Dear All,
Please find attached the GridPP Project Management Board
Meeting minutes for the 551st and 556th meeting.
The latest minutes can be in:
http://www.gridpp.ac.uk/php/pmb/minutes.php?latest
as well as being listed with other minutes at:
http://www.gridpp.ac.uk/php/pmb/minutes.php
Cheers, Dave.
GridPP PMB Meeting 556 (02.03.15)
=================================
Present: Dave Britton (Chair), Andrew Sansum, Dave Colling, Jeremy Coles, Tony Doyle, Steve Lloyd, Dave Kelsey, Claire Devereux, Pete Gronbech, Tony Cass, Roger Jones (Minutes Ð Suzanne Scott)
Apologies: Pete Clarke
1. Collaboration Board (CB) meeting
====================================
In preparation for this meeting, which was due to take place this afternoon at 4pm, emails had been exchanged and various views expressed. A presentation had been circulated by DB; he also suggested an overview of the hardware resources.
DK commented that sites might wish to see the hardware in relation to meeting the Service Level Agreements (SLAs). DB advised that the experiments had to take the lead with this information, particularly ATLAS and CMS. It may not be possible to determine the exact hardware requirements at such short notice. DC advised of issues at Bristol in relation to hardware allocation. DB understood the Bristol case with reference to CMS but noted that there was a broader picture. RJ considered that ATLAS wanted sites primarily to be well-run and not to be a drain on resources. Some sites who are treated as a Tier-2 do not perform as Tier-2s but rather as Tier-3s. DK noted that in the past, the hardware funding model was based on prior delivery of resources, if we moved away from that, what would the new model be? RJ reminded that sites had to meet an appropriate level of service.
DB advised that it was in GridPP's interests to support sites that didn't have effort, or had little/minimum effort, in relation to engagement and VO support. 10% of hardware was 'unassigned' and could be 'distributed', which was an important aspect of engagement and leverage. There ensued a discussion on the status of smaller sites.
It was agreed that a 'holding' statement of intent, with regard to hardware, would have to suffice for now, as further details at the moment were impossible to ascertain. DB would add a slide to the CB presentation.
PG reminded that the GridPP4+ hardware funding would have to be allocated soon.
DB noted that if consensus is achieved at the CB meeting, he would return to writing the GridPP5 Proposal, based on the CB presentation and would frame his arguments around the key objectives. The CB knew that they must focus on the 'big picture' of the objective and strategy of the project as a whole and that it was always incumbent upon them to take this broad view. The valuable resource we have is the expertise of staff, and this is mainly in Deployment, Operations, and Support. We are trying to present a complex strategy in the best practical way possible. DB noted we could optimise across categories, however the structure needed to be self-consistent.
2. Oversight Committee (OC) Documents
======================================
PG noted there were only two documents required for this, the Project Status document and the Financial Status document. He had circulated the latest versions of these, which he considered to be the finalised versions. The deadline for submission was 4th March. DB wished to read them over prior to submission, and would do this as soon as he could.
3. DIRAC
=========
AS had replied to the DIRAC enquiry regarding tape media. It was considered good for GridPP to be engaged with DIRAC Ð we did have tape media and we might be able to commit some effort as well in order to assist them. The hardware cost was over £100k and they could potentially have 16-17% share of the Tier-1. We may be able to commit 0.5FTE in 2018-19. This assistance was more or less agreed now, the next stage was to provide DIRAC with the relevant contacts.
4. AOB
=======
PG noted that regarding the JeS forms for the GridPP5 Proposal, we need to get these in by 10th March, so we should get the process rolling now. Figures were needed for the Institutes. DB advised that if the CB agreed the GridPP5 strategy this afternoon, then we could issue the JeS call tomorrow. If they did not agree, then this would not be possible to do.
STANDING ITEMS
==============
SI-0 Monthly Report from Development (Cloud) Group
---------------------------------------------------
There was no report.
SI-1 Dissemination Report
--------------------------
SL reported on behalf of Tom Whyntie as follows:
### New User Engagement Campaign - Update
* CERN@school student data on the grid: thanks to the efforts of the ICL DIRAC team the LUCID student team is able to upload datasets to the DIRAC DFC and publish via http at Liverpool (QMUL to follow once se01 has been sorted out).
* New VOs: Thanks to Jeremy Coles (JC) we have two new VOs for the campaign - LIGO and SuperNEMO. The former tried and failed to get started a few years ago due to issues with OSG and the latter is actually a VO where all of the users are inactive (so it's more of a VO resurrection). We look forward to seeing how they get on with the new system!
SI-2 ATLAS weekly review & plans
---------------------------------
RJ had left the meeting, but had noted there was nothing major to report.
SI-3 CMS weekly review & plans
-------------------------------
DC had nothing major to report.
SI-4 LHCb weekly review & plans
--------------------------------
PC was absent.
SI-5 Production Manager's Report
---------------------------------
JC reported as follows:
1) As per my message circulated on Friday, GridPP cloud instances are ready to receive additional resources (providing the experiment accounting is reconfirmed as accurate) and a strategy decision is needed at this point to put more resource behind the cloud instances or not. The direction of the GridPP cloud resources at Imperial also needs discussion.
2) The GridPP DIRAC service, whilst proving successful with individual VOs, is encountering issues that need to be resolved in order to properly support multiple VOs. We will review in the coming weeks the intended move to production in May.
3) An LHC VO/WLCG IPv6 working group target for Tier-1s to have enabled a dual-stack perfSONAR (by April) is approaching. Progress is being made but RAL may not make the target date.
4) The LHCb migration to Dirac File Catalogue (DFC) appears to have been smooth Ð some issues were fixed quickly during the initial use.
5) ATLAS has had an issue with multicore job lengths and effective resource utilisation which is being addressed. There have also been some FTS3 share issues for intra-VO groups.
6) At the ops meeting last week we reviewed progress and issues at sites. Some things to note: sites are increasing their multi-core capacities (i.e. moving more resource behind these queues); more sites are looking at HTCondor as a batch system; additional sites are looking at deploying a VAC instance; IPv6 allocations are increasing but dual-stack developments are very slow; perfSONAR is back up at the majority of sites. For RAL T1, Andrew Lahiff said he has been looking at using cloud resources on demand to provide worker nodes to the batch system, and mentioned a plan to allow worker nodes to run SL7 on the hardware, but to run jobs in SL6 containers. Sites are ready for Run-II.
7) There was an EGI OMB last week. Here is a snapshot of some things that came up. EGI will start a review of critical services in the summer Ð a survey to gather NGI views will circulate soon. A new global VO has been created for EUBrazil Cloud Connect (eubrazilcc.eu). A new accounting portal format will be released soon Ð this includes an API to query data. There was a reminder about the EGI conference in May: http://go.egi.eu/reg2015. The cloud security questionnaire responses have been evaluated and recommendations taken forward. FedCloud security challenges are being discussed. An approach to monitoring ARGO is progressing (it is lighter to maintain than Nagios). There was an update on VO SLAs. Finally EGI-Engage core competency centres (exist for BBMRI, DARIAH, EISCAT_3D, ELIXIR, EPOS, INSTRUCT, LifeWatch & Disaster mitigation) are surveying whether there is interest in joining them as unfunded partners.
SI-6 Tier-1 Manager's Report
-----------------------------
AS noted nothing major to report.
SI-7 LCG Management Board Report of Issues
-------------------------------------------
There was no MB.
REVIEW OF ACTIONS
=================
533.3 DK to draft a preliminary response for any future security incidents where the press might contact GridPP (Tom Whyntie should be briefed on this as well in case he is contacted direct). Ongoing.
