Hi,
We have noticed problems, mainly at large sites with the stability of
the site GIIS. The best way to see if your site has this problem is by
looking at the graphs in the gstat too.
http://goc.grid.sinica.edu.tw/gstat/
For example compare the graph of CERN and RAL
http://goc.grid.sinica.edu.tw/gstat/lxn1181.cern.ch/
http://goc.grid.sinica.edu.tw/gstat/lcgce02.gridpp.rl.ac.uk/
Notice that there are many down spikes on the graph for RAL but CERN is
flat.
If your site shows this problem it would be a good idea to use the BDII
as a site GIIS. It was done at CERN and this is why the graph is flat.
How to do this is described in the AdministrationFaq.
http://goc.grid.sinica.edu.tw/gocwiki/AdministrationFaq#had-6aac0335e47055de2e011d2b9fd3df26444f1fe5
The BDII is run in parallel with the site GIIS so it should be transparent.
Laurence
http://goc.grid.sinica.edu.tw/gstat/
On Thu, Aug 12, 2004 at 10:02:30AM +0200 or thereabouts, Laurence wrote:
> Hi Steve,
>
> We had problems at CERN with the LSF CE causing problems with the
> information produced from CERN. It was the usually information provider
> being slow and affecting the GRIS/GIIS. To fix this we replaced the site
> GIIS with a BDII. Compare the graphs of CERN and RAL.
>
> CERN http://goc.grid.sinica.edu.tw/gstat/lxn1181.cern.ch/
>
> RAL http://goc.grid.sinica.edu.tw/gstat/lcgce02.gridpp.rl.ac.uk/
>
> Notice how flat the CERN graph is and all the down spikes in the RAL
> graph. You might also want to consider replacing your site GIIS with a
> BDII. It is very easy to do. The steps can be found here.
Morning Laurence,
It is certainly something I plan to do, it is not urgent since in reality
it is working enough of the time for the farm to fill up. When I get back.
Steve
>
> http://goc.grid.sinica.edu.tw/gocwiki/AdministrationFaq#had-6aac0335e47055de2e011d2b9fd3df26444f1fe5
>
>
>
> Laurence
--
Steve Traylen
[log in to unmask]
http://www.gridpp.ac.uk/
|