1 / 19

Tier1 Grid from users point of view: urge of standards

This article discusses the importance of implementing standards in the grid system to provide an operational and reliable environment for PhD students, researchers, and others. It highlights the challenges faced by users and suggests the need for a more efficient and streamlined process. Examples of operational problems and solutions are also presented.

akenna
Download Presentation

Tier1 Grid from users point of view: urge of standards

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Tier1 Grid from users point of view: urge of standards Dr James Cunha Werner Babar UK Grid Meeting

  2. Users Requirements • PhD students with 3 years scholarship. • Researchers with fixed-term contract. • Researchers with deadlines and competition. THEY NEED AN OPERATIONAL AND RELIABLE ENVIRONMENT TO DO THEIR WORK.

  3. The service provide by RAL for Babar Grid UK • Months to install LCG properly. • Months to develop an initialisation script. • Lack of adequate procedures  Poor service. USERS LOOKING FOR OTHER RESOURCES: SLAC, GRIDKA, ETC User’s waste of time. Idle resources.

  4. Grid at Babar Elba meeting

  5. TauUsers reprocessing: opportunity lost!

  6. Jenny’s request • Date: Mon, 4 Apr 2005 12:58:37 +0100 (BST) • From: Jenny Williams <jenny@hep.man.ac.uk> • To: James Werner <jamwer@hep.man.ac.uk> • Subject: TauUser for CM2 • ok, it works. • Requirements: • for running with analysis-24: • Beta V00-12-03 • BetaMiniUser V00-03-00 • BetaPid V00-04-10-05 • …

  7. Date: Mon, 4 Apr 2005 10:58:11 +0100 • From: Steve Traylen <s.traylen@rl.ac.uk> • To: jamwer <James.Werner@manchester.ac.uk> • Cc: babargrid-uk@lists.man.ac.uk, Chris Brew <c.a.j.brew@rl.ac.uk> • Subject: Re: [BABARGRID-UK] Jobs in Waiting forever... • On Mon, Apr 04, 2005 at 10:11:30AM +0100 or thereabouts, jamwer wrote: • > Dear colleagues, • > Last week I submitted one dataset (26 jobs) to bohr0001.... and the jobs • > were waiting for 4 days. I killed all of them and submitted again in my • > farm bfb... and they still waiting. • > Submission was fine: • > • > JOB SUBMIT OUTCOME • > The job has been successfully submitted to the Network Server. • > Use edg-job-status command to check job current status. Your job • > identifier (edg_jobId) is: • > • > - https://lcgrb01.gridpp.rl.ac.uk:9000/hXbthIXfJCACQeOh-na3_w • Chris, James • I should add , it is only lcgrb01.gridpp.rl.ac.uk that appears to have • this problem. There are not reports from other RBs of them going into • this state. • I'll keep you updated as I get news. • Looking for other RBs that support babar there is also • grid008g.cnaf.infn.it • egee-rb-01.cnaf.infn.it • It would be good to break there RB as well. CNAF has the expertise locally • to fix this kind of thing. • Steve Operational problems At RAL

  8. RAL operational again • Date: Fri, 6 May 2005 09:25:58 +0100 • From: Steve Traylen <s.traylen@rl.ac.uk> • To: Babar Grid UK <babargrid-uk@lists.man.ac.uk> • Cc: James Werner <james.werner@manchester.ac.uk> • Subject: lcgrb01 looks to be okay now. • Hi James and others. • lcgrb01.gridpp.rl.ac.uk the RB at RAL that was having problems • now looks to be okay. It was okay before I went away two weeks • ago and still appears to be. • The fault looked to be a bad a interaction between globus and • nscd. • Please feel free to use lcgrb01 and as normal post questions to • lcg-support@gridpp.rl.ac.uk

