GridPP Ops meeting, 21/01/14
Attendance noted as per transcript.
Expt. Updates
LHCb
====
Raja with Elena/Jeremy:
Nothing much from LHCb, mostly Monte Carlo simulations. Some progress on EFDA-JET. Some jobs yesterday (20th), stalled with problem with voms-proxy-init. Later version is installed than on lxplus, some options not available. ongoing on ticket.
Bristol not picking up jobs on ARC, problem with environment. Same for ARC at RALPP, CREAMs are fine.
Sheffield. Started September last year, until then things were fine, 15th-20th Sept. 2013. Nothing was done at that time at Sheffield. They haven't talked to University about it, not running much LHCb. Will put more time into this after glexec is finished.
- CMS
Daniela:
Very little problems, RALPP has problems with SAM tests, Chris is following that up.
- ATLAS
Elena with Kashif/Gareth:
Not much news. Prod tests are running since Christmas on single cores. New multicore simulations. Rucio renaming almost done. Small problems with QMUL (StoRM) and Manchester (broken disk server). Discussed with sites. Kashif noted problems with WNs crashing running production and asked if anyone else had seen it. Symptoms were a lot of open files. Mostly occurring with 64 core WNs. Gareth noted similar problem at Glasgow, upped the number of files cvmfs could cache which fixed the issue. Glasgow also seeing with 2.1.15 on large core machines some cache corruption. Kashif noted that with CVMFS-nagios plugin, some WNs showing output error detected. May be aftereffect of the problem. Elena suggested emailing cloud support/raising at ATLAS meeting on Thursday, Kashif said he would follow up with cloud support.
- Other
Chris with Kashif/Jeremy/Ewan:
Sent out UI recommendation to update to VOMS, I think most of the configuration is done for that. Please report any problems. News from Janusz Operation Bulletin has DIRAC supported VOs (Mice NA62 Londongrid SNO+ Landslides). Request for something we can use, obvious ones are dteam or gridpp. Jeremy thought there was an agreement before for gridpp.
Busy with CHEP stuff this week, hopefully in the not too distant future. Jeremy; should track which sites have made updates to UI. Kashif: sites failing tests on vo-nagios, wondered who should raise tickets? Chris suggests looking at link so that sites could look at failing tests. Chris was suggesting updating VOMS information in any case. Kashif posted vo-nagios link:
https://vo-nagios.physics.ox.ac.uk/nagios/
Meetings & updates (20')
========================
Jeremy with others:
With reference to: http://www.gridpp.ac.uk/wiki/Operations_Bulletin_Latest
There was a request to approve ipv6.hepix.org VO. Chris wanted to run it by ops before adding to approved list. Jeremy noted procedure for VO being added to approved list (running by ops team and looking at possible additional requirements). Jeremy asked who hosts this - Chris noted that it's hosted out of CNAF. Steve noted that quite a lot of info on VO card. Ewan wanted to clarify where the ipv6 VO would be enabled, presumably just ipv6 kit? Chris agreed, but noted that the request here was to add it to the approved list. Discussion of this being a special case where it was a testing VO, not production.
Matt wondered if this VO is sending tests to anyone that supports it? Mark noted that it depends - some specific data transfer tests, some general job tests. He also noted that RAL is not IPv6 enabled; Jeremy said this was being followed up, not a current priority but would be shortly.
T2 A/R: Some sites that needed to respond based on the last report. VO derived figures - discussion at GDB around the fact that jobs might have different priority and so could be timing out. Hence discussion around different priority for ops/vo tests. Direct link in transcript. Look at these closely at the start of February.
In kind contributions: still waiting on input from some sites, thanks to those that followed up.
WLCG Ops co-ordination. Virtual meeting. Same topics as usual. No updates from ATLAS/LHCb as they had other meeting. glexec/ipv6 validation ongoing. Next meeting for Middleware readiness in Feb 6. Multicore TF meeting this morning. perfsonar TF ongoing. SHA-2 migration - we know OSG move in mid feb. France moved last week, UK March. Tracking tools ongoing. WMS decommissioning ongoing and xrootd ongoing. Chris asked if pilot/power users moved first and did debugging? Jeremy noted that this was already happening.
