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
- UMD Preview repository
- Replace EMI from early 2016 on (will start from a EMI3 mirror)
- while TPs send usual updates to UMD for verification/ integration, the preview repo goal is to make products available to the community before UMD verification/integration (no QA, just “UMD” community repository)
- Release monthly, together with provider information (release info, bug fixes, new features...)
Middleware releases and staged rollout
UMD release
- UMD 3.14.0 released, on Nov26, broadcast delayed to Dec3 (CGSI-gSOAP 1.3.8, GFAL Utils 1.2.1, SRM-IFCE 1.23.1, dCache Server 2.10.42, globus-default-security 6.1.0, GRAM5 6.1.0, GridFTP 6.1.0, MyProxy 6.1.13, StoRM 1.11.9)
- UMD 3.14.1 integration tests ongoing (DPM 1.8.10, FTS3 3.3.1, XROOT 4.2.3), scheduled for this week (deadline is OMB)
- UMD 4.0.0 (FTS3, ARC, Site/Top BDII, dCache), scheduled for this week (deadline is OMB)
- CentOS7 only, SL6 support will be introduced on early 2016 while decommissioning SL5
Staged rollout updates
- UMD3
- myproxy 6.1.15
- gridftp 6.1.0
- globus-default-security 6.2.0
Under Staged Rollout
In Verification
- UMD3
- arc 15.03.4 (sl6/centos7)
- voms-admin-server 3.4.0 (requires java 8)
Ready to be released
- UMD3
- fts3 3.3.1
- xroot 4.2.3
- DPM 1.8.10: Somehow problematic release since it requires xroot 4.2.3 and new globus components which are under staged rollout. (instability issues reported: https://ggus.eu/index.php?mode=ticket_info&ticket_id=118203)
- UMD4
- site-bdii 1.2.1
- top-bdii 1.1.4
- dcache 2.10.42
- fts3 3.3.1
UMD 3/UMD 4 EA
- we need Early Adopters for CentOS7!
Next releases
Operational issues
Decommissioning dCache 2.6
- support for dCache 2.6 ended at May 2015
- we made an assessment to understand how many sites still expose dCache 2.6 endpoints.
- decommission the dCache 2.6 endpoints by the end of January 2016 (or before) https://ggus.eu/index.php?mode=ticket_info&ticket_id=118248
- ref: https://wiki.egi.eu/wiki/PROC16
Decommission of SL5
- SL5 support aligned with RHEL5
- no more "Full support”; end of "Transition" Phase on January 31, 2014 • not getting anymore new software functionalities
- in "Maintenance" until March 31, 2017 • only urgent/critical fixes until then
- Supporting CentOS7 in UMD requires to schedule the end of support of SL5 in UMD
- No more new packages for SL5, only security/important fixes accepted
- SL5 services must be decommissioned by end of April 2016; broadcast at December, probes will be warning since February 2016 to start helping with decommissioning
APEL on SL5
- APEL sl5 publishers send data to the message brokers using the SSLv3 connections, and the brokers still allow them; but, if you remember the poodle vulnerability and its relation to sl5, in case there is a new poodle exploit, these connections will be blocked and the SL5 publishers would fail to send messages to APEL. For sl6 instead there won't be any problem
- we would like to know how many APEL publisher instances are installed on sl5 in your NGI; to avoid confusion, we mean the instances that on GOC-DB are registered as glite-APEL
- Please communicate at operations@egi.eu this number so we can choose how is better proceeding, in coordination with the APEL team, for the migration of the affected instances to sl6. Deadline is TODAY Dec14, discussion will happen at next OMB (Dec17)
- 14 NGIs replied so far (13 instances complessively)
WMS Usage Assessment
- we know there are several non-HEP VOs that make large use of WMS service, and we would like to gather some data about the real usage of this service for better planning its future.
- questionnaire sent to noc-managers (how many WMS servers, how many VOs served, which VOs and load for each VO, job stats) and to VOs.
- AfricaArabia, NGI_DE, NGI_IBERGRID, NGI_NL, NGI_PL, NGI_RO, NGI_TR and ROC_CANADA have replied so far.
New CE/batch system accounting integration
Before sites roll in production new batch system, they should contact APEL team to make sure that accounting logs are properly parsed by the APEL parsers.
AOB
Monthly Availability/Reliability - October Underperforming RCs follow-up
- since October 2015 the reports are no more pubished in pdf format. They are displayed on ARGO: http://argo.egi.eu/lavoisier/ngi_reports?month=2015-10
- as usual, it was found several RCs eligible for suspension. Some of them recovered;
- still problems on:
- SCAI (NGI_DE) currently is not published in the top-BDII, so it is failing all the tests. If no feedback, we are going to suspend it this week;
- at UNI-DORTMUND (NGI_DE) currently the test jobs remain scheduled forever, and so it is failing the job submit probe
- MK-04-FINKICLOUD (NGI_MARGI): it is a CLOUD RC, but it is moniterd by mistake among the GRID RCs:
- On GOC-DB the host nebula.finki.ukim.mk is registered also as Site-Bdii: with this information, ARGO/SAM considers the site as a GRID one, but the monitoring returns a failure because the site information are published only in GLUE2 (it is a CLOUD site).
- the others were suspended:
- T3-TH-CHULA (AsiaPacific)
- INAF-TS, UNI-PERUGIA (NGI_IT)
- MK-01-UKIM (NGI_MARGI)