rbowen / centos / board

Forked from centos/board 3 years ago
Clone
Text Blame History Raw

Board minutes 2016-06-15

Attendees

  • Jim Perrin
  • KB Singh (Chair)
  • Fabian Arrotin
  • Johnny Hughes
  • Karsten Wade (Secretary, meeting #2)
  • Michael McLean
  • Ralph Angenendt
  • Carl Trieloff

Guests

  • Richard Megginson (OpsTools SIG)
  • Matthias Runge (OpsTools SIG)

Agenda & notes

Optools SIG

  • Lack of tools in CentOS (and Fedora) for operators of large infrastructure/installations
  • Doesn’t fit in e.g. Cloud SIG
  • Home for these tools + bits and pieces to install/config infrastructure
  • E.g. if logging, there are good & bad examples of what to log, how to log, etc.
  • So provide a guide of how-to install each specific tool, what to watch, etc.
  • E.g. OpenStack installations log way too much, including things that are not errors but look like errors, so need some filtering afterwards.
  • Useful outside of each specific SIG.
  • Relationship with Config Mgmt SIG
  • Would provide manifests/playbooks/recipes/etc. To get tools installed would first look to existing sources in CentOS & SIGs.
  • Configs ultimately belong to the Config Mgmt SIG.
  • Working against duplicated efforts in tools stacks
  • Diff stacks of software reimplement or bring in the same tools
  • Use case: correlate logs amongst all these things
  • E.g. openshift on top of openstack, want to be able to correlate the logged error into each part of the stack - storage, underlying OS, orchestration, etc.
  • Give operators tools to do deep-cause analysis
  • KB - having ELK, graphon, etc. in the project helps on community side as we have a very operator-heavy base, many potential consumers.
  • Proposal is on the wiki -- https://wiki.centos.org/SpecialInterestGroup/OpsTools
  • What about the lack of upstream connection? Are any SIG members from the relevant upstream projects.
    • Most things are not in our direct control, refer back up to EPEL.
    • Willing to build bridges into upstreams where there are no existing upstream developers here
    • Talking with actual upstream developers.
    • Elastic is changing to an ‘open core’/’commercial open source’ model
    • Have some contributions in
    • Paying attention to open source replacements for problematic source models.
  • Can start with current proposed package set, but can replace components one-at-a-time without disrupting user experience.
  • Provide modularity so we’re not locked in
  • “Log collector package” -- don’t need to know what is underneath, but can choose if you prefer.
  • Vote - Should the OpsTools SIG proposal be accepted and the SIG move forward?
    • +1 from all present
    • Tru vote, comments, questions requested

SIGs ready for rise up?

  • Cloud SIG getting quite close

    • Work is having a lot of influence on downstream
    • Sorting out how that’s going to work
    • Maybe segregate RH parts?
    • Maturing very quickly
    • Expect to see them in self-admin mode soon-ish
  • PaaS SIG very organized

    • Closed group
    • Quite RH focused
    • Not ready yet, maybe next meeting or so?
  • Virt SIG

    • Established, self-organized
    • But … George Dunlap is moving from packaging to developing upstream
    • He’s actively looking for a takeover
    • Waiting for leadership shift

Credit Suisse

  • SE Asia operations invests 5 to 7 mil in corporate social responsibility, including into technology

    • Hosting contests, giving away laptops, etc.
    • Want to spend on FLOSS projects
    • Moving dev environment from Debian to CentOS
    • Have 150 servers to contribute to FOSS projects
    • Spread across: Singapore, Hong Kong, Tokyo, Sydney
    • Need to handle hosting ourselves
    • Can we get that DC space?
    • Yes, via RH sponsorship, we can start this
  • Additional discussions

    • Host Dojos
    • Sponsor local community people in those specific countries
    • Reasonable interest in that piece
  • Sponsorship level?
    • Likely they host the details themselves
    • We may not need to create a ‘level’
    • Support might be 1 to 4 years through their contracts, depending on how much we have installed
    • Might be a foot in the door toward further contributions that are equally or more meaningful.
  • Value

    • What is the value of these machines v. cost to us to host v. canvassing for more hosting?
    • July call to thrash out synergy
  • Figure out hosting costs