- Tier-1 status
Jeremy (on behalf of Gareth):
Operations have generally been smooth. The tape system is down this morning (21st Jan) for an update to the microcode in the tape libraries. Both disk and one CPU tranch now delivered. Last major item (second CPU tranch) being delivered this week.
We are working on the major upgrades to the Tier1 network and plan to have this set-up within around 6 weeks.
- Storage and Data Management
Sam didn't think there was anything critical from the last meeting; quarterly reports, Sheffield's weird LHCb transfer issues. Brian noted the new StoRM version.
- Documentation
Keydocs out of date - discuss this further at Core Ops.
- Interoperation
Updates in Bulletin, Updates for Gridsite and Globus gatekeeper as part of openssl , Gridsite in staged rollout. Flagged the webdav StoRM update 1.11.3 from December
Releases: only major release ARC 4.0.0. SHA-2 update: 99% of services completed. No discussion of NGIs moving to SHA-2 certs. Flagged that the clients are not monitored, but for dcache-srm-client the first version supporting SHA-2 certificates is v2.2.22.
Decommission of CERN WMSes : Planned for April, it was noted by service managers that there was a high usage for ops tests. Mentioned the UKs trial of RPMs for VOMS data; this was favourably received so it was indicated that we'd update in a few weeks with more experience.
- Monitoring
One thing to pull out from last meeting - new probes being written to remove use of WMS - ALICE/LHCb CREAM probes being rolled out Jan/Feb, ATLAS/CMS CondorG probes being rolled out after that.
Jeremy: might be worth looking at this in a bit more detail in another meeting.
- On-duty
Gareth Smith on duty last week. Sussex have put their CE into a downtime. Cleared up some tickets after they did this. The APEL alarms at Brunel reappeared from time to time. Might want to explore publishing more fully at another meeting. Closed these referring to a ticket Daniela raised. See:
https://ggus.eu/ws/ticket_info.php?ticket=100287.
- Security
Maarten gave a good update on openssl last week.
- services
Made clear that BNL perfsonar dashboard isn't being updated. Looking at site dashboards don't see much change over last week although indication of a lot more loss. Are people finding PerfSonar useful? Chris found it useful in past, haven't looked more recently. Raja: part of metrics used to maintain T2D status in UK, set up by Andrew. Jeremy notes glexec test becoming critical for CMS. Ewan Steele noted PerfSonar very useful at Durham with slow connection. Graphs very useful as they show flat line in one direction.
- Tickets
Matt:
EFDA-JET
https://ggus.eu/ws/ticket_info.php?ticket=97485 (21/9/2013)
LHCB jobs failing at Jet. The Jet chaps have just fixed an SSL problem at their site, so would like to see if this has fixed the LHCB problems. Waiting fore reply (20/1) Update - things are still failing, reading the error perhaps JET have picked up some wierd rpms somewhere?
(This also possibly solves the Jet gLeXeC ticket https://ggus.eu/ws/ticket_info.php?ticket=95295 UPDATE-SOLVED, the Jet guys put in a fix to JAVA to solve the keysize problem and things work now )
UCL
https://ggus.eu/ws/ticket_info.php?ticket=100342 (16/1)
Atlas are seeing transfer failures to/from UCL's dpm. Looks like an authentication problem, Ben might need a hand as we only have a bit of his time. In progress (20/1)
TIER 1
https://ggus.eu/ws/ticket_info.php?ticket=100333 (16/1)
Looks like this problem Tom and Chris spotted with one of the RAL WMSii has been solve, case can be closed. In progress (17/1) SOLVED
https://ggus.eu/ws/ticket_info.php?ticket=100343 (16/1)
But the WMSses still bring us pain, here Chris documents that the RAL ones are still producing 512-bit proxies. Chris also helpfully links two other WMS tickets. In progress (17/1)
Chris noted that when proxy is renewed that you get 512-bit proxies. Matt looked up ticket; possible that not all people affected. Jeremy noted that this could be followed up.