549.2 AS to update all sites who hold Extreme kit, regarding the outages over the Christmas period. Ongoing.
550.3 PG to provide an assessment of the needs of non-LHC experiments for the GridPP5 proposal. Done, item closed.
550.4 PG to work on the Finance Report and the Project Manager's Report for the OC meeting. Done, item closed.
550.7 SL/PG to consider the Agenda for GridPP34 (theme is Run-2; 'other users' non-LHC talks to be included). Ongoing.
551.1 ALL: to begin work urgently on the Oversight Committee documents. Done, item closed.
553.1 DC/PC to provide experiment OTP numbers showing effort worked on computing operations who were not ATLAS/CMS/LHCb members. Done, item closed.
554.1 RJ/JC to take forward the Sussex situation regarding their Tier-status.
554.2 PG/SL/DK/DB/JC/DC/CD Ð to review the Risk Register for their risks and let PG know they were happy, or not, with the current levels. Done, item closed.
554.4 DK to contact EDFA JET and discuss a possible move to a more lightweight model than a Tier-2 site. Ongoing.
555.1 PG to sanity-check the hardware and staff effort figures in the GridPP5 proposal. Done, item closed.
555.2 PG to clarify the breakdown of travel costs/OPN figures. Done, item closed.
555.3 ALL: to respond to requests for input on the GridPP5 proposal, by return, this week. Done, item closed.
555.4 AS to look at the figures regarding the possibility of DIRAC using tape media in FY15, and report back to DB. Done, item closed.
ACTIONS AS OF 02.03.15
======================
533.3 DK to draft a preliminary response for any future security incidents where the press might contact GridPP (Tom Whyntie should be briefed on this as well in case he is contacted direct).
549.2 AS to update all sites who hold Extreme kit, regarding the outages over the Christmas period.
550.7 SL/PG to consider the Agenda for GridPP34 (theme is Run-2; 'other users' non-LHC talks to be included).
554.1 RJ/JC to take forward the Sussex situation regarding their Tier-status.
554.4 DK to contact EDFA JET and discuss a possible move to a more lightweight model than a Tier-2 site.
The next PMB would take place on Monday 9th March @ 12.55 pm.
GridPP PMB Meeting 555 (23.02.15)
=================================
Present: Dave Britton (Chair), Andrew Sansum, Dave Colling, Jeremy Coles, Tony Doyle, Steve Lloyd, Dave Kelsey, Claire Devereux, Pete Gronbech, Tony Cass
(Minutes Ð Suzanne Scott)
Apologies: Roger Jones, Pete Clarke
1. GridPP5
===========
DB had circulated the latest version of this document. The main section still to be written was the 70% scenario. It was difficult to know what STFC meant by 'flat cash' in the context of the three, four, and five-year projects. DB presented the hardware costs and the staff effort. PG was asked to sanity-check the figures. DB asked if the travel budget was manageable? What did the 'misc' category cover? PG to clarify the breakdown of travel and OPN figures.
ACTION
555.1 PG to sanity-check the hardware and staff effort figures in the GridPP5 proposal.
555.2 PG to clarify the breakdown of travel costs/OPN figures.
DB turned to the 90% scenario, where a balance between the hardware and the FTE was required. The hardware budget in this scenario would be cut to 90%, however we needed to ask STFC what we do about meeting the pledges etc, guidance was required.
DB noted that the main focus would be on the 70% scenario, and this would be treated differently as it was the 'disaster' scenario and would mean 'damage control' mode with no chance of meeting the MoU. Hardware would be scaled in this model. STFC would have to explain this to the CRRB. Staff effort would be much reduced. The PMB had reached consensus before Christmas that the Tier-1 manpower would be reduced consistent with the original GridPP5 proposal, reflecting experiment priorities. AS confirmed he could present some options. DB asked the PMB whether they could make a strategic decision today? The response was yes. DB advised that we could present different options in the 70% scenario and ask the experiments for their preferences. DB asked for any objections to be voiced to the GridPP5 'Round 2' strategy? There were no objections and consensus was achieved.
DB outlined changes in the 70% funding assumptions, and each point was discussed and agreed in turn. There was a discussion of reductions in the Project Manager's and Production Manager's roles Ð we could request 0.5FTE for work on SKA for the Cambridge post, which would be below the line in order to help manage reductions. DB summarised the outcomes of the 70% scenario overall. JeS forms must be submitted, and University costs would be used. There ensued a discussion on the GridPP4+ hardware budget, and the purchases for this year.
Agreement was reached on the way forward, based on the GridPP5 strategy as outlined by DB, ensuring that we deliver as much as possible to the experiments. The draft proposal must be finished this week in order to allow time to circulate to the CB. There would be a CB meeting taking place next Monday.
ACTION
555.3 ALL: to respond to requests for input on the GridPP5 proposal, by return, this week.
2. Data Preservation Training
==============================
An email had been received from Ian Bird regarding a suggestion by Jamie Shiers of offering training for certification of the WLCG data centres as long term digital repositories. It was agreed that we should send someone, however it was expensive. Someone from the Tier-1 would be the best choice, and Matthew Viljoen had been interested in attending. DB was happy to part-fund his attendance.
3. European Open Science Cloud / Helix Nebula
==============================================
Information had been circulated about this EU initiative. DB needed to speak to Ian Bird. PC/DB had corresponded with Tony Medland, urging caution. TM would attend the proposed meeting.
4. DIRAC
=========
AS reported that he had emailed to DIRAC a cost estimate of tape media use in FY15. They would like to make use of tape storage at UCL in FY15 if we could make them an allocation out of our 'other' 10%. This might be possible for a minimal sum. What did we want to do? DB considered that there were both risks and opportunities here. The risk was that we did it, and they gave us ~£35k and then it turned out not to be possible to actually utilise the tape; or they go ahead and then no money is forthcoming for the allocation. We could try it this year? AS needed to look at the sums more closely. It would be preferable if they paid us. DB thought it would be good to do if it were really feasible. DB asked about slot count? AS needed to run the model, but it should work. JC thought we should do it in time for Run 2. PG noted that MICE had also asked for 1PB and there was already 1PB allocated to T2K. The MICE PB needed to be paid for.
ACTION
555.4 AS to look at the figures regarding the possibility of DIRAC using tape media in FY15, and report back to DB.
STANDING ITEMS
==============
SI-0 Monthly Report from Development (Cloud) Group
---------------------------------------------------
There was no report.
SI-1 Dissemination Report
--------------------------
SL reported on behalf of Tom Whyntie as follows:
### New User Engagement Campaign - Update
#### UCLan - GalDyn
* The groups software has now been compiled using the GridPP-contextualised CERN VM (SL6). TW has received great feedback on the process. (On a personal note, some very satisfying sharing of Open Software techniques such as documenting code with Markdown.) Now in a position to deploy software via CVMFS and determine optimal simulation scaling.
#### New VO creation
* UCLan and the QMUL Proteomics groups are now in a position to create Virtual Organisations (VOs) for their work. Steve Jones (SJ), Jeremy Coles (JC), Jens Jensen (JJ, KeyDoc owner), Robert Frank (RF) and TW have been discussing an update of the procedure documented here https://www.gridpp.ac.uk/wiki/New_VO_deployment There is some work to be done but we have a great opportunity to update it now and get plenty of feedback on the process.
#### Monitoring recent VO activity
* Part of writing the Engagement quarterly reports involves using the EGI accounting portal. Via a few GGUS tickets, TW has become aware of a method for dynamically creating a list of recently used VOs, rather than using a check-box list populated on the entire history of the NGI. In a nutshell, use the development version of the portal here: http://accounting-devel.egi.eu/egi.php?SubRegion=1.65 which populates the VO list for "Custom" based on the date range selected.