  9. Initialisation script From : <jamwer2000@hotmail.com> Sent : 17 February 2005 09:00:07 To : BaBarGrid-hn@slac.stanford.edu Subject : Re: VO-based environment settings Dear Artem, Your question is very important if we want to establish a worldwide grid. LCG grid software defines envvar VO_BABAR_SW_DIR to point the configuration directory, where initialisation scripts, tars etc are stored. At Manchester we defined the script $VO_BABAR_SW_DIR/babar-grid-setup-env.sh to initialise $BFROOT, $BFARCH, ... and call all scripts from hepix (group_siteSpecs.conf.sh, group_aliases.sh, group_sys.conf.sh, and bashrc). If you do not have the release installed, them a tar should be untared following http://babar-hn.slac.stanford.edu:5090/HyperNewws/get/BabarGrid/322.html to provide the necessary infrastructure. We do not use this, because our babar software is installed at AFS. The next step is set 00_FD_BOOT to your last version of condition and configuration database. At this point, you will be able to run BetaMiniApp without any problem, in any computer in the world with follow this elementary standard. I am running Tau11 in parallel in 26 computers from different farms, which allow me analyse more tham 1 million events per hour. For more information, see http://www.hep.man.ac.uk/u/jamwer/ Best regards, James

  10. From : <C.A.J.Brew@rl.ac.uk> Sent : 17 February 2005 09:41:40 To : BaBarGrid-hn@slac.stanford.edu Subject : RE: VO-based environment settings Hi, As someone who sits on both sides of this fence (site admin and grid application developer/user) James's solution is, I think, the only practical one and the one I've been pushing. …

  11. Date: Mon, 9 May 2005 10:59:34 +0100 (BST)From: jamwer <James.Werner@manchester.ac.uk>To: Hep-grid@lists.man.ac.uk, babargrid-uk@lists.man.ac.ukSubject: [BABARGRID-UK] Grid needs standardsWould you please write a script for analysis-24, called. $VO_BABAR_SW_DIR/babar-grid-setup-env-analysis-24.shwhich initialise all babar environment and 00_FD_BOOT.The commands users have to run after run your script will be:local=`pwd`cd /afs/rl.ac.uk/bfactory/dist/releases/analysis-24srtpath analysis-24 $BFARCHcd $localln -s $BFROOT/dist/releases/analysis-24 PARENTedg-rm --vo babar cp lfn:jamwer_bfb.tier2.hep.man.ac.uk_BetaMiniApp_16file:///tmp/BetaMiniAppchmod 777 /tmp/BetaMiniApp/tmp/BetaMiniApp JobTau11-Run4-OnPeak-R14-1.tclrm /tmp/BetaMiniAppI am trying to run using the same parameters I had in the batch system andit is not working.We need a standard way to initialise the environment,if we want to allow users in grid in any site.Let me know when you have the job done, or if you have a best way to doit.Best regards,James

  12. Date: Tue, 10 May 2005 13:51:59 +0100To: jamwer <James.Werner@manchester.ac.uk>Cc: babargrid-uk@lists.man.ac.ukSubject: RE: [BABARGRID-UK] Grid needs standardsHi James,I've not dealt with this because I'm away at the HEPiX Workshop at the moment and this will need some dicussion before it's implemented. The script you suggest is very highly taylored to your specific needs and will have to very much more generalised before it can go into use.Also as you say in the subject line "Grid needs standards" but thosestandards need to be agreed and useful for many people.I suggest you report this as a suggestion to the main BaBarGrid listwhere we can discuss it and find a general solution which will work for more situations than just yours.…

  13. Publishing site resources/releases • > GlueHEPSup= Babar, Atlas, ... <= different softwares • > GlueOS= RH7.2, RH7.3 or SL3 ... <= Operating System • > GlueAplic= BetaMiniApp, Moose, ... <= Available Application • > GlueReleases= 14.5.2, 14.5.2d, 16.0.1 etc <= Releases available • > GlueCondDB= local, AMS, xrootd, ... <= Cond & Config DB • > GlueBackgroundDB= local, AMS, xroot, ... <= Background DB • > GlueBbk= local, xrootd, ... <= Experimental Data • We would be able to seach the configuration we want to run the software • and optimise resources. I am able to know how many jobs are in queue, and • what would be the best strategy. • If a massive software (taking days) we can use data remotely • through xrootd: them GlueBbk=xrootd would be used. If a program test use • GlueBbk=local, and only a few sites would be able to run it. • A consulta fornecera a lista com o nome dos CE com o release disponivel.

More Related