Print

Print


Matt,

Although we updated with YAIM, we had the symptoms you describe. The 
LDIFs contained the lines, but they didn't show up in the LDAP output 
from our site BDII. In our case, we had to update the BDII to version 
glite-BDII-3.2.9-0.sl5 (it had been 3.2.3-0). After that, the values 
showed up.  It's just something to try. Use this command on your BDII to 
check whether you need to do this:

grep gluececapability /opt/bdii/bin/bdii-update

If things are fine, you should see  the "gluececapability" line.

Steve


Matt Doidge wrote:
> Thanks to everyone for clearing things up for me. I've been trying to 
> slot the VO Share information into our static ldif's manually. It 
> worked for our CREAM CE, but when I try to slide the line 
> "GlueCECapability: Share=atlas:91" into our lcg-CE's 
> static-file-CE.ldif it doesn't show up  at all (even after a bdii 
> restart, at both the site and resource level). Following advice given 
> to Winnie previously I've hunted for rogue whitespace in the ldif but 
> didn't find any, and the CE publishing is otherwise okay. The bdii.log 
> doesn't seem to have any interesting entries in it either. I've also 
> experimented with altering currently entries within the 
> static-file-CE.ldif to see if all the static entries are "frozen in 
> time", but any changes I made trickled through straight away.
>
> Does anyone else have any other suggestions that I may try? I'm out of 
> ideas on what I thought would be a simple thing to impliment.
>
> Thanks in advance,
> Matt
>
>
>
> John Gordon wrote:
>> The cpus and cores=0 for a VO is only in gstat. It is not used in 
>> scheduling, only for management reporting.
>>> -----Original Message-----
>>> From: Testbed Support for GridPP member institutes [mailto:TB-
>>> [log in to unmask]] On Behalf Of Matt Doidge
>>> Sent: 11 February 2011 15:48
>>> To: [log in to unmask]
>>> Subject: Re: Publishing VO shares in gstat
>>>
>>> Hello,
>>>> If you don't publish SHARE for a VO then the #cpus and cores will
>>> appear as 0 for that VO.
>>> How does a 0 for #cpus affect a VO? Will VOs (if they are paying any
>>> attention to the information system), read this as "no resources for
>>> you
>>> here" or as "opportunistic use only for your VO at this site" (seeing
>>> as
>>> you are bothering to publish something about that VO at your site you
>>> assume you provide some support for it).
>>>
>>> Lancaster is predominantly an atlas site, so giving atlas a share of
>>> 93%
>>> (from the hardware numbers) seems about right. But we also support 17
>>> other VOs on one of our CEs. If we gave each of these a minimum share
>>> of
>>> 1... well I think you can see where I'm going.
>>>
>>> I think the crux of my query is do I just need to explicitly write out
>>> the share for the "major players" at my site?
>>>
>>> Cheers, and have a good weekend all,
>>> Matt
>>>
>>>
>>>> I've asked Laurence to produce a page which explicitly shows the
>>> shares for each site.
>>>> -----Original Message-----
>>>> From: Testbed Support for GridPP member institutes [mailto:TB-
>>> [log in to unmask]] On Behalf Of Winnie Lacesso
>>>> Sent: 11 February 2011 14:28
>>>> To: [log in to unmask]
>>>> Subject: Re: Publishing VO shares in gstat
>>>>
>>>> Happy Friday!
>>>>
>>>>> From:    Jeremy Coles <[log in to unmask]>
>>>>> Subject: Publishing VO shares in gstat & Jan availability figures
>>>>>
>>>>> I have just taken a snapshot (http://tinyurl.com/5tmecj3 - the ALL
>>> VOs
>>>>> column is taken from the tool. The total LHC column is simply the
>>> sum
>>>>> over the LHC VO columns) and this suggests the following sites are
>>> not
>>>>> yet publishing anything:
>>>> <snip>
>>>>> Bristol
>>>> Bristol was in scheduled MAINT on 10 Feb, all nodes down most of day.
>>>> Up now & all well.
>>>>
>>>>> If you want to look yourself then use this link:
>>>>> http://gstat-wlcg.cern.ch/apps/capacities/sites/
>>>>> Select the VO to check (top left) and then scroll to your site (or
>>> use
>>>>> search). The total figure
>>>> I'm so sorry, Doctor, but what are we to see on that page w.r.t. VO
>>>> shares?
>>>> And what's missing after "The total figure" ?
>>>>
>>>>> If you believe that your site is publishing yet appearing in this
>>> list
>>>>> please could you contact me.
>>>> Done.


-- 
Steve Jones                             [log in to unmask]
System Administrator                    office: 220
High Energy Physics Division            tel (int): 42334
Oliver Lodge Laboratory                 tel (ext): +44 (0)151 794 2334
University of Liverpool                 http://www.liv.ac.uk/physics/hep/