SI-2 ATLAS weekly review & plans
---------------------------------
RJ was absent.
SI-3 CMS weekly review & plans
-------------------------------
DC noted nothing major to report.
SI-4 LHCb weekly review & plans
--------------------------------
PC noted nothing major to report.
SI-5 Production Manager's Report
---------------------------------
JC reported as follows:
1)
a. LHCb has had problems with its central DiRAC configuration service (request load) and has not been running many jobs. They are also just now migrating from the LFC to DFC; they will run both catalogues in parallel until confident with the new setup.
b. ATLAS job loads have increased. MC15a (will take 3 months) starting very soon.
c. CMS site readiness tests are showing some intermittent problems at one GridPP site, this is being investigated as the cause is unclear.
2) We are still resolving issues with LIGO (no active VOMS) and LSST (it has VOMS but OSG hosted and at EMI-2 so requiring UI downgrades! We also need to understand the CVMFS needs).
3) Many VOs have either been renamed or stopped functioning over the years. Whilst their presence in accounting/support portals has little operational impact, they can lead to confusion and ideally they would have been removed. Tom has given a kick-start to our clearing of obsolete VOs: https://www.gridpp.ac.uk/wiki/VO_Cleanup_Campaign.
4) CERN@school first data has now been uploaded. Tom reports good progress with running jobs on a contextualized installation on a virtual box instance.
5) In recent weeks we have made steady progress with open tickets. We are currently down to 16 open across GridPP. Of these 5 are on-hold (e.g. awaiting a glexec tarball), 2 are waiting for reply and all others are in-progress. Thanks to everyone at the sites for supporting this campaign!
6) An HTTP task force has been proposed to form under the WLCG ops coordination area (this was presented at the MB and ops meeting: ). The experiments have been asked to confirm their interest in wider http deployment before a decision is taken. We have other use-cases that will benefit in the UK (e.g. MICE), but are unclear on the LHC experiment directions and thus whether a "task forceÓ is needed right now to push deployment.
A) Jamie is advertising a ÒTraining Course, run by the authors, on ISO 16363 for Auditors and Managers of Digital RepositoriesÓ. Few people have responded so far.
SI-6 Tier-1 Manager's Report
-----------------------------
AS reported as follows:
Fabric:
1) CPU procurements - one tranch delivered - second due by end of month.
2) Disk procurement - one tranch delivered - second no due date available yet - although build is reported to be proceeding.
3) Router issue continues - waiting for a diagnosis from vendor following second test in operation last Thursday.
4) On Friday both our core Z9000 switches restarted in close succession. No break in service owing to resilience of the core Tier-1 network. There appears to be a pattern that these switches have restarted every 75 days. Cause still to be identified.
Service:
1) Operations report at:
A generally quiet week.
http://www.gridpp.ac.uk/wiki/Tier1_Operations_Report_2015-02-18
Staff:
1) Seeking approval for a 3-month contract to work on developing a GRIDFTP plugin for CEPH.
2) Recruiting year-in-industry post to work on cloud and storage projects. Start date - making progress but further delay leading to expected March.
3) Two posts in Fabric team: adverts closed, interviewing shortly.
AS added that Catalin was currently working with LIGO to get them onto the Tier-1, discussions were ongoing. Also, they were meeting with LOFAR tomorrow to see if they could get them onto the cloud infrastructure.
SI-7 LCG Management Board Report of Issues
-------------------------------------------
DB reported that there had been an MB meeting Ð there was nothing major on Service; there had been an update on the experiment status for Run-2; there was forward planning from Ian Bird: for the HL-LHC regarding costs, also the computing model for HL-LHC Ð the WLCG workshop at CHEP would help get discussions underway; there was also discussion on FCC 2035 and the timeframe involved.
There was no time to check all of the actions Ð could everyone check their own actions.
REVIEW OF ACTIONS
=================
533.3 DK to draft a preliminary response for any future security incidents where the press might contact GridPP (Tom Whyntie should be briefed on this as well in case he is contacted direct).
549.2 AS to update all sites who hold Extreme kit, regarding the outages over the Christmas period.
550.3 PG to provide an assessment of the needs of non-LHC experiments for the GridPP5 proposal.
550.4 PG to work on the Finance Report and the Project Manager's Report for the OC meeting.
550.7 SL/PG to consider the Agenda for GridPP34 (theme is Run-2; 'other users' non-LHC talks to be included).
551.1 ALL: to begin work urgently on the Oversight Committee documents.
553.1 DC/PC to provide experiment OTP numbers showing effort worked on computing operations who were not ATLAS/CMS/LHCb members.
554.1 RJ/JC to take forward the Sussex situation regarding their Tier-status.
554.2 PG/SL/DK/DB/JC/DC/CD Ð to review the Risk Register for their risks and let PG know they were happy, or not, with the current levels.
554.3 SL to ask Tom to send to CD the GridPP publications list on our website and highlight which ones were relevant to EGI (default was that they were probably all relevant, unless obviously not so). Done, item closed.
554.4 DK to contact EDFA JET and discuss a possible move to a more lightweight model than a Tier-2 site.
554.5 SL to do doodle poll to the CB for possible meeting from 2nd to 4th of March to discuss GridPP5. Done, item closed.
ACTIONS AS OF 23.02.15
======================
533.3 DK to draft a preliminary response for any future security incidents where the press might contact GridPP (Tom Whyntie should be briefed on this as well in case he is contacted direct).
549.2 AS to update all sites who hold Extreme kit, regarding the outages over the Christmas period.
550.3 PG to provide an assessment of the needs of non-LHC experiments for the GridPP5 proposal.
550.4 PG to work on the Finance Report and the Project Manager's Report for the OC meeting.
550.7 SL/PG to consider the Agenda for GridPP34 (theme is Run-2; 'other users' non-LHC talks to be included).
551.1 ALL: to begin work urgently on the Oversight Committee documents.
553.1 DC/PC to provide experiment OTP numbers showing effort worked on computing operations who were not ATLAS/CMS/LHCb members.
554.1 RJ/JC to take forward the Sussex situation regarding their Tier-status.
554.2 PG/SL/DK/DB/JC/DC/CD Ð to review the Risk Register for their risks and let PG know they were happy, or not, with the current levels.
554.4 DK to contact EDFA JET and discuss a possible move to a more lightweight model than a Tier-2 site.
555.1 PG to sanity-check the hardware and staff effort figures in the GridPP5 proposal.
555.2 PG to clarify the breakdown of travel costs/OPN figures.
555.3 ALL: to respond to requests for input on the GridPP5 proposal, by return, this week.
555.4 AS to look at the figures regarding the possibility of DIRAC using tape media in FY15, and report back to DB.
The next PMB would take place on Monday 2nd March @ 12.55 pm.
GridPP PMB Meeting 554 (16.02.15)
=================================
Present: Dave Britton (Chair), Andrew Sansum, Dave Colling, Jeremy Coles, Tony Doyle, Roger Jones, Steve Lloyd, Dave Kelsey, Claire Devereux (Minutes Ð Suzanne Scott)
Apologies: Pete Gronbech, Tony Cass, Pete Clarke
1. PRACE pilot calls
=====================
CD reported that PRACE had not been willing to attribute acknowledgements of resources Ð all accreditation would be given to PRACE and not to EGI or to resource owners Ð EGI had therefore withdrawn from the Call.
JC voiced his concern about the end section of Tiziana's email, which seemed to contradict the parameters of the Call. CD commented that EGI does not own resources, the institutes do, and these are pre-allocated to projects. DB thought the wording was certainly misleading, but probably not what Tiziana had meant.