https://ggus.eu/ws/ticket_info.php?ticket=98122 (17/10/2013)
But Tom provides another win, this time with the cern@school cvmfs repo. He's managed to get it working, able to put data into it, so this ticket can probably be closed too. In progress (17/1) SOLVED
https://ggus.eu/ws/ticket_info.php?ticket=100114 (8/1)
But then the WMS try to spoil our buzz again with another ticket. Although I believe this is the forerunner to 100343 above. In progress (16/1)
BRUNEL
https://ggus.eu/ws/ticket_info.php?ticket=100188 (10/1)
Raul has provided Brian with the database dump from his SE (it should have landed in Brian's inbox), I think this ticket can be closed if the dump looks alright. In progress (16/1)
BRISTOL
https://ggus.eu/ws/ticket_info.php?ticket=99910 (20/12/2013)
LHCB problems at Bristol, due to ARC doing strange things to the environment. A few brave fixes have been tempted, but no joy. Waiting on feedback from the ARC developers - if that takes a while this ticket will need to be On Holded. In progress (14/1)
ECDF
https://ggus.eu/ws/ticket_info.php?ticket=99794 (16/12/2013)
Poking holes in the Edinburgh firewall for the perfsonar box. Any news from the IT overlords? On hold (13/1)
Andy would check into it and update the ticket.
GLASGOW
https://ggus.eu/ws/ticket_info.php?ticket=98253 (21/10/2013)
The "getting CMS working at Glasgow" ticket. It's looking almost as neglected as my gym membership. On hold (16/12/2013)
MANCHESTER
https://ggus.eu/ws/ticket_info.php?ticket=97066 (5/9/13)
Getting the Manchester perfsonar boxes back up and running. How goes it? On hold (7/1)
SHEFFIELD
https://ggus.eu/ws/ticket_info.php?ticket=98594 (4/11/2013)
The LHCB job uploading problem at Sheffield. It seems all parties have gotten stuck, so we need to decide where to go with this. On hold (8/1)
Jeremy noted that Elena would come back to this shortly.
DURHAM
https://ggus.eu/ws/ticket_info.php?ticket=99621 (10/12/13)
Just making sure this ticket, with a bad node needing offlining, isn't forgotten about. On hold (19/12)
Similar with the Durham GLEXEC ticket https://ggus.eu/ws/ticket_info.php?ticket=95302 - it was On Holded over Christmas, but Christmas was a while ago now. In fact, with Creme eggs out, it must be nearly Easter already... right?
EXTRA EXTRA
RALPP https://ggus.eu/ws/ticket_info.php?ticket=100401 (20/1) This nagios glexec alarm ticket which Chris quickly jumped on has been reopened on you guys. Just bringing it up as reopened tickets have a habit of sneaking under the radar. Reopened (21/1)
OXFORD
https://ggus.eu/ws/ticket_info.php?ticket=100348 (17/10) Atlas are getting a little ansy for some news on this ticket. And also don't seem to understand the waiting for reply state is for... Waiting for reply (21/1)
(Kashif mentioned this earlier in the meeting)
- Tools
Pete wanted to flag the HS06 benchmarking, with a request that sites update their benchmarking following SL6 updates on the wiki if they have not yet done so. New HS06 value should be reported in Q4 2013 reports.
Affected sites:
Brunel, UCL, Lancaster, Liverpool, Manchester, Sheffield, Durham, ECDF, Glasgow, Birmingham, JET, RALPP, RAL.
Jeremy noted that might be worth ticketing in due course. Sheffield have benchmarked but not updated. Lancs benchmarked but didn't update the wiki.
- VOs
- SIte updates
11:40
GDB review (15')
- GDB agenda http://indico.cern.ch/conferenceDisplay.py?confId=272795
- There was a pre-GDB on cloud issues: http://indico.cern.ch/conferenceDisplay.py?confId=272783.
Andrew:
(see links above and on meeting page)
pre-GDB Cloud discussion: A couple of things about general direction opf accounting. Question of wall-clock/cpu-clock. Complicated discussion on how to measure/scale this. Imagine most sites would want wall clock - scaling? Discussion complicated vs simple solution. Glexec within VMs, vs sudo? Fair-share in cloud based installations. Simple vs complicated again.
(with reference to https://indico.cern.ch/getFile.py/access?contribId=4&resId=1&materialId=slides&confId=291923)
Possible topics discussed for next meeting, nothing final.
Experiments considering instrumenting their pilot frameworks rather than relying on prioritising some of their jobs. Ops tests priority/expt testing.
Accounting: 2 talks about APEL-3. John Gordon involved temporarily due to staff changes. Under control.
