Week of 150504

WLCG Operations Call details

• At CERN the meeting room is 513 R-068.

• For remote participation we use the Vidyo system. Instructions can be found here.

General Information

• The purpose of the meeting is:
  • to report significant operational issues (i.e. issues which can or did degrade experiment or site operations) which are ongoing or were resolved after the previous meeting;
  • to announce or schedule interventions at Tier-1 sites;
  • to inform about recent or upcoming changes in the experiment activities or systems having a visible impact on sites;
  • to provide important news about the middleware;
  • to communicate any other information considered interesting for WLCG operations.
• The meeting should run from 15:00 until 15:20, exceptionally to 15:30.
• The SCOD rota for the next few weeks is at ScodRota
• General information about the WLCG Service can be accessed from the Operations Web
• Whenever a particular topic needs to be discussed at the daily meeting requiring information from site or experiments, it is highly recommended to announce it by email to wlcg-operations@cernSPAMNOTNOSPAMPLEASE.ch to make sure that the relevant parties have the time to collect the required information or invite the right people at the meeting.

Monday

Attendance:

• local: Ian (IT-PES), Luca (IT-DB), AlessandroDiG (ATLAS), Maarten (ALICE), Xavi (IT-DSS+SCOD)
• remote: Felix (ASGC), Lisa (FNAL), IN2P3 (Rolf), Sang-Un (KISTI), Dimitri (KIT), Ulf (NDGF), Onno (NL-T1), Rob (OSG), Pepe (PIC),

Experiments round table:

• ATLAS reports (raw view) -
  • Central Services
    • FTS server - client communication issue, timeouts, is affecting ATLAS badly. There was a request from FTS3 devs to upgrade FTS servers, can FTS3 service managers clarify?
      • Upgrade will be performed in PPS at RAL(Tue) and CERN (Wed) and put in PROD on Thursday.
    • RAL Frontier down, GGUS:113409 - solved, thks.

• CMS reports (raw view) -
  • This week is CMS and therefore almost for sure nobody can join the call
  • New xrootd re-director for Tier-0 name space in EOS seems to need a port opening in the fire wall, GGUS:113032 (at the very end)
    • Not clear why this causes trouble with files out side of the Tier-0 name space
  • Please send a mail to Christoph, if there is the need to followup something

• ALICE -
  • mostly high activity
  • KIT: team ticket GGUS:113438 opened on Sat because of network issues (solved)

• LHCb reports (raw view) -
  • Operations dominated by MC jobs and user analysis
  • T0:
  • T1
    • Problem to contact SARA SRM (GGUS:113324)

Sites / Services round table:

• ASGC: ntr
• BNL: np
• CNAF: np
• FNAL: SIR report submitted (unrecoverable data loss in dCache 15th/April)
• GridPP: np
• IN2P3: ntr
• JINR: np
• KISTI: ntr
• KIT: np
• NDGF (mail form Ulf): Atlas unevenly hit our tape write systems, causing one of them to overload. It got turned RO on Thursday evening to even it out, and that then fixed the problem during the weekend. FI_HIP_T2 for CMS is also seeing the FTS "upload-timeout-reupload-delete" issue in their logs, so it's not only Atlas. (It's quite easy to spot the pattern in the dcache billing log)
• NL-T1: ntr
• NRC-KI: np
• OSG: ntr
• PIC: ntr
• RAL: np
• TRIUMF: np

• CERN batch and grid services:
  • FTS fts3.cern.ch upgrade fts-server 3.2.32 -> 3.2.33 08:00 UTC Wednesday 6th May. A 2 or 3 minute downtime. This is the version running on the pilot for sometime.
• CERN storage services:
  • EOSATLAS namespace unavailable for about an hour (9h40 to 10h40). The reason was a corrupted entry on the namespace metadata.
• Databases: ntr
• GGUS:
  • Network issues at KIT on Monday at 14:30. The network experts are looking into it
• Grid Monitoring:
• MW Officer:

AOB:

• the SARA SRM access issue reported by LHCb may well be due to the same problem that has been causing intermittent proxy renewal failures on the ALICE VOBOXes at NIKHEF and SARA (GGUS:113240):
  • the CRL of the TERENA-eScience-SSL-CA-3 used by NLT1 service hosts has a lifetime of a few days (~1 week at the time of writing)
  • when it is downloaded, the corresponding HTTP header may advertise a cache validity that exceeds the CRL lifetime
  • fetch-crl < 3.0.16 will trust the advertised validities to avoid putting unnecessary load onto CRL web services
  • any CRL with a "short" lifetime might then expire before it gets downloaded again
  • fetch-crl 3.0.16 works around such mismatched promises
  • it currently is available in EPEL-testing:
    • EPEL-testing-6 repo
      • fetch-crl-3.0.16-1.el6.noarch.rpm
    • EPEL-testing-5 repo
      • fetch-crl3-3.0.16-1.el5.noarch.rpm
  • documentation:
    • https://dist.eugridpma.info/distribution/util/fetch-crl/
      • https://dist.eugridpma.info/distribution/util/fetch-crl/CHANGES