2. GridPP5 Services & Operations
=================================
JC had sent an update to DB. DB reported that JC had compiled a 'service view' divided into three categories: EGI/WLCG/UK by and for UK, however there were overlaps in many places. JC had added some figures and DB needed to analyse these and ascertain their relevance/usability for the Proposal. JC considered that he had underestimated in a lot of categories. DB would look at this today if he could.
3. GridPP5
===========
DB had circulated the introductory sections of this document. He asked for actions from PMB members: DC, CD, PC, AS.
DB had received information from CD. DC reported he was awaiting a response from someone else at present. AS was working on his section.
DB had written the start of Section 4, however the work packages were not currently aligned with the 'service view' table. The 'service view' was not the best way of presenting the work packages; Section 4 was nearing the 'first draft' stage. Section 5 was on experiment hardware resources. AS/DB had been iterating on hardware planning. DB noted he must move onto the scenarios next.
SL asked about the Collaboration Board (CB)? Could we show them a full draft once this was available? It was noted that the timing was very tight indeed. DB agreed we must schedule something. The Proposal had to be submitted by 10th March. Would the week prior to this be feasible? Week commencing 2nd March? Would we have a fully completed draft by then? TD advised that the deadline for Consolidated Grants' equipment purchases (CapEx) bids was that Thursday 5th of March, so that might impact availability of people. SL thought that the Monday to Wednesday of that week, 2nd to 4th of March would be possible. He would do a doodle poll for the CB meeting. SL had sent DB the Impact information. DB advised he was hoping to send round the latest draft document soon.
ACTION
554.5 SL to do doodle poll to the CB for possible meeting from 2nd to 4th of March to discuss GridPP5.
4. Oversight Committee (OSC)Documents & Quarterly Reports (QRs)
================================================================
PG had circulated the latest draft of the Finance Report and the Project Status document; also the Risk Register. These needed work and DB must look at them. AS/PG had done the Risk Register together last week and had made changes and additions. There ensued discussion on the wording of some Risk Register sections, and it was agreed that revisions were required. This could be revisited in due course once the other documents were near completion.
ACTION
554.2 PG/SL/DK/DB/JC/DC/CD Ð to review the Risk Register for their risks and let PG know they were happy, or not, with the current levels.
It was noted that the Risk Register was the version on the website. DB asked if there were any QRs outstanding? AS had done his. DK had done his. DC had done his as well.
5. AOB
=======
- Dirac: AS had circulated emails received from Rachel Reynolds regarding the storing of data for Dirac. AS had done costings for Jeremy Yates. We could charge them for the hardware, however the key issue was 5PB of Dirac data could equate to 16%, which was a significant operational share. AS had not yet received a reply from JY. It was noted that the 'World Object Store' (WOS) was not the same as backing-up data to tape. It was not clear exactly what JY required. Disk would be very expensive and very complicated.
PC had emailed to say that Edinburgh could help with data storage. DB said we would wait and see what the response was before doing anything further. From a technical point of view we would need to investigate the options/possibilities. The PMB agreed that we would be happy to do all that we could to assist. AS was given a mandate to pursue this issue with Dirac.
- EGI publications: CD reported that EGI were currently collating lists of publications, and she needed sight of the papers instigated by the Collaboration. The UK had not submitted any to EGI thus far. The GridPP ones were eligible to be included in the EGI listings. Could Tom look through the GridPP publications and add them? It was agreed that SL would ask Tom to send to CD the GridPP list on our website and highlight which ones were relevant to EGI. It was good that the UK was represented on the EGI listings.
ACTION
554.3 SL to ask Tom to send to CD the GridPP publications list on our website and highlight which ones were relevant to EGI.
DB noted that all of the LHC papers related to EGI because they used the infrastructure but were not attributed to a particular country.
- GridPP34: DB noted that Registrations had to be opened by this Friday 20th. He would discuss with SL.
STANDING ITEMS
==============
SI-0 Monthly Report from Development (Cloud) Group
---------------------------------------------------
There was no report.
SI-1 Dissemination Report
--------------------------
SL noted that a large report had been submitted last week; there was nothing new or major to report.
SI-2 ATLAS weekly review & plans
---------------------------------
RJ noted nothing major to report. Regarding his actions, 553.1 was done; 553.2 was done; for 553.3 RJ queried the wording, as this was not comprehensible to him. DB wished the action changed to read: 'We in the UK should decide how those sites should be presented, possibly as Tier-3 sites rather than Tier-2'. It was agreed that UCL should be removed and it had to be discussed with Sussex regarding their status. DB noted this was about supporting Sussex, consistent with the effort they had available.
ACTION
554.1 RJ/JC to take forward the Sussex situation regarding their Tier-status.
SI-3 CMS weekly review & plans
-------------------------------
DC noted nothing major to report.
SI-4 LHCb weekly review & plans
--------------------------------
PC noted nothing major to report.
SI-5 Production Manager's Report
---------------------------------
JC reported as follows:
1. RAL T1 has been working to establish some LIGO (http://www.ligo.org) users in Cardiff for access to GridPP resources (e.g. CVMFS). OSG appears to host a LIGO VO, but all supporting services appear to have expired (VOMS for example just hangs). I am trying to make contact with OSG colleagues to check the status but it is very likely that we will want to setup LIGO on our VOMS. In the meantime we will enable the users within a regional VO so they can get familiar with the tools used. Are there any comments on, or opposition to, this plan?
2. Marcel Stanitzki from DESY joined the Storage Group meeting last week to report on the ILC data model. They have a smilier approach to that used by Belle-II. More details in the minutes: http://storage.esc.rl.ac.uk/weekly/20150211-minutes.txt. GridPP was thanked for its contributions.
3. WLCG T2 reliability/availability figures for January have been circulated. As usual sites below the 90% WLCG target have been asked to provide details and these appear below where a response was obtained.
ALICE (http://sam-reports.web.cern.ch/sam-reports/2015/201501/wlcg/WLCG_All_Sites_ALICE_Jan2015.pdf)
- All okay.
ATLAS (http://sam-reports.web.cern.ch/sam-reports/2015/201501/wlcg/WLCG_All_Sites_ATLAS_Jan2015.pdf)
- UCL-HEP: 21%:21%
- ECDF: 77%:77%
- Sussex: 26%:29%
CMS (http://sam-reports.web.cern.ch/sam-reports/2015/201501/wlcg/WLCG_All_Sites_CMS_Jan2015.pdf)
- All okay.
LHCb (http://sam-reports.web.cern.ch/sam-reports/2015/201501/wlcg/WLCG_All_Sites_LHCB_Jan2015.pdf)
- ECDF: 79%:79%
- EFDA: 72%:72%
Feedback from the sites:
EFDA: No response. No GOCDB entries.
ECDF: This was primarily due to our Argus service being offline due to a hardware failure on two separate occasions. Unfortunately the first instance occurred on a Friday which meant it was not picked up when service was interrupted over the weekend. The service was hosted on a virtual instance and we have now migrated the service to a (helpfully) more robust hosted infrastructure. The original server hosting the instance is awaiting repair - backup service will be set up to avoid further outages when this service is offline.
Sussex: we spent most of January in downtime due to initially a problem with our lustre storage system, where multiple drive failures meant we needed to manually transfer and recover data. We also had problems with the kernel upgrade due to our use of IPoIB which has un-reported (other than by EGI) problems in the latest 6.5 and 6.6 kernels.
UCL-HEP: No entries in GOCDB and no feedback.