Volunteer computing: BOINC. Might be useful for other computing facilities on campus. Problems with LHC@home using VMs, doesn't work with Windows protected environment, can't be running in the background. Now a way round this, might be interesting and useful for sites to extend resources.
Detailed ops co-ordination report. VO SAM tests, going over to using VO ones. Lots of status reports. 12 pages of TF updates, middleware deployment, SL6, glexec, perfsonar, SHA-2, IPv6, WMS decommissioning , Savannah -> JIRA, Machine/Job Features, Multicore. Also glexec TF, more interesting to us, details in the talk. 68 tickets are closed, 27 site tickets are open. ALICE and ATLAS still need dev work. LHCb testing pilots. CMS almost done. Aiming to put glexec test critical in Jan for CMS.
Openssl talk. To get things to work, need 1024-bit proxies. Culprit turns out to be Gridsite, it gets to choose what size of proxy. Sites need to install this. Interesting to see how this came out. No central person
EGI News. Verification of security contacts. Send test emails to security contacts requiring clicking on a URL to make sure that email works. Thinking about extending this to other contacts. Noticed that some GOC emails have spam traps in them, might be worth looking at contact details in the goc. Proposing changing A/R thresholds, 70%-80% Availability, 75-85 Reliability (for EGI).
HEPSPEC benchmarks. Going to do repeat of what was done with HS06 - use part of SPEC to represent HEP workload, following SPEC14 this year. Might involve a lot of work at the beginning of next year once SPEC14 is finalised but other work can be done beforehand.
Status reports. Not directly related to Horizon 2020. Nothing really stood out from an operational standpoint. Jeremy: May 2014, intend to move testbed into production, federated cloud services.
Davix: http toolkit. Put together lots of extensions for things that people had been talking about and put an interface in front. Looked like a complete, credible storage access system. Final pieces to making http an alternative to the kind of things we're using xrootd for.
Jeremy: postscript from John Gordon, the GOC v5.2 which takes into account with Glue 2 put key values which means to can use custom key properties. Definition of class of sites, can use goc to group things in a way we want to. The other thing was the new benchmarks, there is a WLCG validation team (hepix), other non-WLCG HEP VOs that should be part of this (t2k, eg).
Mark: Opportunistic CPU - standardised model? Andrew - uses BOINC, there are versions for Linux,Windows and Mac, they use Virtual Box now. You run that VM, in LHCb we use the same VM that we use for clouds and vac for this. The downside is that you have to have the cooperation of the person in front of the keyboard, which is a factor in a student cluster, eg. If it can be run as a service, can wake up based on a policy (disk activity, eg), wake up and start up a linux VM and can do the same things as we can on any kind of SL6 machine - potentially could access local storage.
Jeremy: One of the issues is local caching, cvmfs. Andrew noted that they rely on squid caches, machines have to be able to find a local squid cache. One option is to run a squid within the VM.
====
Jeremy: Talking to Janusz, we'd like gridpp to be enabled on DIRAC, (see also transcript) - also documentation.
Action list
===========
With reference to: https://www.gridpp.ac.uk/wiki/Operations_Team_Action_items
Jeremy highlighted some actions:
All sites to query the IPv6 status of their respective institutions: Review this in February about where sites Are
Explore possibilities for LFC backup at a Tier-2: LFC Backup put at another RAL site, perhaps Daresbury, Jeremy to check on the status of that.
Plan out the future of CE/Batch System integration. Torque/maui are not supported by EGI. Layout an agenda with proposals.: This is a concern, mentioned at GDB. Plan to have a pre-GDB to look at batch system issues, also in hepix. Ongoing in terms of a solution but Torque/Maui are going away.
Explore LFC/SE consistency checking status and develop tools if needed.: Sam: Moving along at low priority. It's part of the small VOs policy stuff that has been moving along for storage that Chris has written a wiki page on, but Sam has never felt that LFC/SE consistency checking was important enough to the small VOs to write a tool for it. Sam wonders if we wrote a tool (which Sam agrees with Chris would be useful) whether it would be used. Note from Tom that cern@school isn't at a point to use a LFCtool, but maybe in the future. Order of magnitude file usage ~hundreds of files (size GBs, maybe TBs in the next year). Jeremy: may be a need, perhaps check with other VOs. Jeremy to update action item.
Following up on benchmarking of sites.
Progression to using Backup VOMS.
