Date: Mon, 18 Jun 2007 14:55:11 +0200 From: Maria Dimou-Zacharova To: ggus-info CC: it-dep-gd-ops-us@cern.ch, Ian Bird Subject: Re: stockholm Dear All, FYI impressions from the Grid Operations' Workshop sessions I attended, i.e. Wednesday 13/6/07 after 16:30hrs and the whole of Thursday 14/6. Full agenda in http://indico.cern.ch/conferenceTimeTable.py?confId=12807 They only relate to various GGUS topics in view of the ESC meeting http://indico.cern.ch/conferenceDisplay.py?confId=7187 1. Many speakers emphasised the need for up-to-date documentation. The most precise reports were from Alessandra in http://indico.cern.ch/materialDisplay.py?contribId=25&materialId=slides&confId=12807 following which I contacted Joachim with suggestions concerning the glite doc. on which he is already working. I 'll see now what else derives specifically from this talk. 2. Site admins expressed distress due to lack of config. information coming with releases. As I know from TPM monitoring and the VOM(R)S WG, a big number of tickets are assigned to the "Installation & Configuration" Support Unit (SU) and info about new site-info.def, vomses, gridmap.conf and other such files takes months to propagate. The good news are that all VO-related config. info should, in the future, be taken from the cic portal (https://cic.gridops.org/index.php?section=vo&page=idcardupdate) by the sites. As I am particularly interested in this, I'll keep you all informed as the VO-IDs get enhanced. 3. The related issue of GGUS FAQs for VOs (please see appendix for reminder) was not discussed in Stockholm as planned, at least while I was there. There simply was no time. Thursday lunch break was 45 minutes and everyone had to find a restaurant in the neighbourhood, so we didn't manage to group and hold a meeting. We should discuss it at the ESC this week. It is important for me that we do this because I approached some VOs for updates and then told them to put the issue on hold. 4. ROCs complained about some VOs often bypassing the grid 'hierarchy' and claiming bad service by the sites to national and/or academic authorities, without reasonable justification. This comment was 'inspired' by the last slide "Challenges" in my presentation: http://indico.cern.ch/materialDisplay.py?contribId=20&materialId=slides&confId=12807 The lesson, as I understand it, from this is that, we, supporters can help avoiding conflict via our ticket monitoring and VO Support activities. This will also encourage VOs and sites to use GGUS persistent ticket URLs for such follow-up activities. 5. Ian(B) said that there is a perception that questions emailed to the rollout mailing list are answered correctly and instantly but GGUS tickets take too long to be solved. Steve(T) observed the visibility one gets via a 'Reply-All' in email, which doesn't exist when one does a good job in a ticketing system. Therefore I shall put now in the shopping list the suggestion to publish on the GGUS home page the "TPM of the month" and the "SU of the month" with the aim to get an award at each Operations' Workshop. I copy Ian for approval of whatever the award could be. When I mentioned this idea in my presentation, he seemed favourable to it. I wouldn't hesitate, as I said in my presentation, to turn the whole rollout into a GGUS SU for some days just to get users and supporters familiar with our facilities (but not without the agreement of all parties involved). Yours - maria -------------- APPENDIX: GGUS FAQs for VOs --------------- Date: Tue, 29 May 2007 13:11:55 +0200 From: Maria Dimou-Zacharova To: project-eu-egee-sa1-esc@cern.ch CC: it-dep-gd-ops-us@cern.ch, cms-grid-support@cern.ch, atf , alice-lcg-task-force@cern.ch, lhcb-grid-support@cern.ch, project-lcg-vo-geant4-admin@cern.ch, na48-vo-admin@cern.ch, egee-biomed-vo-manager@healthgrid.org, frederic.schaer@cea.fr, Gilles Mathieu Subject: Re: GGUS FAQ review for your VO Dear All, many thanks for studying these FAQ documents for your VO. Please put the content review on hold for a few days. Reason: The May 24ht ESC http://indico.cern.ch/conferenceDisplay.py?confId=7186 discussed the issue and decided to *review the template of this type of FAQs in the coming Operations Workshop in Stockholm* (June 14th). Input for that discussion: 1. Today the VO-ID card at the CIC operations' portal: https://cic.gridops.org/index.php?section=vo&page=idcardupdate contains some of this information. We should move ggus-support-related data i.e. yes(email)/no in the VO-ID card for each VO to avoid maintaining the information in 2 places. 2. GGUS-specific Questions and Answers (e.g. Items 4604, 4607, 4608, 4609 etc in the http://egee-docs.web.cern.ch/egee-docs/support/documentation/doc/4600_FAQ_for_cms_05.doc example) should appear in a common, ggus-maintained "FAQ for VOs" document, as this type of information doesn't depend on the expert supporters within the VO, anyway. Comments are welcome with many thanks and regards maria PS some asked "who is supposed to be informed by these pages". The answer is: *supporters* in other GGUS Support Units, most importantly the TPMS, i.e. ticket dispatchers. Maria Dimou-Zacharova wrote: > Dear all, > > could you please click on the document concerning your VO in the > appended list, apply changes, make an update the document change log to > record it was recently checked it and send it back to us by simply > attaching the new version in your reply to this message? > > http://egee-docs.web.cern.ch/egee-docs/support/documentation/doc/4100_FAQ_for_alice_06.doc > > > http://egee-docs.web.cern.ch/egee-docs/support/documentation/doc/4200_FAQ_for_atlas_04.doc > > > http://egee-docs.web.cern.ch/egee-docs/support/documentation/doc/4600_FAQ_for_cms_05.doc > > > http://egee-docs.web.cern.ch/egee-docs/support/documentation/doc/4900_FAQ_for_lhcb_06.doc > > > http://egee-docs.web.cern.ch/egee-docs/support/documentation/doc/6600_FAQ_for_na48_01.doc > > > http://egee-docs.web.cern.ch/egee-docs/support/documentation/doc/6800_FAQ_for_geant_01.doc > > > http://egee-docs.web.cern.ch/egee-docs/support/documentation/doc/4400_FAQ_for_biomed_02.doc > > > PDF versions of these FAQ pages are linked from > GGUS_home --> Support Staff --> Info about "Responsible Units" > so they should be up-to-date. > > As we try to better organise the GGUS documentation all together, > if you find some things are not useful, functional, well linked > please let us know. > > Thanks very much in advance > maria and the rest of Grid Deployment Support Team ============= End of APPENDIX =========================================== Notes by Torsten Antoni (GGUS developer) on the Grid Operations' Workshop ------------------------------------------------------------------------- here our comments about the meetings in Stockholm. COD: * Failover: Oracle streaming is not an ideal solution for replication; after each update of data the whole database is replicated. COD now using/trying to use DataGard: Tool the manages replication of updated data !!!!!! SAM: problems getting more Oracle licenses; currently only one Oracle license for one CPU at Cyfronet -> see slides from Alfredo Pagano and Alessandro Cavalli * Downtime reporting: -> see slides from Osman approach: to have only one place were the downtime has to be announced GGUS downtime also on GOC DB? * SAM - ENOC: Integrate ENOC tests into SAM? -> if several sites are down due to network problems it does not make sense to start the SAM tests ARM: * Change request procedure: GGUS procedure most advanced and well received by ROC managers. Here other core services can learn from us. OPS: Maria gave a very good and clear presentation on GGUS and also summarized the GGUS relevant topics nicely in her email from earlier today. Especially Points 4 and 5 from her email came up several times during the Ops meeting and are of course closely linked. I agree that we have to do something to better advertise GGUS and also to honor the work done by supporters and make it more public. I'm not fully convinced that an "employee of the month"-scheme would work, but it could be worth a try. Concerning the Rollout-List I don't see an easy mechanism to link this to a ticketing system, since these are two complementary ways of doing support. But what could be done is to include the rollout archive in the database for the GGUS semantic search. Summarising I would say that, even though it is a slow process, the reception of GGUS at these meeting is improving. Let's work on better marketing. ----------------------------------------------------------------------- Dr. Torsten Antoni torsten.antoni@iwr.fzk.de