Hi Luke,
Looking at Gstat the subclusters for lcgce01 are not being shown, but for lcgce02 they do appear. I think the problem is that your subclusters for each CE have the same name, i.e. both lcgce01 and lcgce02 have subclusters called gridAMD and gridIntel, and I believe that the names of subclusters at a site have to be unique.
Regards,
Andrew.
________________________________________
From: Andrew Lahiff [[log in to unmask]]
Sent: Saturday, May 31, 2014 9:59 AM
To: [log in to unmask]
Subject: Re: Debugging 'BrokerHelper: no compatible resources'.
I eventually found that if you run the secret command:
echo "[ command = \"ism_dump\"; ]"> /var/workload_manager/jobdir/new/req0
the file /var/workload_manager/ismdump.fl is generated containing all the CEs known by the WMS.
lcgce02 doesn't appear there at all. I think Catalin is back on Monday so I'll ask him if there's a way to force the WMS to refresh it's cache. Hopefully that will help.
Regards,
Andrew.
________________________________________
From: Stephen Burke [[log in to unmask]]
Sent: Friday, May 30, 2014 3:20 PM
To: [log in to unmask]
Subject: Re: Debugging 'BrokerHelper: no compatible resources'.
Andrew Lahiff [mailto:[log in to unmask]] said:
> I'm not actually sure how to do that. Last year I found a way
> of dumping all the info a WMS has about all CEs, but I'm
> still trying to find out again what this was...
ISTR it used to be in a file called ismdump.fl, I'm not sure if it still is.
Stephen
--
Scanned by iCritical.
--
Scanned by iCritical.
--
Scanned by iCritical.
|