General information
- the Operations meeting will be on the 2nd Monday of the month
- the EGI Operations Meeting schedule for first half of 2016 is available on Indico: https://indico.egi.eu/indico/categoryDisplay.py?categId=32 and on the new summary page: https://wiki.egi.eu/wiki/Operations_Meeting
News from URT
- A Critical bug which causes file loss.has been discovered on the DPM dmlite-shell new drain command released in DPM 1.8.10. One site in production has been affected https://svnweb.cern.ch/trac/lcgdm/wiki/Dpm/Dev/Dmlite/Shell#Newfunctionality:Drain
- broadcast sent on March 10th
- if you have run the new drain commands at your site, contact the DPM Development team through GGUS (data consistency check is needed)
- DO NOT use the new drain commands (documented at https://svnweb.cern.ch/trac/lcgdm/wiki/Dpm/Dev/Dmlite/Shell#Newfunctionality:Drain) and until the fixed components are released please continue to use the old dpm-drain command
- UMD 3.14.1 release notes updated
Staged rollout updates
- frontier-squid 2.7.24.2 (centos7)
- voms-admin 3.4.1 (sl6)
- storm 1.8.10 (sl6)
Next releases
Preview repository
On March 9th it was released the first update of Preview:
- STORM 1.11.10
- VOMS Admin server 3.4.2
- VOMS Server 2.0.13
- VOMS API Java 3.0.6, 3.1.0, 3.2.0
see details in https://wiki.egi.eu/wiki/Preview_1.1.0
Generic information about Preview repository: https://wiki.egi.eu/wiki/Preview_Repository
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.
Operational issues
Globus GSI clients moving to STRICT_RFC2818 by default
- the release of the update that will change the default name compatibility mode from "HYBRID" to "STRICT_RFC2818" is planned for April 1, 2016.
- EGI Broadcast sent in August already warning about the change, already advising "site managers to make sure that all the hostnames and aliases used to connect to a service are included in its host certificate Subject Alternative Name field, at the latest by the end of the year"
- sites that could be affected by this future change are the ones running services whose clients may use globus-gssapi-gsi for authentication (CE, FTS, SRM, GridFTP, MyProxy, WMS) and using DNS aliases which are not included within the SAN (Subject Alternative Name) field of the certificate (including the host name itself)
Aligning Fedcloud sites to the A/R procedures
- EGI Operations proposal to align Fedcloud sites to the A/R related procedures used for the grid sites
- based on the availability reliability of monitored services in cloudmon, EGI Operations will start follow up with underperforming sites as we are doing for every grid sites
- sites will NOT be suspended for a/r performance at least until end of May
- in parallel EGI Operations will start PROC08 to include cloud probes in the EGI_CRITICAL and EGI profiles used for A/R computations (IN PROGRESS)
The proposed timeline is:
- February 2016:
- EGI Operations will check the status of the production cloud services in order to understand which issues (if any) the site has and provide help to NGIs and sites;
- Start of the integration of cloud probes in EGI CRITICAL profile(current set+openstack): To be agreed with the ARGO team, PROC08 will be followed
- June 2016:
- Starting notification of sites eligible for suspension
FedCloud status
Old issues
Grouped by NGI, please follow up with sites.
- NGI_UK
- 100IT (OpenStack)
- vmcatcher issues https://ggus.eu/index.php?mode=ticket_info&ticket_id=116358#update#19 IN PROGRESS
- BDII and GOCDB have different Endpoint URLs https://ggus.eu/index.php?mode=ticket_info&ticket_id=119002#update#5 FIXED
- 100IT (OpenStack)
- NGI_PL
- CYFRONET-CLOUD (OpenStack)
- VMCatcher https://ggus.eu/index.php?mode=ticket_info&ticket_id=116363#update#29 IN PROGRESS
- CYFRONET-CLOUD (OpenStack)
- NGI_DE
- GoeGrid (OpenNebula)
- NGI_GRNET
- HG-09-Okeanos-Cloud (Synnefo)
- VMCatcher, issue with large metadata, on hold (it requires some development) https://ggus.eu/index.php?mode=ticket_info&ticket_id=116368 ON HOLD
- HG-09-Okeanos-Cloud (Synnefo)
- NGI_TR
- TR-FC1-ULAKBIM (OpenStack)
- Missing GLUE2DomainID and image description looks wrong https://ggus.eu/index.php?mode=ticket_info&ticket_id=119005#update#15 IN PROGRESS
- TR-FC1-ULAKBIM (OpenStack)
New issues
- New tickets opened to track issues in publishing appliances on AppDB for fedcloud.egi.eu: https://ggus.eu/index.php?mode=ticket_info&ticket_id=120010
- Issue with OCCI and fedcloud.egi.eu VO at MK-04-FINKICLOUD (NGI_MARGI): https://ggus.eu/index.php?mode=ticket_info&ticket_id=120027
Actions
- EGI Operations have been asked by user support to contact sites with unresolved technical problems in the support of the fedcloud.egi.eu VO since a long time
- if issues cannot be fixed quickly, sites will be asked to remove the support to fedcloud.egi.eu
- they will re-enable the VO support as soon as they are able to fix the issues
- sites will be contacted directly by EGI Operations
Getting help
- the whole FedCloud wiki has been reviewed, removing redundancies, updating links and instructions
- from the operations point of view: https://wiki.egi.eu/wiki/Federated_Cloud_resource_providers_support
- see in particular the manual for the installation of a cloud site
- TBD: review the support units associated with FedCloud (in progress)
Decommissioning Debian
- Debian support for squeeze (6.0) has been reached (Feb2016) https://www.debian.org/News/2016/20160212
- only one service published on BDII and production in GOCDB, but in GOCDB it is indicated as SL5.8, site is UA-MHI (NGI_UA)
dn: GlueSubClusterUniqueID=arc.hpc-mhi.org,GlueClusterUniqueID=arc.hpc-mhi.org ,Mds-Vo-name=UA-MHI,Mds-Vo-name=local,o=grid GlueHostOperatingSystemName: Debian GlueHostOperatingSystemRelease: 0 GlueHostOperatingSystemVersion: 0
Decommissioning SL5
- Tracked on SL5_retirement wiki
- No checks for dCache, DPM, ARC, UNICORE --> Action on NGIs/ROCs to follow up directly with sites
AOB
Monthly Availability/Reliability
A/R report on ARGO: http://argo.egi.eu/lavoisier/ngi_reports?accept=html
List of the underperforming RCs for (at least) 3 consecutive months:
- AfricaArabia https://ggus.eu/?mode=ticket_info&ticket_id=117094:
- EG-ZC-T3: unresponsive since months, must be suspended
- ZA-UJ
- AsiaPacific:
- MY-UM-SIFIR
- NGI_DE:
- LRZ-LMU
- NGI_GRNET:
- GR-04-FORTH-ICS
- NGI_IT https://ggus.eu/index.php?mode=ticket_info&ticket_id=118846:
- INFN-NAPOLI-PAMELA: in decommissioning
- NGI_MARGI https://ggus.eu/index.php?mode=ticket_info&ticket_id=118465 no monitoring data since January
- ROC_LA:
- UFAL: new site but the monitoring data are missing
Next meeting
- 11 Apr 2016 https://indico.egi.eu/indico/event/2738/