Alert.png The wiki is deprecated and due to be decommissioned by the end of September 2022.
The content is being migrated to other supports, new updates will be ignored and lost.
If needed you can get in touch with EGI SDIS team using operations @ egi.eu.

Difference between revisions of "Agenda-05-12-2016"

From EGIWiki
Jump to navigation Jump to search
Line 69: Line 69:
* [https://indico.egi.eu/indico/event/2810/contribution/3/material/0/ ARGO Proposal (September OMB)]
* [https://indico.egi.eu/indico/event/2810/contribution/3/material/0/ ARGO Proposal (September OMB)]
* [https://indico.egi.eu/indico/event/2814/contribution/6/material/slides/ ARGO] and [https://indico.egi.eu/indico/event/2814/contribution/7/material/slides/ GOC-DB] updates from November OMB
* [https://indico.egi.eu/indico/event/2814/contribution/6/material/slides/ ARGO] and [https://indico.egi.eu/indico/event/2814/contribution/7/material/slides/ GOC-DB] updates from November OMB
* The plan:
 
**Develop new features on GOC-DB and ARGO:
*Timescale:
***GOC-DB: create the new�boolean attribute "Monitored" on the ServiceEndpoints
**New GOC-DB release on Dec 7th including a boolean ‘monitored’ flag for the service endpoints
***ARGO: change the GOCDB connectors in order to take into account also this attribute
**Then creation of a web UI view for uncertified sites in ARGO  
**Start with the uncertified sites, then all the others
**Uncertified sites will be asked to fill in the service endpoints information. Follow the [https://wiki.egi.eu/wiki/HOWTO21 How to add URL service endpoint information into GOC-DB]
**Interim period: consuming both BDII and GOC-DB
***('''OPTIONAL''') use the [https://gocdb-test.esc.rl.ac.uk/portal/index.php GOC-DB test instance] for testing the procedure
***If no issues, only GOC-DB will be kept as topology source
**As information is added in the GOCDB, uncertified sites/services will be picked up by the ARGO Monitoring Engine and they will start to be monitored
* Proposed timescale:
** By Q2 2017: support for multiple service endpoints
**November
***Complete developments on the ARGO/GOCDB sides
***New report will be configured on ARGO for uncertified sites
**December:
***Uncertified sites will be requested to update their information on GOCDB
***As information is added in the GOCDB, uncertified sites/services will be picked up by the ARGO Monitoring Engine and they will start to be monitored
**End of February:
***Present at the OMB the results of the pilot and decision for the next step regarding the production infrastructure
* For site-admins: use the [https://gocdb-test.esc.rl.ac.uk/portal/index.php GOC-DB test instance] for testing the procedure
** Follow the [https://wiki.egi.eu/wiki/HOWTO21 How to add URL service endpoint information into GOC-DB]


== VAPOR ==
== VAPOR ==

Revision as of 13:42, 2 December 2016


General information

UMD/CMD/Preview

Preview repository

  • Released on 2016-11-25:
    • Preview 1.6.0 AppDB info (sl6): ARC 15.03 update 10, frontier-squid 3.5.22-2.1, gfal2 utility 1.4.0, XRootD 4.5.0
    • Preview 2.6.0 AppDB info (CentOS 7): ARC 15.03 update 10, dCache 2.16.18, frontier-squid 3.5.22-2.1, gfal2 utility 1.4.0, XRootD 4.5.0, UI/WN tarball.

Note: EGI provides the preview repository without any additional quality assurance process, but the products are released as they are provided by the product team. EGI recommends the use of the UMD repositories, which contain software verified through the quality assurance process of UMD.

