2022-05-27 - Sys Ops & Management SIG Agenda and Meeting notes

Date

Attendees

  •  review Technical Pain Points 

Discussion items

TimeItemWhoNotes
5WelcomeIngolf
55FOLIO Technical Pain Points

FOLIO Technical Pain Points - DRAFT - Google Docs

---------------------------

SysOps SIG to answer these questions:

  • Is this the right set of pain points ?
  • Why and how is something painful ?
  • What negative impact does it have  ?
  • How do we prioritize it ? 



FOLIO Technical Pain Points  - ConclusionsNew document: SysOps pain points and priorities Spring 2022

There has been some misunderstanding about the ownership of this document. The ownership is TC, not SysOps SIG.

We think the sections in this document need to be re-grouped and re-written. We could not make any prioritization at this state of the work.

  • Ingolf has re-winded the document to 05/20/2022 5 EST , thus restored changes that have been accidentally lost. Then he has re-inserted the comments made during the May 20th meeting.
  • Wayne will take that document and compile a SysOps-Version of the document; with pain points from the system administrators' perspective

--------------------------------------------

Notes :

There is overlap by the two documents by design.

Are there any topics that are missing in the document ?

  • Operational documentation: The documentation is scattered. It is time consuming to apply it.
    • Improve architectural operational documentation
    • Installation procedural documentation is pretty informal and unequally maintained
    • Module documentation is very distributed, not well summarized
    • Release Upgrade documentation
    • A legitimate response of the Tech Council could be: Hey, TAMU, contribute 20 hours of your time to write a documentation (common agreement)
  • Okapi bootstrapping
    • Tenant bootstrapping. This is an awkward process
    • Registering modules with Okapi discovery. There is at least an extra step. → Platform service discovery
    • Tooling and managing Okapi.
    • Tod: Pulling in two different directions on Okapi. Come to a consensus in what direction we want to go. Okapi pain points. How big is Okapi and is it doing the right things ? There is a lot of work that Okapi does that is not really useful in a production setting. Tod: This doesn't seem to be a main point. Sit down with Jakub and others: Here is what we want to expand in Okapi, here are pain points in Okapi. Wayne: Some of the tooling could be provided by the project. Tod: There are conflicting prescriptions for how to proceed.
    • Wayne: I hestitate to give a prescription "This is how Okapi should work". Tod agrees. It should be rather: "This is what Okapi does". 
    • Harry: What is the ideal procedure you would like to see ? Installation, development and operational.
  • Basis FOLIO Tenant Installation
    • It is quite complex to get a system up and running. As a hosting provider, I may have only one tenant, but a good piece of infrastructure to support this tenant. → Reduce the scope of a minimal installation. Reduce the scope of the basic ILS profile.   
    • This will also simply the extermely complex dependency graph that we have (all agree that this is a pain point for operations)
    • Operators should be able to klick a button and install a tenant.
  • Stripes platform build and deploy
    • Wayne added this point. It could be streamlined
  • Observability / Logging / Troubleshooting
    • This is a real pain point
    • When something goes wrong, it is hard to know what is going on.
    • Logs are inconsistent in both what they log, how they are formatted and they spread everywhere.
    • Chase down a set of events when you are troubleshooting
    • Marc: We shouldn't aggregate the aggregation. Even the logs from a single module spread out if you have multiple instances of a single module.
    • We can provision somebody else to aggregate (ELK, Splunk, etc.)
    • Making things easy to find, making things available and making things ... are three different things (Jason)
    • Wayne: We should get better with what we have before we get better with what we should do.
  • Performance
  • Release Upgrades
    • Indexing takes a lot of time; significant down time
    • not very well documented
    • not very well documented to implement things like blue-green deployment
  • Security

-----------------

What needs to be done now:

  • some sections need to be worked out
  • then a small group would meet
  • then the group should come together and bring the proposals to the TC
  • Wayne: Not a plain dump of complaints, but a set of suggestions; I followed a "consistency of style" in the document.
  • Wayne: Keep it constructive and actionable.
  • Wayne: I edit "summary of the impact"
  • Others (Jason etc.) will add suggestions (for the sections which are not fleshed out)
  • Use "suggestion mode" instead of "editing mode".

A small group will meet on Wednesday, June 1st, 12 (noon) EST. Ingolf will invite.

Please put your suggestions to the document SysOps pain points and priorities Spring 2022 - Google Docs in editing mode.

Action items


  • Ingolf Kuss Plan Operational Needs session for WOLFCon 2022