arie.