Regarding EDFA JET and UCL, there ensued a discussion about whether or not to continue them as part of the active site list, or to move them to a more lightweight model. DB considered that we must do something different with them, they could not continue as they were presently. DB noted there was no more hardware or manpower there. Did they want to continue as GridPP sites? UCL would now be removed as an ATLAS Tier-2, beyond that, it depended on these sites' aspirations as lightweight models? DB advised that discussion was required, but that continuing the way things were was not an option. Regarding EDFA JET, DK would contact them to discuss the way forward.
ACTION
554.4 DK to contact EDFA JET and discuss a possible move to a more lightweight model than a Tier-2 site.
4. Given the observations last week regarding poor ATLAS ASAP site performance for a couple of sites and the lack of GridPP funding for those sites questions have arisen about whether we need to take more action in this area. For example moving such sites to Ôlight weightÕ options including being diskless, becoming a VAC/vcycle instance or simply putting the site into downtime (and if no resolution occurs within a set period returning it to un-certified). Does the PMB want GridPP to start examining these possibilities or would it rather leave it to the experiments to delist the sites in their queues?
5. Following the issue of CERN accounts mentioned at the PMB last week, a quick process to sign-up GridPP staff with LCG association has been agreed.
6. Many thanks to Adam Huffman for his contributions to GridPP during his time at Imperial. Adam is moving on to work at the Francis Crick Institute in London.
The PMB wished to thank Adam for his valuable contributions to GridPP, and they wished him well for the future.
7. A summary of last weekÕs GDB is available: https://twiki.cern.ch/twiki/bin/view/LCG/GDBMeetingNotes20150211. At the LHCOPN/LHCONE meeting last week Brocade introduced a CERN partners program, which can be used for savings on their equipment and support. The pre-GDB on traceability in clouds/VMs (UK led) produced a number of actions (https://twiki.cern.ch/twiki/bin/view/LCG/20150210PreGDB) to take the area forward in areas such as setting up a central syslog test and investigating the use of net flow.
SI-6 Tier-1 Manager's Report
-----------------------------
AS reported there were no major issues, all was routine.
SI-7 LCG Management Board Report of Issues
-------------------------------------------
There had been no meeting.
REVIEW OF ACTIONS
=================
533.3 DK to draft a preliminary response for any future security incidents where the press might contact GridPP (Tom Whyntie should be briefed on this as well in case he is contacted direct).
548.2 DB to revisit the European context descriptions and rewrite/expand them to cover such areas as the HEP S/W foundation, EU-T0, Cloud SIG and ZEPHYR. Done, item closed.
549.2 AS to update all sites who hold Extreme kit, regarding the outages over the Christmas period.
550.3 PG to provide an assessment of the needs of non-LHC experiments for the GridPP5 proposal.
550.4 PG to work on the Finance Report and the Project Manager's Report for the OC meeting.
550.7 SL/PG to consider the Agenda for GridPP34 (theme is Run-2; 'other users' non-LHC talks to be included).
551.1 ALL: to begin work urgently on the Oversight Committee documents.
553.1 RJ/DC/PC to provide experiment OTP numbers showing effort worked on computing operations who were not ATLAS/CMS/LHCb members. RJ done.
553.2 RJ/DC/PC to send to DB any relevant sections from the Consolidated Grants' experiments' submissions re their statements on the GridPP5 interface. Done, action closed.
553.3 ALL: Regarding the exposure of site availability information at the ICB, we in the UK should decide how the 'goodwill' sites (UCL, Sussex) should be presented, possibly as Tier-3 sites rather than Tier-2. Done, action closed.
ACTIONS AS OF 16.02.15
======================
533.3 DK to draft a preliminary response for any future security incidents where the press might contact GridPP (Tom Whyntie should be briefed on this as well in case he is contacted direct).
549.2 AS to update all sites who hold Extreme kit, regarding the outages over the Christmas period.
550.3 PG to provide an assessment of the needs of non-LHC experiments for the GridPP5 proposal.
550.4 PG to work on the Finance Report and the Project Manager's Report for the OC meeting.
550.7 SL/PG to consider the Agenda for GridPP34 (theme is Run-2; 'other users' non-LHC talks to be included).
551.1 ALL: to begin work urgently on the Oversight Committee documents.
553.1 DC/PC to provide experiment OTP numbers showing effort worked on computing operations who were not ATLAS/CMS/LHCb members.
554.1 RJ/JC to take forward the Sussex situation regarding their Tier-status.
554.2 PG/SL/DK/DB/JC/DC/CD Ð to review the Risk Register for their risks and let PG know they were happy, or not, with the current levels.
554.3 SL to ask Tom to send to CD the GridPP publications list on our website and highlight which ones were relevant to EGI (default was that they were probably all relevant, unless obviously not so).
554.4 DK to contact EDFA JET and discuss a possible move to a more lightweight model than a Tier-2 site.
554.5 SL to do doodle poll to the CB for possible meeting from 2nd to 4th of March to discuss GridPP5.
The next PMB would take place on Monday 23 February @ 12.55 pm.
GridPP PMB Meeting 552 (02.02.15)
=================================
Present: Dave Britton (Chair), Pete Gronbech, Pete Clarke, Andrew Sansum, Claire Devereux, Dave Colling, Jeremy Coles, Dave Kelsey, Tony Doyle, Roger Jones (Minutes Ð Suzanne Scott)
Apologies: Steve Lloyd, Tony Cass
1. STFC Strategy Consultation
==============================
PC had circulated the document and had received no comments, so he assumed it was fine. JC offered to help to tidy it up, then it would go to DB to send off. It would be good to give feedback to STFC regarding the framing of their questions and their underlying assumptions. CD reported that the nature of the survey was due to be addressed at high-level.
2. PRACE Pilot Calls
=====================
PC had circulated the information. PC was on the DIRAC Resource Allocation Committee. He understood that these calls were PRACE trying to engage with the community, eg: EUT0. DB asked if this had been initiated by PRACE? CD wasn't sure, however EGI were doing this activity across communities at present. DB asked, on behalf of GridPP, whether/how much we should commit storage? We could not guarantee ring-fenced resources in this regard. Tiziana's request to us was for us to guarantee some storage Ð we can respond in two ways: 1. we cannot guarantee any at all; 2. we can guarantee a small amount. AS advised that EUT0 were currently discussing collaboration with PRACE. Under the EUT0 banner we had already committed some TB to PRACE Ð 200TB is therefore available as already agreed. PC noted we should therefore indicate willingness. JC had received a Survey. DB considered that the danger was that we generally invite new groups to use resources therefore it was difficult for us to allocate some TB which are actually allocated elsewhere and unknown to us. We need to know where the resources are being used Ð we can't give away resources without some element of oversight. DB suggested that we will count the Tier-1 commitment as 200TB and see how this goes. JC could complete the survey with assistance from AS.
ACTION
552.1 JC to complete the PRACE Survey with inputs from AS/DB.
3. GridPP5 Services & Operations
=================================
DB reported that progress was being made, JC had circulated a revised spreadsheet of Ops and Services. JC had taken on board the suggestions and feedback, and had sub-divided the sections; there was simple colour-coding at present and contributions had been added. DB thought the spreadsheets looked good Ð some more info. could be included. Can we add Appendices within the GridPP5 Proposal word limit? Small tables might work, however the GridPP/NGI effort sum was presently incorrect. We need to consider exactly what this column represents. JC noted that these were all task-based percentages across all tasks, but that it was extremely difficult to sub-divide in this way. DC considered that the level of detail here was useful internally but not externally. DB agreed and suggested that we need small tables with colour-coded areas, without the detailed figures analysis. These tables could then link to larger tables showing the GridPP4 effort. These elements must be re-visited prior to making the tables public. There was very useful detail here for the Proposal. JC would continue compiling the table. DB noted we could extract relevant information for the body of the Proposal at higher-level.