Thursday

Attendance:

• local: Joel (LHCb), Ian (PES), David (ATLAS), Maria (SDC), Pablo (Monitoring+GGUS), German (DSS), Maarten (ALICE), Andrea (MW), Andrei (DB), Xavi (DSS+SCOD)
• remote: Felix (ASGC), Michael (BNL), Sonia (CNAF), Lisa (FNAL), Rolf (IN2P3), Thomas (KIT), Ulf (NDGF), Dennis (NL-T1), Rob (OSG), John (RAL), Di (TRIUMF)

Experiments round table:

• ATLAS reports (raw view) -
  • Grid is running mostly full
  • Thanks to sites for upgrading FTS3 to address serious issue we had - we see much fewer submission timeouts
  • Computing Run Coordinator shift starts this week - he/she will report for ATLAS in WLCG daily meetings

• CMS reports (raw view) -
  • New xrootd re-director for Tier-0 name space in EOS still causing some problems GGUS:113032 (at the very end) - more likely a configuration issue than a firewall issue (as suspected on Monday)
  • Network problems in the US late Sunday until ~Tuesday
    • There were some issues in ESnet
  • Problems with FTS3 at FNAL last weekend, GGUS:113449
  • CVMFS: Stratum 1 on cvmfs.fnal.gov <http://cvmfs.fnal.gov> had not been updated since 48 hrs.
    • SLS status is “degraded” to 65%. SLS details show indeed a catalog time of 2015-05-02 21:00:51
    • FNAL issue seems to have recovered (was perhaps related to US network issue)

• ALICE -
  • high activity

• LHCb reports (raw view) -
  • Operations dominated by MC jobs and user analysis
    • T0:
    • T1
      • Problem to contact SARA SRM (GGUS:113324) no news from SARA
        • SARA reported they applied a workaround and will update the ticket.

Sites / Services round table:

• ASGC: DPM diskserver damaged reparation process ongoing.
• BNL: ntr
• CNAF: TAPE system in scheduled downtime next week for 3 days (12th to 15th of May)
• FNAL: ntr
• GridPP: np
• IN2P3: Plan to update dcache to the newer version, plan to do it on the 22nd of May.
• JINR: np
• KISTI: ntr
• KIT: Plan to update to update dCache next week: LHCb (Tue), ATLAS+CMS (Wed)
• NDGF:
  • We updated today to latest dcache 2.12, since the fix to 2.10, 2.11 and 2.12 got released on Tuesday.
  • The new version fixes a leak into chimeras t_tags_inodes table, where every upload via SRM leaves rows behind. At NDGF-T1 we had 300 million rows (40G data on disk), which made it extremely hard to delete without killing production from database IO-starving. It's important to update to stop the leak. There will be cleanup later, but the most important thing is to stop the leak. The update will also create a new index, which will take quite some time. Our cleaned DB with 17M rows took 1h 20 min to create the index.
• NL-T1: ntr
• NRC-KI: np
• OSG: ntr
• PIC: np
• RAL: FTS3 updated
• TRIUMF: ntr

• CERN batch and grid services: Experiencing issues with two CREAM-CEs, put into downtime in GOCDB while investigating.
• CERN storage services:
  • CASTORALICE upgraded to 2.1.15-6
  • Proposed upgrades for next week: CASTORATLAS (Monday the 11th), CASTORCMS and CASTORLHCb (Wednesday the 13th) announcements to be sent.
  • German (IT-DSS-TAB) announced a robot cleaning campaign during June: 17th and 24th.
    • Implications: writing to tapes is guaranteed but reading (recalls) might be delayed by about 8h if data is sitting on the affected robot. Days have been selected to make them coincide with the foreseen LHC technical stop (+ scrubbing).

• Databases:
• GGUS:
  • Question for our OSG colleagues: how should GGUS map contact lists from MyOSG.
• Grid Monitoring:
  • Draft availability reports for April sent, and available at the SAM3 UI
• MW Officer:
  • Asked for the FTS3 update at BNL, Michael replied that will be performed today.
  • Asked FNAL for plans to upgrade dCache version (currently running 2.2 with many later fixes backported): no plans on the short term.