Meeting notes: Workstation WG, 6 Jan 2020

[Allan Day]

Hi all; see below for today's meeting notes.

Thanks,

Allan
-- 

# Fedora Workstation WG meeting, 6 January 2020, 9:00 EST

## Attendance

 - Present: Allan, Chris, Neal, Michael, Langdon, Matthias, Jens
 - Missing: Kalev, Christian

## Agenda

### Permanent chair issue - #114

 - Chris, Michael and Allan have volunteered. Michael proposes that we
appoint Chris as chair and Allan as vice-chair, until F32
 - The vice-chair will be responsible for meeting minutes and publishing.
If the vice-chair is chairing the meeting, a secretary will be named for
that meeting.
 - Proposed: vice-chair to serve as secretary unless the chair is
unavailable, in which case a temporary secretary will be selected while the
vice-chair fills in.
 - Do we want to use Etherpad for notes and then publish through Zodbot?
Langdon describes some advantages of Zodbot. ACTION: Allan will look into
it.
 - Do we want to publish video recordings of the meetings? Various views;
in the end we decide to wait and see if anyone requests them.

### New meeting time - #118

 - Due to time zones and members' existing commitments, the current options
are: 7:30 EST (any day); 9:30-11 (T, W, Th)
 - ACTION: Michael to send out a poll with these options

### Taiga and/vs Pagure #112

 - Some reservations about having issue tracking across two systems -
potential for duplication of information and conversations happening in two
places.
 - Some opposing views:
    - Neal: there are benefits to curating a Kanban - so it doesn't get
backed up with massive numbers of potential work items.
    - Langdon & Allan: Taiga as a transient work view - an extra layer of
management. Shouldn't have the same material as the Pagure issues.
    - Langdon: avoiding duplication will require work and maybe some
corrective action if we stray from the path
 - The decision is to let the chair decide. Chris is happy to try Taiga and
see how it goes.

### Enabling earlyoom by default - #119

 - Background:
   - earlyoom is being proposed for F32 as a partial solution to systems
locking up due to memory shortages (the classic swap system lock-up)
   - earlyoom finds the process that's causing the memory issue and
sigterms it (slightly kinder than just killing it). This will typically
return the system to a usable state in a couple of minutes.
 - Matthias: the real issue is the kernel's swap implementation being too
slow. Neal: it seems that the kernel people aren't interested in working on
this.
 - earlyoom is not a great/complete solution, but it will help.
 - Chris has some concerns about earlyoom being adequately tested.
 - The work in #98 will help
 - Neal: earlyoom will be a stability improvement. The potential problem is
it killing processes prematurely/unnecessarily. With earlyoom we can reduce
the default swap size.
 - Langdon: you need to have a large swap for hibernate to work. Neal:
earlyoom could break hiberate (?) Chris: there's some related discussion in
#120.
 - earlyoom is being raised so that people are aware of it. Langdon: the WG
should have a position on the change proposal.
 - Allan: a two minute hang isn't much of a solution from a UX perspective
- a lot of users will give up and do a hard reboot within that time.
There's some interest in using hibernation more extensively; earlyoom could
be an issue if it makes that harder.