ACTION
552.2 JC/DB to continue to work on the Services & Ops spreadsheets for the GridPP5 Proposal.
4. GridPP5
===========
DB reported that he had made progress on the hardware planning. He would send the details to PG after he had done a bit more work to finalise it. PG noted that he had received no responses to his request for estimates of usage. DC advised that there were references to projections to 2020 in an ECFA meeting towards the end of 2014. DB checked the Agenda and some of the talks, but could find nothing relevant. DC would check. DB noted that the Consolidated Grants had now all been submitted, therefore this should free-up people's time to work on the GridPP Proposal. DB would circulate a list of tasks for documents and would prioritise PC's contributions as PC had travel constraints.
ACTION
552.3 DB to circulate a list of tasks for GridPP5 Proposal documents and would prioritise PC's contributions as PC had travel constraints.
5. Oversight Committee (OC) and Quarterly Reports
==================================================
PG had received inputs for the document from LHCb and Dissemination. Quarterly Reports were URGENTLY required Ð he had received around 6 Reports but others were very urgently required, especially the Tier-1 Report, Security, Operations. Following receipt of these PG could work on the Manpower and Finance sections of the document. PG had an updated GridPP Plan with SLA figures. Other sections relied on the Quarterly Reports and also information from DK. DB urged the PMB members to finish their QRs ASAP as a priority.
6. GridPP34
============
DB needed to talk to SL. PG had created a skeleton Agenda. DB had invites to send out, including inviting LSST to do a talk. PC would send round an introductory mail to LSST in the first instance.
STANDING ITEMS
==============
SI-0 Monthly Report from Development (Cloud) Group
---------------------------------------------------
There was no report.
SI-1 Dissemination Report
--------------------------
There was no report, SL was absent.
SI-2 ATLAS weekly review & plans
---------------------------------
RJ reported nothing major was happening, their 2014 work was finished, queues were not full; Computing & Software Week was this week; there were reboots at RAL.
SI-3 CMS weekly review & plans
-------------------------------
DC noted nothing major to report, Cloud work was ongoing; CMS was empty, ready for the new campaign.
SI-4 LHCb weekly review & plans
--------------------------------
PC advised that they had lost files during stripping recently, but this happened outwith the UK Ð the PM had been done and it appeared that this problem can occur due to the nature of the process and appears legitimate.
SI-5 Production Manager's Report
---------------------------------
JC reported as follows:
1) There is a low number of multi-core jobs at the moment. As a consequence, sites are asking for a check on their single-core job queues to make sure they get their proportion of that work.
2) The final WLCG Tier-2 availability/reliability figures are now available at: https://espace2013.cern.ch/WLCG-document-repository/ReliabilityAvailability/2014/december-14/. No GridPP changes since the figures were reported to the PMB a few weeks ago.
3) ECDF is ready for LSST test jobs but the activity is awaiting clarification from VO users. The VO-admins have moved on since their registration in EGI tools, so others in the VO are being consulted.
4) A buffer overflow vulnerability CVE-2015-0235 in the GNU C Library (glibc) called the "GHOST" vulnerability has been announced which may be exploited remotely. So far no exploit has been found in software commonly used on the EGI Infrastructure. Sites are updating as a precaution.
5) The February pre-GDB (10th Feb) topic is on cloud traceability. GridPP has several levels of engagement Ð Ian Collier is leading the working group, Andrew McNab is contributing on VAC/Vcycle/VMS and David Crooks will represent our core-operations interests.
6) There was an EGI OMB last Thursday: There was a proposal to stop EGI.eu tracking ROD performance; there is a campaign to upgrade FedCloud site accounting probes (note FedCloud now has now received 43 NatSci, 3 Medical and 3 Humanities use cases); there was a report from the Earth science e-infrastructures workshop (e.g. on the grid approaches of earthquake and seismology research). There was a reminder about the EGI 2015 conference (18th-22nd May in Lisbon - http://conf2015.egi.eu/ ) and early bird deadline of 5th April.
SI-6 Tier-1 Manager's Report
-----------------------------
AS reported as follows:
Fabric:
1) CPU procurements - delivery expected in February.
2) Disk procurement - delivery expected in February.
3) Work ongoing to diagnose and resolve the Tier-1 router fault. Currently running without resilience on the top router. A test carried out last week succeeded in reproducing the fault within a few minutes. No cause yet identified - issue is being escalated within vendor. We have identified a spare router available on site in case remaining router fails.
4) JANET wish to re-engineer our OPN backup link (as currently implemented on a network due to be retired). Expect work to be carried - Tuesday 3rd. The new link will be tested by failing-over to the link and testing it in production.
5) glibc security updates were carried out last week and are still ongoing this week on the CASTOR service.
Service:
1) Operations report at:
http://www.gridpp.ac.uk/wiki/Tier1_Operations_Report_2015-01-28
Staff:
1) Seeking approval for a 3-month contract to work on developing a GRIDFTP plugin for CEPH.
2) Recruiting 'year-in-industry' post to work on cloud and storage projects. Start date (delayed) but expected February.
3) Two posts in Fabric team: adverts closed Friday - shortlisting this week.
SI-7 LCG Management Board Report of Issues
===========================================
There had been no meeting.
AOB
===
DK reported that the post of Security Officer was now filled Ð they had made an offer to a candidate, which had been accepted. Start date would be 1st March 2015.
REVIEW OF ACTIONS
=================
533.3 DK to draft a preliminary response for any future security incidents where the press might contact GridPP (Tom Whyntie should be briefed on this as well in case he is contacted direct). Ongoing.
548.1 JC to look at ways to expand the Deployment, Operations and Support material to explain the necessity of the core operations work beyond individual site and experiment tasks. Done, item closed.
548.2 DB to revisit the European context descriptions and rewrite/expand them to cover such areas as the HEP S/W foundation, EU-T0, Cloud SIG and ZEPHYR. Ongoing.
548.3 DB & PG to develop the hardware planning for 2016+. RJ to provide ATLAS input asap. DC to provide CMS info to 2020 over the next few days. Done, item closed.
549.1 JC/CD to build a case for Tier-2 staff for the GridPP5 submission, showing how the staff effort supports the infrastructure aside from running the hardware. Done, item closed.
549.2 AS to update all sites who hold Extreme kit, regarding the outages over the Christmas period. Ongoing.
550.1 PC to lead the response on behalf of the PMB to the STFC Strategy Consultation. Done, item closed.
550.3 PG to provide an assessment of the needs of non-LHC experiments for the GridPP5 proposal. Ongoing.
550.4 PG to work on the Finance Report and the Project Manager's Report for the OC meeting. Ongoing.
550.6 DC to check why CMS CPU efficiency for November-December at the Tier-1 had been low at around 50-60%. Ongoing.
550.7 SL/DC to consider the Agenda for GridPP34 (theme is Run-2; 'other users' non-LHC talks to be included). Ongoing.
550.8 PC to look into the EGI Security Alert/kernel update, and the fact that ECDF had been close to suspension, to see if any lessons could be learned. Done, item closed.
551.1 ALL: to begin work urgently on the Oversight Committee documents. Ongoing.
551.2 JC to work on the Services & Operations spreadsheet to ensure all information was included and expanded where necessary. Ongoing.
551.3 DB to contact Jeremy Yates re his latest survey, and note to him that GridPP could not complete this on the timescale required. Done, item closed.
ACTIONS AS OF 02.02.15
======================
533.3 DK to draft a preliminary response for any future security incidents where the press might contact GridPP (Tom Whyntie should be briefed on this as well in case he is contacted direct).
