 |
 |
Hi Mario, It very much depends on the site’s configuration. The MON box (and the glite-APEL) has the database where records are written. If you are using two databases you need to make sure that they are both receiving all the records that APEL needs (from batch system log files, from the accounting log files, from the site’s GRIS or GIIS). If every CE contains all these log files, then yes, you can have one of them publishing via the MON and the rest on glite-APEL (or viceversa). However, if for instance the batch system is installed in a different machine, then it gets a bit tricky, because the parser in that machine can only write to one db, and the other db will be missing this information. As I said, depends on your site’s configuration and it can very easily go wrong… Hope this helps, Cristina -------------------------------------------------------------------------------- Cristina del Cano Novales Scientific Computing Technology Group / e-Science Science and Technology Facilities Council Rutherford Appleton Laboratory Harwell Science and Innovation Campus OX11 0QX Didcot Tel: +44 (0) 01235 445778 [log in to unmask] <mailto:[log in to unmask]> From: LHC Computer Grid - Rollout [mailto:[log in to unmask]] On Behalf Of Mario David Sent: 02 July 2010 10:14 To: [log in to unmask] Subject: Re: [LCG-ROLLOUT] gLite 3.2 :: MON (???) hi since the mon or apel nodes collect records from CE's, if you have more than one CE you can put one of them publishing the records to the apel node, and see how that goes, am I correct? cheers Mario On Jul 2, 2010, at 9:02 AM, Cristina DelCanoNovales wrote: HI Santanu, You can have both nodes specified in GOCDB, that’s not a problem. However, having both the new glite-APEL and the old MON running at the same time is a bit more tricky. Are you planning to have both publishing records for a while? Cheers, Cristina -------------------------------------------------------------------------------- Cristina del Cano Novales Scientific Computing Technology Group / e-Science Science and Technology Facilities Council Rutherford Appleton Laboratory Harwell Science and Innovation Campus OX11 0QX Didcot Tel: +44 (0) 01235 445778 [log in to unmask] From: LHC Computer Grid - Rollout [mailto:[log in to unmask]] On Behalf Of Santanu Das Sent: 01 July 2010 11:26 To: [log in to unmask] Subject: Re: [LCG-ROLLOUT] gLite 3.2 :: MON (???) Greetings all, Just trying to confirm thisg with others: Install glite-APEL (v3.2.2-0) couple of days ago, without having any errors. As far as I understand, MON is not needed anymore for this release, so it's just the glite-APEL needs to be installed on that box. /opt/glite/yaim/bin/yaim -c -s site-info.def -n glite-APEL Another question: Is there any way to test the new node without putting in the GOCDB or production? Or, even better, is it possible to put the new glite-APEL node in the GOCDB along with the still-running old SL3 MON-box, for testing? Cheers, Santanu Hi Alessandra, We've got four sites beta testing the new transport mechanism right now and all have been successful. Two of these sites have been moved to the beta after problems sending records via R-GMA. As noted in the other messages we've also recently certified the latest version of the publisher that uses the new transport mechanism. Kindest Regards -----Original Message----- From: LHC Computer Grid - Rollout [mailto:[log in to unmask]] On Behalf Of Alessandra Forti Sent: 19 May 2010 14:10 To: [log in to unmask] Subject: Re: [LCG-ROLLOUT] gLite 3.2 :: MON (???) Hi Steve, We have high hopes for the change in transport layer that is being coded to make this go away. is there any time scale for this? It has been known for a while now that RGMA was going to be phased out. cheers alessandra Steve Traylen wrote: On Thu, May 13, 2010 at 12:00 AM, Jeff Templon <[log in to unmask]> <mailto:[log in to unmask]> wrote: Hi we've had problems getting all our accounting records in for two months now. the solution is to republish frequently and often, I am not sure whether RGMA is dropping the records, or is it somehow not making it from RGMA into the DB. perhaps we could save lots of effort and just email the account records once per week or so? kind of silly to keep sacrificing a machine just to support the deprecated transport mechanism, which often just doesn't work. This has been my recent proposal for CERN as well. Even sending our back logs took long enough to process that we built up a back log again big enough that it could not be sent. => doomed. JT On 12 May 2010, at 22:38, Cristi Magherusan wrote: On Mon, 2010-05-10 at 09:40 +0100, Stephen Burke wrote: LHC Computer Grid - Rollout [mailto:[log in to unmask]] On Behalf Of Cristi Magherusan said: Is MON still needed for APEL? Yes, at the moment. This is the patch for the R-GMA-free APEL, if you want to watch the progress. https://savannah.cern.ch/patch/index.php?4014 Stephen Thanks Stephen, It seems this patch got stalled for almost a month now, so I wonder what's blocking this from getting into production. @Will: what do you think about this patch? Thanks, Cristi -- Cristi Măgherușan, alumnus System/Network Engineer Technical University of Cluj-Napoca, Romania http://cc.utcluj.ro +40264 401247