SA2.6 meeting: EMI-1 RC deployment on Integration Testbed
CERN
Issue to discuss:
- How many servers needed for gLite, ARC, UNICORE (we can partially re-use some server deploying non EMI products, or decommissioned (sl4 gLite versions for example)
- Platforms
-
Time schedule:
- EMI 1 RC0 pre-release is available: 18 February. Packages made available in testing repository will contain untested, non-certified packages. The affected components https://twiki.cern.ch/twiki/bin/view/EMI/EMI-1#ETICS_configuration_schedule
- SA2 testbed: RC0 rolled out on the testbed: 23 February
- Installation of new packages from RC repository. Update issues: what happens when packages change in the repo?
- Information System issue: separate bdii for RC versions? We already went through this a while ago, but for EMI-1 it is worth to review the issue.
- Distribution of RC across testbed sites + Effort.
- AOB
Title: SA2.6: EMI-1 RC deployment on testbed
Community: Universe
Meeting Access Information:
- Meeting URL
http://evo.caltech.edu/evoNext/koala.jnlp?meeting=MeMMMu2M2DDvDi9a9BDv9e- Phone Bridge ID: 292 0304
RC-0 details: Cristina
-
2 repositories will be available on 18, 1 for Production and 1 for EMI-1 RC0.
-
All services are supposed to be in the repository by 18, although UNTESTED, so from testbed point of view we will deploy all hosting servers and make them point to the RC0 repo.
-
platform will be just sl5 64.
-
decomissioning: keep previous middleware production versions.
Alberto: all requests thorught GGUS + timing coordination in the deployment order of services?
Deployment ISSUES:
-
Bjiorn, Unicore:
-
no HW resources problem
-
no problem of deployment order coordination: all services ready together.
-
-
Jozef, Arc:
-
no HW resources problem
-
no coordination problems: all services ready at same time. No resources problem.
-
-
Danilo, gLite:
-
we need a deployment plan, on the base of which we'll know about resources availability
-
Frantisek- CESNET: 2-3 virtual servers available. Possibly migrate LB to other site, due to low HW disk performace.
-
-
Christian, dCache: ?
Action: Danilo+ Cristina: detailed plan of deployment for gLite.
Action: Danilo: open GGUS tkts to testbed to track deployment of RC0 versions, for all middlewares
Action: Danilo+ Tomasz + Frantisek: service deployment for gLite.
VO to be used:
-
Danilo: PT still asking for VO dteam to be supported.
-
Alberto, Cristina: No support. Testers is the VO to be used as stated in PEB.
Information System
-
We set up a separate information system instance for production resources and RC resources:
-
Tomasz: OK for gLite
-
Bjorn: OK for Unicore
-
Jozef (ARC): need feedback from developers
-
Action: Jozef feedback about info sytstem.
Training on Testbed:
-
Date: we keep 1st of march 10-12 am.
-
Tempative Outline:
-
Danilo: introduction going thorugh the Testbed twiki (30 minutes)
-
Tomasz: gLite testbed usage basic example: connect to UI, proxy creation with testers VO, see Testbed resources (prod), job submission, data management (Coordinate with Christian (also with Arc and Unicore if dCache testbed is reachable by their testbed)) (20 minutes, including dCache?)
-
Bjorn: unicore testbed basic usage examples: similar to gLite? (10 minutes)
-
Marek/Jozef: arc testbed basic usage examples: similar to gLite? (10 minutes)
-
QUESTION TIME (40 minutes): (ALL answer)
-
-
Show things in the stable production testbed, not on the RC-0.
-
Suggestion about examples material to be attached in the presentation: list of commands/operations with comment.
Action (all): MEETING: 9:30 next tuesdays Feb.22 about Training.
Action (Alberto): get information about the training audio-video presentation system.
Action (Alberto-Danilo): send registration invitation, asking for questions in advance.
AOB:
-
Christian: Problem with HW migration occurring these days. Trying to postpone the migration.