548.2 DB to revisit the European context descriptions and rewrite/expand them to cover such areas as the HEP S/W foundation, EU-T0, Cloud SIG and ZEPHYR.
549.2 AS to update all sites who hold Extreme kit, regarding the outages over the Christmas period.
550.3 PG to provide an assessment of the needs of non-LHC experiments for the GridPP5 proposal.
550.4 PG to work on the Finance Report and the Project Manager's Report for the OC meeting.
550.6 DC to check why CMS CPU efficiency for November-December at the Tier-1 had been low at around 50-60%.
550.7 SL/DC to consider the Agenda for GridPP34 (theme is Run-2; 'other users' non-LHC talks to be included).
551.1 ALL: to begin work urgently on the Oversight Committee documents.
551.2 JC to work on the Services & Operations spreadsheet to ensure all information was included and expanded where necessary.
552.1 JC to complete the PRACE Survey with inputs from AS/DB.
552.2 JC/DB to continue to work on the Services & Ops spreadsheets for the GridPP5 Proposal.
552.3 DB to circulate a list of tasks for GridPP5 Proposal documents and would prioritise PC's contributions as PC had travel constraints.
The next PMB would take place on Monday 9th February @ 12.55 pm.
GridPP PMB Meeting 551 (26.01.15)
=================================
Present: Dave Britton (Chair), Pete Gronbech, Steve Lloyd, Pete Clarke, Andrew Sansum, Claire Devereux, Dave Colling, Jeremy Coles, Dave Kelsey (Minutes Ð Suzanne Scott)
Apologies: Tony Doyle, Roger Jones, Tony Cass
1. Oversight Committee Documents
=================================
PG noted that the Project Status Document should be added to the list. He could look through the last document produced. Quarterly Reports were still awaited and were required urgently now. He would also re-visit the Risk Register. DB had advised that the following documents must be initiated:
Introduction - DB
International Context - CD
Summary of GridPP4 status - PG
(includes summary of project map, deliverables and risks)
Tier-1 Status - AS
Deployment Status - JC
ATLAS Status - RJ
CMS Status - DC
LHCb Status - PC
Other VOs - ??
Impact and Dissemination Ð SL
PG would circulate the latest version of the September 2014 document. DB noted that Milestones and Metrics could be required for the GridPP4+ Grant. DB advised that the key message was that all of the above documents were required on the same timescale as the GridPP5 Proposal. These Oversight Committee documents must be completed ASAP. PC asked what level of view was required? DB confirmed 'high-level' view.
ACTION
551.1 ALL: to begin work urgently on the Oversight Committee documents.
2. GridPP5 Services & Operations
=================================
JC had circulated a spreadsheet in an attempt to analyse all tasks/areas in GridPP including those run under the auspices of EGI and WLCG. He had outlined the critical services in EGI, as well as the funded and unfunded GridPP contributions. Project months were noted per year. DB considered that this information was very useful. JC should expand the information and tighten-up the overview. DB advised that we would need a matrix showing this information, with coloured boxes etc, which will help get the main points across Ð use green for example, to show services run by GridPP. It was a very useful exercise to show all the things run by the Grid. JC should try to capture all of the relevant information and then see how it could be condensed. CD would help with the EGI services, she would add-in some columns. DB thought that a few tables might be required to summarise everything.
ACTION
551.2 JC to work on the Services & Operations spreadsheet to ensure all information was included and expanded where necessary.
3. GridPP5 Proposal
====================
DB asked if DC was yet in a position to provide the CMS hardware numbers? DC had already tried to elicit this information; he would send round another reminder. DB needed to insert the relevant numbers soon. The primary task this week was the Operations & Services info.
4. CHEP Attendance
===================
DK had circulated a summary spreadsheet of applicants. For CHEP 2013 there had been five accepted talks and 25 accepted posters. We funded 19 people @ 50% funding, which was around a £12k spend.
DK asked whether, for this time round, it was reasonable to accept 16 people given the current climate? Should we accept talks only, or accept everyone?
DB considered that individual institutes should optimise their travel. We should fund the talks first. DC observed that this was an important conference so we should fund as many people to go as we could. PG noted that we were ensuring 'due diligence' by reviewing applicants at PMB level. What was the current budget status? DK could not provide detail, however he had received, to date, 155 Visit Notices which were already approved, to the value of around £100k. He considered that we could probably afford to send people to CHEP. SL thought that it was hard to justify sending many people round the world for a poster only. There was also the WLCG Collaboration meeting to consider. DB thought that we could not justify three people from SCD for posters alone, however it was reasonable that one person from SCD should attend. One attendee from Lancaster could cover the posters from there, and the remainder of the attendees should give talks. DK noted that no-one from Imperial had actually been in touch with him (apart from DC), however four persons attending from Imperial may be too many. DC advised that at least two of the persons concerned were reluctant to attend CHEP. PG considered that this was a good self-selection. DK would assume two from Imperial.
DK would take this issue forward, noting that attendees from Imperial and RAL needed to be finalised; he would also need to speak to RJ. Apart from these, all of the other applicants were approved.
5. Jeremy Yates' latest survey
===============================
Another survey had been received from Jeremy Yates, with a close deadline of 9th March. This co-incided with many other important deadlines which converged at that time. PC noted that the survey would not affect GridPP and in addition we could not complete the survey on the timescale required due to GridPP5 constraints. It was agreed that it was not possible to complete this Survey. DB would respond to JY noting that we could not do this on the timescale required.
ACTION
551.3 DB to contact Jeremy Yates re his latest survey, and note to him that GridPP could not complete this on the timescale required.
6. GridPP34 @ QMUL
===================
PG reported that he had set up the Indico page this morning and would begin work on the Agenda. There ensued a discussion on the Hepix Agenda and attendance Ð this could be reviewed nearer the time. There were no UK Registrations as yet.
7. STFC Strategy Consultation
==============================
PC had now completed a draft of this document. It contained all of the data that he could think of. It now required clear instructions on any changes Ð email suggestions were welcome now, but only on clear items missed or specific items that needed to be changed/deleted Ð generic comments would not be helpful at this stage; then PC would send round a revised draft. PG raised some questions. PC went through the document:
Q2c: PC thought that the shortest statement was the strongest; DB agreed. Following discussion PC would add for clarity: 'WLCG MoU signed by STFC'.
DB would send the numbers on 'scale' in 2(a). PC should amend 'Technology'.
Q4: PC to add-in: non-LHC and non-Particle Physics.
Q5: bullet IV: add 'how to provide the increasing hardware and staff resources'
Q8: would Tier-2 staff be included? PC assumed no, but he could emphasise us as clients and what we need them to do in collaboration with Universities. GridPP was distributed facility so include the Tier-2 staff at Universities.
Q11: 'added value' would be to fund the expertise that was extant, so that we could apply for other funding opportunities.
Q12: neither the model nor the questions were appropriate! PC would phrase accordingly.
PC would do another draft, and provide a marked-up version of any specific changes required.
STANDING ITEMS
==============
SI-0 Monthly Report from Development (Cloud) Group
---------------------------------------------------
A Report (Minute) had been circulated by DC last time.
SI-1 Dissemination Report
--------------------------
SL reported on behalf of Tom Whyntie, as follows:
###Small VO/New User Campaign - update
* UCLan galaxy simulations: the ICL DIRAC team & TW have been working hard to setup two cosmologists at the University of Central Lancashire(UCLan) with access to grid resources via the ICL DIRAC instance. Currently we are setting up the DIRAC user interface (UI) on their local CentOS cluster. They have successfully joined the NorthGrid regional VO and should be ready to submit test jobs soon.
