SLATE Collaboration Face-to-Face

Name: SLATE Collaboration Face-to-Face
Start: 2019-01-21T08:30:00-05:00
End: 2019-01-23T17:00:00-05:00
Location: Other Institutes

21 Jan 2019, 08:30 → 23 Jan 2019, 17:00 America/Detroit

Other Institutes

University of Michigan Physics 450 Church Street Ann Arbor, MI

Description

Please register so we can track number of attendees.

Meeting details in this Google doc

Contact

smckee@umich.edu

+1 734 255-8293

Registration

SLATE Collaboration Face-to-Face Meeting Registration

Participants

Ben Kulbertis
Bob Killen
Chris Weaver
Gabriele Carcassi
Jason Stidd
Joe Breen
Lincoln Bryant
Robert William Gardner Jr
Shawn Mc Kee
Shelly Johnson
Todd Raeker

Monday 21 January
- Mon 21 Jan
- SLATE Demonstration: SLATE Workshop Startup
  - 08:45
    
    Coffee and breakfast
    
    Coffee and breakfast outside the room
  - 1
    
    Workshop Introduction
    
    Introduction (Rob)
    Workshop logistics (Shawn)
    ATLAS Centrally Managed Build Process:
    https://docs.google.com/document/d/1c7Lmmp_-vsrCKg2Ggie6WWfjcaO-oDw4lOIqFRNXalE/edit#heading=h.x803ifi2xymt
  - 2
    
    Build Utah machine from raw hardware
    
    Lincoln and Ben facilitate
    https://docs.google.com/document/d/1c7Lmmp_-vsrCKg2Ggie6WWfjcaO-oDw4lOIqFRNXalE/edit#heading=h.x803ifi2xymt
    Boot machine from USB image
    Install software using Puppet
    Provision network (Calico, metallb)
    Manual registration steps
    Validate basic Kubernetes and SLATE client functionality
- 12:30
  
  Lunch
  
  Find places nearby along South U
- SLATE Storage: ATLAS Storage Discussion
  
  Bob Killen facilitates
  - 3
    
    ATLAS Storage Discussion
    
    ATLAS project defines disk requirements.
    How do we layout disks on ATLAS nodes?
    How do we provision and advertise the storage consistently for xcache and other SLATE apps?
    ATLAS XCache is relying on manual partition in current packaging
    Does this storage packaging allow for other SLATE or ATLAS apps seamlessly?
- SLATE Node Level Monitoring: ATLAS Node Level Monitoring
  - 4
    
    Discussion on SLATE Monitoring with ATLAS as a Driver
    
    Gabriele and Bob facilitate
    
    Monitoring doc:
    https://docs.google.com/document/d/1lMji2dwLPkHPgtgYk5U5f0H500kOLeQswAd9SbBl1RU/edit#heading=h.n9pkgznjiiyp
    Goal for planning session: come up with the general deployment strategy that will be implemented over the next few months and record it in JIRA
    Metrics
    Logs
    Goal for working session: Prometheus/Grafana installed at least on one cluster standalone
    Items to discuss:
    Deployment strategy for metrics monitoring (standard helm charts? create our own?)
    Deployment strategy for logs
    Outside access (ingress, dns name, …)
    Location for central services (i.e. Grafana and Prometheus aggregator)
    Data aggregations from the clusters (Push Gateway, prometheus/thanos/cortex?)
    Cluster level permission (what requires admin privileges and what doesn’t?)
    Metrics (what do we want to monitor from each node?)
    User level permissions (i.e. logins and permissions for Grafana)
- 15:30
  
  Coffee and Snacks
- SLATE Monitoring Discussion: Monitoring Discussion for SLATE
  - 5
    
    SLATE Monitoring Discussion
    
    Discuss what SLATE should do about monitoring for at least two use-cases: 1) Monitoring for the SLATE application developer: what does the "SLATE pane of glass" provide for a developer and 2) Monitoring for the SLATE platform managers: what does the "SLATE pane of glass" provide for a platform deployer/manager?
    
    Other cases?
    
    How to monitor the network?
- Wrap-Up Day1: Wrap-Up and Action Items
  
  Summarize day's work, review action items.