Document how to use robot certificates to perform routine tasks without regular human intervention.& Document how to renew a server's grid certificate without loading it into a browser. : Jens joining meeting next week to go through some issues, particularly around certificates so can follow up then.
AOB
===
T2 Quarterly reports are now due. Drafts going round, waiting for some updates, if those could be addressed in coming days that would be useful. Also a reminder about the in kind contributions.
Mark: Requested any comments to add to the networking section of proposal. Pete, Duncan and Mark discussed to what extent sites were connected at 10G to JANET; Mark noted that no one has dedicated bandwidth, all metro areas have been updated to 40G. 10G is the highest campus connection at this time. Pete wanted to check into what the average site connection was. Mark; majority of sites are connected at 10G, on a shared environment in JANET. Some sites at 1G, could be part of future development.
Brian and Mark discussed the experiments expectations with respect to bandwith.
Jeremy noted that this was a summary of the current position.
Tom checked about whether there were any news items; Stephen Jones' Instant UI tool picked up at EGI recently.
Meeting closes.
Transcript
==========
[10:53:10] Gang Qin joined
[10:53:19] Brian Davies joined
[10:53:23] Elena Korolkova joined
[10:53:31] Robert Frank joined
[10:53:39] Gareth Roy joined
[10:53:42] Robert Frank left
[10:54:19] RECORDING David joined
[10:56:41] John Bland joined
[10:56:47] Rob Fay joined
[10:57:46] Jeremy Coles joined
[10:58:09] Christopher Walker joined
[10:58:30] Steve Jones joined
[10:58:41] Mohammad kashif joined
[10:58:43] Matt Doidge joined
[10:58:56] Jeremy Coles David is taking minutes today.
[10:59:06] Mark Slater joined
[10:59:19] Sam Skipsey joined
[10:59:55] Daniel Traynor joined
[11:00:01] Andrew McNab joined
[11:00:06] Gareth Smith joined
[11:00:11] Daniela Bauer joined
[11:00:11] Pete Gronbech joined
[11:00:22] IPPP1 UofDurham joined
[11:00:26] John Hill joined
[11:00:44] Mark Mitchell joined
[11:00:49] Raja Nandakumar joined
[11:01:08] Tom Whyntie joined
[11:01:51] Ewan Mac Mahon joined
[11:02:58] Elena Korolkova left
[11:03:04] Elena Korolkova joined
[11:03:28] Matt Williams joined
[11:04:16] Matt Williams left
[11:04:58] Duncan Rand joined
[11:05:01] Andrew Washbrook joined
[11:06:44] Chris Brew joined
[11:07:24] Matt Williams joined
[11:07:56] Matt Williams left
[11:08:53] Linda Cornwall joined
[11:09:32] Ian Loader joined
[11:09:42] Matt Williams joined
[11:09:51] Ewan Mac Mahon So where exactly did you tweak the limit?
[11:11:37] Gareth Roy /etc/cvmfs/default.local
[11:11:39] Gareth Roy CVMFS_NFILES=131072
[11:11:46] Robert Frank joined
[11:12:09] Gareth Roy is what we have at the moment
[11:12:12] Ewan Mac Mahon Thanks.
[11:12:39] Elena Korolkova for LHCB problem in Sheffield : the mid of September we started to move to sl6.
[11:13:15] Ewan Mac Mahon For gridpp, probably Chris.
[11:13:31] Ewan Mac Mahon It's a bit less wide open than dteam would be.
[11:15:14] John Bland elena: do your SL6 nodes still have the TCP buffer tweaks for WAN transfers?
[11:15:24] Mohammad kashif https://vo-nagios.physics.ox.ac.uk/nagios/
[11:16:24] Raul Lopes joined
[11:16:33] Duncan Rand left
[11:16:53] Jeremy Coles https://www.gridpp.ac.uk/wiki/Operations_Bulletin_Latest#
[11:17:17] Brian Davies Telling me I have to be a mameber of southgrid vo to access that webpage kashif
[11:17:44] Matt Williams System Message to user Matt Williams: SeeVogh has detected significant packet loss in your incoming video stream.
[11:17:46] Matt Williams System Message to user Matt Williams: SeeVogh has reduced the quality of your incoming video to optimize your performance.
[11:18:01] Elena Korolkova to John: yes, we have. I'll appreciate if you can send me what do you have in Liv so I can compare.