* The DIRAC new user checklist continues to evolve, now incorporating network tests for remote clusters: https://www.gridpp.ac.uk/wiki/DIRAC_new_user_checklist Huge thanks to the ICL team (DB, JM, DR, SF) for debugging support.
* CVMFS repositories for regional VOs: Catalin Condurache has kindly setup CVMFS repositories for the regional VOs:
** /cvmfs/londongrid.gridpp.ac.uk
** /cvmfs/scotgrid.gridpp.ac.uk
** /cvmfs/southgrid.gridpp.ac.uk
** /cvmfs/northgrid.gridpp.rl.ac.uk
The regional VO CVMFS repository policy continues to evolve; currently upload rights are assigned to the lcg-admin role. See
https://www.gridpp.ac.uk/wiki/CVMFS_Use_Cases_for_GridPP for more information
Huge thanks to Alessandra Forti (AF), DB, DR, Robert Frank (RF) for
testing and feedback during this process
SI-2 ATLAS weekly review & plans
---------------------------------
RJ was not present.
SI-3 CMS weekly review & plans
-------------------------------
DC noted nothing major to report.
SI-4 LHCb weekly review & plans
--------------------------------
PC noted nothing major to report.
SI-5 Production Manager's Report
---------------------------------
JC reported as follows:
1) An unscheduled outage of the EGI message broker (GRNET) caused a short-lived disruption to GridPP site monitoring (jobs failed) last Thursday 22nd January.
As operationally things were relatively quiet last week, the rest of the items included are effectively news highlights/for information:
2) Some ATLAS DQ2 tool issues have been arising. Everyone is being encouraged to move to the new Rucio clients.
3) There was an IPv6 F2F meeting at CERN last week. The Tier-1s were reminded of an April 2015 deadline to enable dual-stack on their perfSonar instances. A perfSonar dashboard showing the IPv6 network measurements via IPv6 across the WLCG that have enabled IPv6 on perfSonar has been proposed.
4) An updated draft of the WLCG workshop (Okinawa 11th-12th April) high-level agenda is available for comment: https://indico.cern.ch/event/345619/other-view?view=standard.
5) The deadline to decommission VOMRS at CERN has moved to 16th February so that issues with the VOMS-Admin replacement can be resolved. The tentative date for AFS-UI decommissioning remains 2nd February.
6) A RAID controller failure at SARA (NLT1) has led to significant data loss across the experiments and disrupted workflows.
SI-6 Tier-1 Manager's Report
-----------------------------
AS reported as follows:
Fabric:
1) CPU procurements - delivery expected in February.
2) Disk procurement - delivery expected in February.
3) Work on electrical safety testing is now complete. Over 6 days each power circuit powering the Tier-1's non-UPS services was powered down and inspected/tested. Disruption to the main production service was minimal.
4) Work ongoing to diagnose and resolve the Tier-1 router fault. Currently running without resilience on the top router. A test carried out last week succeeded in reproducing the fault within a few minutes. Now waiting on a response from supplier.
5) JANET wish to re-engineer our OPN backup link (as currently implemented on a network due to be retired). Expect work to be carried out on Tuesday 3rd. The new link will be tested by failing-over to the link and testing it in production.
6) The migration of all the T10KA and T10KA media to T10KB is now complete. 7729 A/B tapes were read and then rewritten onto B media. Only one tape had problems, these were for VO dteam and were associated with problems with CASTOR 2.1.8.
Service:
1) Operations report at:
http://www.gridpp.ac.uk/wiki/Tier1_Operations_Report_2015-01-21
2) STORAGE
a) CASTOR - upgrade CASTOR head-nodes to SL6 was completed.
Staff:
1) Seeking approval for a 3-month contract to work on developing a GRIDFTP plugin for CEPH.
2) Recruiting 'year in industry' post to work on cloud and storage projects. Start date (delayed) finalising but expected February.
3) Two posts in Fabric team still to advertise (SBS and Tier-1 finalising).
SI-7 LCG Management Board Report of Issues
===========================================
There had been no meeting.
REVIEW OF ACTIONS
=================
533.3 DK to draft a preliminary response for any future security incidents where the press might contact GridPP (Tom Whyntie should be briefed on this as well in case he is contacted direct).
548.1 JC to look at ways to expand the Deployment, Operations and Support material to explain the necessity of the core operations work beyond individual site and experiment tasks.
548.2 DB to revisit the European context descriptions and rewrite/expand them to cover such areas as the HEP S/W foundation, EU-T0, Cloud SIG and ZEPHYR.
548.3 DB & PG to develop the hardware planning for 2016+. RJ to provide ATLAS input asap. DC to provide CMS info to 2020 over the next few days.
549.1 JC/CD to build a case for Tier-2 staff for the GridPP5 submission, showing how the staff effort supports the infrastructure aside from running the hardware.
549.2 AS to update all sites who hold Extreme kit, regarding the outages over the Christmas period.
550.1 PC to lead the response on behalf of the PMB to the STFC Strategy Consultation.
550.2 PG to circulate an email to 'Other' Experiments to get an update on their requirements Ð we need to show the demand level for the 10% 'Others' allocation. Done, action closed.
550.3 PG to provide an assessment of the needs of non-LHC experiments for the GridPP5 proposal.
550.4 PG to work on the Finance Report and the Project Manager's Report for the OC meeting.
550.5 DB to look at the last Oversight Committe documents Ð the Finance Report and the Project Management Report, and identify the sections which other people should update. Done, action closed.
550.6 DC to check why CMS CPU efficiency for November-December at the Tier-1 had been low at around 50-60%.
550.7 SL/DC to consider the Agenda for GridPP34 (theme is Run-2; 'other users' non-LHC talks to be included).
550.8 PC to look into the EGI Security Alert/kernel update, and the fact that ECDF had been close to suspension, to see if any lessons could be learned.
ACTIONS AS OF 26.01.15
======================
533.3 DK to draft a preliminary response for any future security incidents where the press might contact GridPP (Tom Whyntie should be briefed on this as well in case he is contacted direct).
548.1 JC to look at ways to expand the Deployment, Operations and Support material to explain the necessity of the core operations work beyond individual site and experiment tasks.
548.2 DB to revisit the European context descriptions and rewrite/expand them to cover such areas as the HEP S/W foundation, EU-T0, Cloud SIG and ZEPHYR.
548.3 DB & PG to develop the hardware planning for 2016+. RJ to provide ATLAS input asap. DC to provide CMS info to 2020 over the next few days.
549.1 JC/CD to build a case for Tier-2 staff for the GridPP5 submission, showing how the staff effort supports the infrastructure aside from running the hardware.
549.2 AS to update all sites who hold Extreme kit, regarding the outages over the Christmas period.
550.1 PC to lead the response on behalf of the PMB to the STFC Strategy Consultation.
550.3 PG to provide an assessment of the needs of non-LHC experiments for the GridPP5 proposal.
550.4 PG to work on the Finance Report and the Project Manager's Report for the OC meeting.
550.6 DC to check why CMS CPU efficiency for November-December at the Tier-1 had been low at around 50-60%.
550.7 SL/DC to consider the Agenda for GridPP34 (theme is Run-2; 'other users' non-LHC talks to be included).
550.8 PC to look into the EGI Security Alert/kernel update, and the fact that ECDF had been close to suspension, to see if any lessons could be learned.
551.1 ALL: to begin work urgently on the Oversight Committee documents.
551.2 JC to work on the Services & Operations spreadsheet to ensure all information was included and expanded where necessary.
551.3 DB to contact Jeremy Yates re his latest survey, and note to him that GridPP could not complete this on the timescale required.
The next PMB would take place on Monday 2nd February @ 12.55 pm.
|