Operations

Software upgrades for OpenStack cloud RCs

  • keystone-VOMS and cloud-info-provider updates available, need to be installed on all OpenStack sites
  • as keystone-VOMS last version is only compatible with Liberty and Mitaka, in case OpenStack is Kilo (or older) an upgrade plan of OpenStack has been asked
  • according to EGI policies https://wiki.egi.eu/wiki/PROC16_Decommissioning_of_unsupported_software OpenStack Kilo or older should NOT be running on the infrastructure! we are asking for discussing this point at the next OMB (October 27)
  • as many sites are finding difficulties in planning upgrades against the very tight release cycle of OpenStack, please come with suggestion and reply with details in the tickets in order to shape the best (shared) proposal
  • ticket campaign ONGOING for all OpenStack sites, asked to upgrade to keystone-VOMS >=8.0.3, cloud-info-provider >=0.6, and plans for the future (OpenStack version currently deployed, plans for upgrades, usual specific RC upgrade schedule), some results:
      • TR-FC1-ULAKBIM FIXED, using FedCloud Appliance (updated version), using Liberty
      • IN2P3-IRES FIXED, using Liberty, upgrading to Mitaka
      • INFN-PADOVA-STACK FIXED, using Liberty, following Indigo indications about the version (staying with Liberty at the moment)
      • NCG-INGRID-PT, using Mitaka, up to date
      • IISAS-GPUCloud up to date, using Liberty, upgrading to Mitaka
      • FZJ using Kilo, not up to date, no plans for migrations due to user isolation bug https://ggus.eu/index.php?mode=ticket_info&ticket_id=121685 -> fixed in Newton, not in Mitaka (investigating about a backport to Mitaka)
      • BIFI using Grizzly, upgrading to Mitaka by mid of Nov
      • RECAS-BARI using Juno, no upgrades so far waiting for resolution of the user isolation bug, preparing upgrade to Mitaka from Ubuntu 16.04 LTS
      • IFCA-LCG2 FIXED, using Liberty
      • IISAS-FedCloud, downtime, upgrading to Mitaka from Ubuntu 16.04 LTS
      • SCAI FIXED, using Kilo, planning short term upgrade (no version specified yet)
      • CETA-GRID using Icehouse, planning mid-term upgrade
      • INDIGO-CATANIA-STACK, INFN-CATANIA-STACK, CYFRONET-CLOUD didn't reply yet
  • feedback/issues reported from sites:
    • setup a pre-production environment that replicates the production environment and we use this testbed to implement/test/verify the upgrade procedure
    • suggested pilot RCs provide and share docs and guidelines when approaching to EOLs of CMFs (NCG-INGRID-PT)
    • New OpenStack versions are often buggy, need some grace time before upgrading
    • Unavailability of EGI Components for new versions
    • OpenStack shared with other projects and integrated with other tools
    • User identity auth issue still unsolved https://ggus.eu/index.php?mode=ticket_info&ticket_id=121685

Monthly Availability/Reliability

  • November A/R figures not definitive yet
  • Underperformed sites in the past A/R reports with issues not yet fixed:
    • NGI_DE GGUS 123836
      • TUDresden-ZIH: set-up a new CREAM-CE, the CA probes were failing. Issues on SRM service
    • NGI_IT: GGUS 123531
      • INFN-CAGLIARI: underperforming for more than 3 months, no further feedback provided, eligible to suspension
    • NGI_IBERGRID: GGUS 124371
      • CIEMATIC-TIC: decommissioning the storage element
    • NGI_NL: GGUS 123532
      • BelGrid-UCL: UNKNOWN status returned by CREAM probes, asked a recomputation
  • Sites suspended after October A/R report:
    • UNI-DORTMUNT (NGI_DE): downtime longer than a month
    • TW-eScience (AsiaPacific): long standing issues

ARGO proposal to use GOCDB as the only source of topology information

  • Timescale:
    • New GOC-DB release on Dec 7th including a boolean ‘monitored’ flag for the service endpoints
    • Then creation of a web UI view for uncertified sites in ARGO
    • Uncertified sites will be asked to fill in the service endpoints information. Follow the How to add URL service endpoint information into GOC-DB
    • As information is added in the GOCDB, uncertified sites/services will be picked up by the ARGO Monitoring Engine and they will start to be monitored
    • By Q2 2017: support for multiple service endpoints

VAPOR

AOB

Next meeting

Retrieved from "https://wiki.egi.eu/w/index.php?title=Agenda-05-12-2016&oldid=91935"