Project

General

Profile

March-12-2019

Slides: https://indico.fnal.gov/event/17324/


Present

Parag Mhashilkar - Project Lead
Marco Mambelli - Technical Lead
Dennis Box - Project Member
Lorena Lobato - Project Member
Marco Mascheroni - Project Member/OSG Factory Operations
Tony Tiradani - HEPCloud Technical Lead
Steve Timm - HEPCloud Technical Advisor
Krista Majewski - HEPCloud Project Manager/GCO
Antonio Perez-Calero Yzquierdo - CMS
Brian Lin - OSG
Jeff Dost - OSG Factory Operations
Tanya Levshina - FIFE
Ken Herner - FIFE
Joe Boyd - FIFE
Mike Kirby - FIFE
Edgar Hernandez - OSG/GLOW


Communication


Support


Project Management

  • Antonio: Based on the roadmap in the slides, whats the translation for high priority items in terms of timeline?
    • Mambelli: They will be assigned to releases based on the stakeholder's input

Roadmap


Technical

Python3
  • Steve: W.r.t migration to python3. What type of testing is done related to containers and python3?
    • Mambelli: SL7 has python3 support and migration to python3 is not related to containers. We have unit testing etc for python3 that we will be using as part of migration
  • Edgar: For Non HEP users, first thing is to have a frontend for RHEL7. Once it is working we start thinking about the migration
    • Brian: Python 3.4 is available in SL7, so they can start trying out anytime
RHEL6 & RHEL7
  • Jeff: We don't want glideinwms dropping python 2 support before we are ready for python 7.
    • Mambelli: We are gathering input and like to know expected timing for migration to RHEL7?
    • Jeff: Expect RHEL7 in a month or so
    • Jeff: We need factories to move to SL7 before we consider python3
    • Tony: At FNAL we are running factories under SL7
  • Parag: what are OSG's plans for dropping support for RHEL6?
    • Brian: Expect osg 3.5 later this year and will not have RHEL6 support. We won't drop RHEL6 support mid stream for existing versions.
Containers
  • Mambelli: Input for containers?
    • Brian: we are interested in containerizing and distributing containers and happy to collaborate.
    • Edgar: To containerize nicely will give ability to have config files same as factory config.d. Will make transition to container easier and make it easy to manage new VOs. We have couple of students to work on that.
Monitoring
  • Edgar: Have more thoughts about monitoring and can share with you. Factory should produce information that will allow them to do elastic search etc. All non CMS VOs are using json/xml etc. Right way to say is we support json/xmli/grafana etc and community can decide what they want.
    • Mambelli: Thats the plan, generate info in JSON or XML and make it available for factory and frontend users to consume and they can decide how to build their monitoring dashboards
  • Mambelli: Is the suggestion to drop support in factory for the historical info?
    • Jeff: thats what we are doing right now. We do not use GlideinWMS supported history info web pages.
  • Mambelli: What about the frontends
    • Edgar: Don't like to spend time to make html pages better. Leave them right now and keep them for historical reasons and let VO decide on how to consume the current data
    • Jeff: RRDs are headache and we do not want to keep supporting them. We are not ready to move right now
    • Edgar: we need both for a while
Other Topics
  • Jeff: What is the status of GlideinWMS with max walltime to be periodically removed issue $20301
    • Mascheroni: its in the v3.4.3 but not deployed in the factory

ACTION ITEMS

  • Marco Mambelli will follow up with Edgar about the students to work in containerization
  • Zoom coordinates in the meeting invite for phone are incorrect and need to be fixed
  • Publish GlideinWMS roadmap to the wiki
  • Dennis will check back with Brian about retrieving html artifacts from Travis CI