[11:19:42] Duncan Rand joined
[11:20:13] Jeremy Coles https://voms2.cnaf.infn.it:8443/voms/ipv6.hepix.org/admin/home.action
[11:20:14] John Bland elena: email sent
[11:20:29] Matt Williams System Message to user Matt Williams: SeeVogh has detected significant packet loss in your incoming video stream.
[11:20:29] Matt Williams System Message to user Matt Williams: SeeVogh has turned off your incoming videos due to poor network performance.
[11:20:59] Steve Jones Dave Kelsey is named as VO Manager
[11:21:08] Steve Jones Hosted at CNAF
[11:22:25] Steve Jones Scope is Global.
[11:25:03] Jeremy Coles http://sam-reports.web.cern.ch/sam-reports/2013/201312/wlcg_vos_unnoficial/
[11:26:28] Mohammad kashif Brian, I have added you to ACL list now. By default it add all members of VO being monitored.
[11:27:23] Steve Jones Well put a report together of estimates.
27:24] Christopher Walker Can we allow anyone in dteam?
[11:27:29] IPPP1 UofDurham we have figures coming but its taking a little while to collate
[11:27:32] Christopher Walker That would cover all the site admins
[11:27:47] Steve Jones I mean "we'll put a report together"!
[11:27:52] Jeremy Coles https://twiki.cern.ch/twiki/bin/view/LCG/WLCGOpsMinutes140116
[11:28:41] Mohammad kashif Chris, good idea, I will add all members of dteam
[11:29:38] Mohammad kashif If some one is not any category then I can add them manually. Please let me know.
[11:31:39] Gareth Smith Thanks Jeremy. Nothing to add.
[11:32:22] Christopher Walker The problems don't seem to be stability - it's performance under load.
[11:32:34] Elena Korolkova @Kashif: I'm not in dteam. Can you add me, please.
[11:32:36] Christopher Walker Also applied to the previous storm version
32:48] Christopher Walker Elena - you should join dteam anyway.
[11:33:55] Steve Jones ipv6.hepix.org: https://www.gridpp.ac.uk/wiki/GridPP_approved_VOs#Approved_Global_VOs
[11:34:16] Steve Jones Thanks for that feedback.
[11:37:35] IPPP1 UofDurham yes
[11:39:22] Duncan Rand you might find this useful too http://maddash.aglt2.org/maddash-webui/index.cgi?dashboard=UK%20sites
[11:42:08] Mohammad kashif Elena, you are already part of nagios ACL becuase you are a member of t2k
[11:42:38] Tom Whyntie Thank you!
[11:43:38] Tom Whyntie Wiki page here if anyone's interested: https://www.gridpp.ac.uk/wiki/CERN%40school_and_CVMFS
[11:45:52] Elena Korolkova Thanks, Kashif
[11:50:16] Elena Korolkova In Sheffield we have done benchmarking
[11:50:45] Elena Korolkova but we haven't update yet
51:19] Ewan Mac Mahon https://www.gridpp.ac.uk/w/index.php?title=HEPSPEC06&action=edit§ion=13
[11:51:45] Mark Slater Bham: I don't think we've done the Wiki but I believe Matt is on it
[11:52:07] Ewan Mac Mahon Lancashire have figures recorded on SL4.6, apparanetly.
[11:53:41] Jeremy Coles The slides on the agenda are my own notes from the meeting... Andrew is using his own notes.
[11:54:04] Ewan Mac Mahon With a clock, on a wall?
[11:56:03] Jeremy Coles Alessandra will be the T2 rep in February. Chris in March.
[11:58:26] Mark Slater left
[11:59:34] Matt Williams System Message to user Matt Williams: SeeVogh has detected significant packet loss in your incoming video stream.
[11:59:36] Matt Williams System Message to user Matt Williams: SeeVogh application will connect to another server to see if it improves
[12:01:52] Steve Jones Will this be an on-going problem? Will min key size move to 2K, then 4K etc. Or is 1K enough "forever"?
[12:02:28] Jeremy Coles That was mentioned and people asked if we should move to longer keys. The concern was that would cause other things to break!
[12:03:01] Steve Jones We should use "medium length" keys - not too long, not too short.
[12:03:14] Janusz Martyniak joined
[12:03:19] Daniela Bauer So they want me to click on a link in a random email ???
