►
From YouTube: Kubernetes SIG Scheduling Meeting - 2019-06-20
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
All
right,
as
you
know,
this
meeting
is
recorded
and
will
be
uploaded
to
public
Internet,
so
chances
are
whatever
you
say
remains
for
a
very,
very
long
time.
Okay,
we
have
a
few
items.
I
want
to
actually
give
you
one
everybody
who
is
here
and
also
people
who
may
see
this
video
later.
We
want
to
increase
the
number
of
our
reviewers
and
approvers
in
the
sake,
because,
right
now,
we
sometimes
feel
like
there
are
PR
that
are
not
getting
enough
attention
by
our
reviewers.
A
Current
pool
of
reviewers
is
relatively
small,
and
some
of
them
are
not
very
active.
In
fact,
I
myself
may
become
a
little
bit
less
active
as
I'm
getting
him
off
with
other
projects,
and
I
have
spoken
with
Abdullah.
Luckily,
he
is
gonna
cover
for
me
enough
and
in
the
second
he's,
gonna
continue
reviewing
PRS,
but
we
want
to
actually
have
more
folks
if
you
have
D,
if
you
have
the
requirements
to
become
a
reviewer
or
approver,
please
volunteer
to
do
so.
A
That's
about
that
with
respect
to
items
that
we
were
working
on,
there
are
a
number
of
PRS
for
adding
extension
points
for
the
scheduling
framework.
I've
reviewed
some
of
those
some
of
them
are
in
progress.
There
was
one
particular
PR
for
adding
filter
extension
points
that
that
was
that
actually
became
a
little
bit
technically
challenging
not
much,
but
it
was
a
little
bit
technically
challenging
because
of
the
way
that
we
are
integrating
the
scheduling
framework
with
the
current
scheduler,
and
we
want
all
the
predicates
that
currently
exists
to
work
without
any
without
any
changes.
A
Basically,
so
we're
working
on
that.
Hopefully
we
can
get
that
done
soon.
I
know
that
Abdullah,
you
have
sent
PRS
for
adding
new
scoring
extension
points,
so
we're
reviewing
those
as
well
and
then,
after
these
extension
points
are
done,
then
we
can
focus
on
converting
some
of
our
extension
point
of
some
sorry,
some
of
these
existing
predicates
and
priority
functions
into
plugins
for
the
scheduler.
A
That's
about
that
there
is
one
actually
got
two
relatively
large
projects.
One
is
even
hotter
spreading
among
failure.
Domains
that
way
is
working
on
that
is
in
progress.
Some
of
the
PRS
are
waiting
for
API
approval.
So
as
soon
as
that
is
done,
we
can
probably
get
the
rest
of
the
PRS
in
as
well,
and,
of
course,
we
need
to
really
wanna
prove
those.
A
Then
there
is
a
kind
of
a
relevant
project
to
this
and
that's
a
physical
host
spreading
I,
don't
know
if,
if
I
have
it
covered
here
in
our
spreadsheet,
let
me
see
I
guess:
yeah
I
have
I,
have
it
it's
actually.
The
third
item
third
feature
in
that
spreadsheet,
so
this
one
does
not
have
any
owner
at
the
moment.
If,
if
you
are
willing
to
take
it
any
of
you
folks
is
willing
to
take
it.
Please
let
me
know
the
idea
behind
that.
You
know
there
is
I
believe
there
is
an
issue.
A
You
can
read
a
little
bit
more
about
that.
The
idea
behind
that
is
that
we
want
to
find
a
way
to
mark
physical
house
on
the
nodes
of
a
cluster.
Basically,
once
you
know
that
a
node
in
a
cluster
runs
on
physical
host
X,
we
want
to
add
a
label
to
the
node
saying
that
physical
host
of
this
node
is
X.
This
helps
us
identify
nodes
that
run
on
the
same
physical
host,
and
basically
this
allows
us
to
in
the
scheduler
to
spread
thoughts
among
different
physical
host
as
well
as
different
nodes
of
a
cluster.
A
So
if
this
is
an
interesting
project
for
the
first
phase,
what
we
had
in
mind
was
to
support
VMware.
Basically,
there
is
a
VMware
plug-in
I
believe
that
if
you,
if
you
look
at
that
the
issue
that
is
linked
in
our
spreadsheet,
you
will
find
more
information
about
that.
There
is
already
a
plugin
for
VMware
and
for
kubernetes,
basically
to
go
and
query
VMware
api's
to
find
a
physical
host
information.
A
A
A
Right,
actually,
one
of
the
reasons
is
that
this
part
became
this.
This
project
and
I
got
delayed
from
the
last
quarter
is
the
fact
that
there
were
there
were
some
discussion
is
about
how
to
add
that
label
and
what
it
should
be
called
stuff
like
that
so
right.
But
someone
needs
to
follow
up
on
that
and
get
consensus
consensus
from
the
community
and
at
that
label,
and
then,
of
course
there
is
this
other
part
of
it,
which
is
communicating
with
like
a
virtualization
environment,
whatever
it
is
to
to
find
information
about
physical
notes.
A
A
C
A
A
A
A
A
D
Yes,
I'm
new
here
so
I
think
I
have
with
a
simple
question
so
for
the
D
scheduler
I
am
trying
to
figure
out
how
to
get
an
official
docker
image
created
and
pushed
to
GC
r
dot
IO,
so
I
I
guess
I'm
looking
for
some
direction
on
how
I
could
help
out
doing
that
and
will
be
the
process
to
to
get
that
completed.
Yeah.
A
I,
remember
that
I,
you
know
I
in
the
past
I
tried
to
do
this
for
rescheduling
and
I
kept
running
into
some
permission.
Issues.
I
I
think
eventually,
I
had
to
go
to
some
of
our
folks
here
at
Google
and
ask
them
to
push
it
for
me.
It
might
be
the
case,
but
the
best
person
to
speak
to
is
Ravi
Santosh.
Do
you
know
him?
You
know
because
Robby
Robby,
Rav
ie,
sorry
I
got
it
Robbie
and
Santosh's.
C
D
The
main
issue,
the
doctor
imager,
is
the
main
issue:
push
into
DC.
Aha,
so
I
submitted
a
pull
request
to
update
the
make
file,
so
the
make
file
now
has
an
option
to
build
an
image
and
it
could
push
to
GC
R.
So
the
main
permission
main
question
is:
who
would
have
the
right
permissions?
There's
an
automated
process
that
would
do
the
pushes
okay
and.