September OpenZFS Leadership Meeting

[Matthew Ahrens]

The video recording has been posted:
https://www.youtube.com/watch?v=fVcNnFHDwHY

meeting notes:

   -

   OpenZFS hackathon (Matt)
   -

      Spreadsheet with Hackathon Ideas/Breakout Sessions:
      https://docs.google.com/spreadsheets/d/1qH-qST3uSYVh7fzWHWIB1eM1y5rAYCQs6tRDjOhVYx0/edit#gid=0

      -

      Feel free to put your ideas in the doc even if you are just working
      by yourself!
      -

      Since it will be held virtually, the hackathon will be slightly more
      structured this year. Each idea proposer will be given a couple
of minutes
      in the beginning of the day to pitch their idea. Then a Zoom
Room (and/or a
      Slack channel) will be created for each team project.
      -

   Per-pool ARC STATS (Richard Elling)
   -

      Unlike the L2ARC, the ARC statistics are not broken down to per-pool
      statistics. It would be nice to be able to break them down into that
      fashion when dealing with issues.
      -

      For the short-term we could just add a few trace points but in the
      long term we may want something more integrated to ZFS.
      -

      This is still in the idea phase and the most straightforward solution
      so far seems to be adding more specific arcstats under the SPL kstat code
      (the upside of this approach being that there is no need for a userland
      change whatsoever).
      -

      There were also a couple of tangential issues brought up to the
      discussion:
      -

         The idea to potentially attempt to break this info even further to
         a per-dataset level
         -

         Existing aggsum implementation being slightly more heavy-weight
         and not as scalable as existing percpu counter
implementations in FreeBSD
         and Linux. We could either learn from these platform-specific
solutions and
         improve the aggsum code’s design, or just get rid of its
existing code and
         make it a thin wrapper on top of the existing
platform-specific solutions.
         -

   ZED all-syslog.sh (Don Brady)
   -

      Illumos’ FMA logs were retained after a crash but Linux’s zpool
      events are not. There exists all-syslog.sh in ZED but its output is terse
      and misses any useful information.
      -

      Don has prototyped a change in all-syslogs.sh to provide more useful
      information for some events (e.g. the vdev, and offset fields
for checksum
      error events).
      -

      Even though changing that code may result in the breakage of some
      existing stuff, all agreed that this is a step into the right direction.
      -

      Don expects to open a PR for this soon.
      -

   OpenZFS 2.0RC (Brian)
   -

      The first 2.0 release candidate was created 2 weeks ago and a new one
      is expected to open this week.
      -

      People who want to try it out are encouraged to do so while the code
      is soaking
      -

      There is no strict schedule for these RC releases but for now we
      expect to see one every two weeks approximately.
      -

      As an overall status, the project is still on track for the official
      2.0 release before the end of the year.


On Mon, Sep 14, 2020 at 11:58 AM Matthew Ahrens <mahrens at delphix.com> wrote:

> The next OpenZFS Leadership meeting will be held tomorrow, September 15,
> 1pm-2pm Pacific time.  We don't have many topics on the agenda for
> tomorrow's meeting, so let me know if you'd like to add anything.
>
> FYI, it's 3 short weeks until the 2020 OpenZFS Developer Summit
> <https://openzfs.org/wiki/OpenZFS_Developer_Summit_2020>!
>
> Everyone is welcome to attend and participate, and we will try to keep the
> meeting on agenda and on time.  The meetings will be held online via Zoom,
> and recorded and posted to the website and YouTube after the meeting.
>
> For more information and details on how to attend, as well as notes and
> video from the previous meeting, please see the agenda document:
>
>
> https://docs.google.com/document/d/1w2jv2XVYFmBVvG1EGf-9A5HBVsjAYoLIFZAnWHhV-BM/edit
>
> --matt
>