[12:04:19] Ewan Mac Mahon That thought occured to me to. Any 'Please click this linlk to verify your email' type thing is going to look pretty phishy.
[12:05:15] Matt Williams System Message to user Matt Williams: SeeVogh has detected significant packet loss in your incoming video stream.
[12:05:15] Mark Mitchell As long as you dont supply your bank details...
[12:05:17] Matt Williams System Message to user Matt Williams: SeeVogh has reduced the quality of your incoming video to optimize your performance.
[12:06:43] Gareth Smith left
[12:07:55] Christopher Walker gfalfs allows you to use the Davix stuff to mount srm or webdav as a filesystem
[12:09:22] Christopher Walker I've filed a number of bugs...
[12:09:30] Christopher Walker But it looks interesting.
[12:09:59] Andrew Washbrook No he is not here
[12:10:06] Andrew Washbrook (Wahid)
[12:10:14] Jeremy Coles Ok thanks.
[12:14:18] Janusz Martyniak i' unable to speak ...
[12:14:26] Janusz Martyniak
[12:14:39] Janusz Martyniak gridpp VO is enabled
[12:14:56] Ewan Mac Mahon Yay, that's a very good thing.
[12:15:17] Ewan Mac Mahon But we could do with some 'how to' type instructions on accessing it starting from a normal UI.
[12:15:41] Jeremy Coles https://www.gridpp.ac.uk/wiki/Operations_Team_Action_items
[12:15:41] Janusz Martyniak ther is an 'official' dirac doc ...
[12:18:00] Matt Williams System Message to user Matt Williams: SeeVogh has detected significant packet loss in your incoming video stream.
[12:18:00] Matt Williams System Message to user Matt Williams: SeeVogh has turned off your incoming videos due to poor network performance.
[12:19:31] Ewan Mac Mahon From what I recall of the official dirac docs, they're generally quite good, but the 'how to configure your client' stage mostly assumed that someone would have given you a suitable config file that you could just point your client at to set everything up.
[12:19:40] Ewan Mac Mahon Do we have such a config file somewhere?
[12:20:08] Andrew Washbrook left
[12:20:46] Steve Jones How much work is it to write a tool?
[12:20:58] Steve Jones If it's easy, just do it.
[12:21:00] Sam Skipsey that depends on how fast you want it to be, Steve
[12:21:14] Steve Jones Ah...
[12:21:15] Christopher Walker Complete a consistency check in 2 years
[12:21:34] Sam Skipsey (the trivial implementation suffers from the fact that the LFC isn't really designed to have fast metadata lookups on directories)
[12:21:35] Christopher Walker Sorry, less than two years.
[12:22:04] Sam Skipsey but, sure, I'll write the trivial version, and we can at least benchmark it
[12:22:13] Christopher Walker QMUL has 9333 files for snoplus and 266195 for t2k
[12:22:32] Christopher Walker There's a recursive ls in the gfal2 svn
[12:22:44] Christopher Walker That gets you most of one way.
[12:23:11] Sam Skipsey That's the SRM recursive ls?
[12:23:54] IPPP1 UofDurham we're not
[12:24:02] Christopher Walker Put some impressive numbers in about how much data we transfer
[12:24:54] Andrew McNab IPv6 is something can could stress that we can test between institutions for real work, since we're a small interlinked community that does a lot of work though
[12:24:59] Ewan Mac Mahon Big up how much we're blazing a trail that others are following; it's all about spinning us as valuable to people that aren't us, not just making us look awesome in our own right.
[12:25:17] Christopher Walker From the High throughput networking sig, we are much more capable of doing this than anyone else in the UK.
[12:25:25] Ewan Mac Mahon The key thing is awesome AND useful.
[12:25:36] Christopher Walker QMUL is peaking at 10Gig now.
[12:26:10] Pete Gronbech Although we are not driving it very hard at the moment
[12:26:36] Ewan Mac Mahon I expect that will change when the boatload of empty new disk comes online.
[12:26:36] Matt Doidge Lancaster's Gridsite is routed over the Universitiy's backup link, so we usually have 10Gb all to ourselves.
[12:28:08] Ewan Mac Mahon The VOs never know what their expectations are.
[12:28:12] Ewan Mac Mahon They never have.