[Ksummit-discuss] [TECH TOPIC] Open panel for discussing MM problems

[Ksummit-discuss] [TECH TOPIC] Open panel for discussing MM problems

[Mel Gorman]

There is at least one MM topic proposed so there is a chance that some
MM people will be around. I hear that a few people have been struggling
with MM-related topics recently but not getting a lot of traction. It's
hard to get a handle on how big of a problem this is or if it's even a
problem. It's possible I just happened to hear from the only people who
are unhappy with MM.

I think it would be beneficial to have an open MM panel on the morning of
the day that overlaps with Plumbers to get a handle on how many people
have MM-related problems and what class of problems they're having. It
would start by having each member spend 1-3 minutes talking about what
they're currently then open it up to the floor.

People would be welcome to describe any MM-related issues they have and
ask the panel to either propose solutions or at least note that it's
something that should be fixed in the future. Ideally this would be a
better description than "MM sucks" or "sometimes the system is slow and I
feel the MM is responsible". If there is a concrete problem, there would be
no guarantee it will get fixed by anyone on the panel but at least we'd be
aware. It might be a bust if we don't have enough for a panel or everyone
is perfectly happy with the MM as it is. However, if it works and there
are some problems discussed then it'll be interesting to see how many,
if any, are fixed by the next LSF/MM.

The people I know are willing to participate if they are at Kernel Summit
are Vlastimil Babka, Johannes Weiner, Rik van Riel and myself. There is
a strong preference for the morning due to potential flight conflicts but
hopefully that can be arranged without colliding with Plumber keynotes. If
anyone else was interested in being up there, they'd be more than welcome
assuming they know which way is up with mm/.

-- 
Mel Gorman
SUSE Labs

Re: [Ksummit-discuss] [TECH TOPIC] Open panel for discussing MM problems

[Jan Kara]

Hi!

On Fri 29-07-16 15:11:26, Mel Gorman wrote:
> There is at least one MM topic proposed so there is a chance that some
> MM people will be around. I hear that a few people have been struggling
> with MM-related topics recently but not getting a lot of traction. It's
> hard to get a handle on how big of a problem this is or if it's even a
> problem. It's possible I just happened to hear from the only people who
> are unhappy with MM.
> 
> I think it would be beneficial to have an open MM panel on the morning of
> the day that overlaps with Plumbers to get a handle on how many people
> have MM-related problems and what class of problems they're having. It
> would start by having each member spend 1-3 minutes talking about what
> they're currently then open it up to the floor.
> 
> People would be welcome to describe any MM-related issues they have and
> ask the panel to either propose solutions or at least note that it's
> something that should be fixed in the future. Ideally this would be a
> better description than "MM sucks" or "sometimes the system is slow and I
> feel the MM is responsible". If there is a concrete problem, there would be
> no guarantee it will get fixed by anyone on the panel but at least we'd be
> aware. It might be a bust if we don't have enough for a panel or everyone
> is perfectly happy with the MM as it is. However, if it works and there
> are some problems discussed then it'll be interesting to see how many,
> if any, are fixed by the next LSF/MM.

Sounds useful.

> The people I know are willing to participate if they are at Kernel Summit
> are Vlastimil Babka, Johannes Weiner, Rik van Riel and myself. There is
> a strong preference for the morning due to potential flight conflicts but
> hopefully that can be arranged without colliding with Plumber keynotes. If
> anyone else was interested in being up there, they'd be more than welcome
> assuming they know which way is up with mm/.

I'd be happy to join as well. I'm not really an MM guy but I can bring
knowledge of writeback code and MM-FS interactions which sometimes cause
issues.

								Honza

-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR