►
From YouTube: 2021-12-06 Multi Large Working Group
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
Okay
good
morning,
good
afternoon,
good
evening,
everyone
today
is
december
6th
and
this
is
the
kubernetes
migration
working
group
and,
let's
get
started
from
the
agenda.
The
first
all
has
been
done
and,
of
course,
our
one
big
achievement.
Last
week
we
completed
the
playlist
migration.
So
congratulations.
B
Yes,
thank
you.
So,
yes,
we're
just
wrapping
up
on
the
tidy
up
of
pages
and
then
we
can
close
out
the
epic.
So
we
have
a
couple
of
things.
We
do
have
a
bug
to
fix
stackdriver.
B
We
don't
think
it's
like
critical
enough
to
have
triggered
a
rollback,
so
we've
left
it,
but
we
will
be
fixing
that
as
soon
as
we
can-
and
we
just
have
to
clean
up
tasks
to
tear
down
the
old
vms
before
we
close
out
the
epic
when
we
do
close
this
epic,
that
is
actually
the
conclusion
of
the
stateless
service
migration.
A
That's
cool
anything.
You
need
help
from
the
development
team.
A
Thank
you,
okay.
Next
one
still,
yours,
yes,.
B
So,
as
we
conclude
on
the
stateless
service
migration,
we
have
now
just
started
work,
so
we're
pairing
up
delivering
a
scalability
of
paired
up
in
q4
and
we're
looking
at
migrating
red
assets.
So
the
first
one
we're
going
to
look
at
is
rate,
limiting,
which
is
one
of
our
most
simple
redisters.
B
But
the
goal
is
to
try
and
kind
of
understand
how
to
get
redis
running
safely.
In
kubernetes,
like
the
migration
is
the
the
trickier
part
rather
than
the
actual
operating,
and
then
the
goal
will
be
to
use
kubernetes
to
kind
of
address.
Our
redis
scalability
concerns
so
we're
still
in
the
investigation
stage.
Right
now
we
do
have
rate
limiting
reddit
running
on
kubernetes
and
pre,
and
from
there
we
can
do
testing
the
main
thing.
B
We're
testing
on
is
performance
and
just
seeing,
if
there's
any
like
slow
down
when
we
run
radius
on
kubernetes.
But
then
one
of
the
big
areas
that
it's
still
a
little
question
mark
at
the
moment
is
the
chart.
So
for
now
we're
setting
things
up
with
the
bitnami
chart,
but
we
intend
to
come
back
with
kind
of
the
requirements
of
what
would
it
take
to
get
rid
of
supported
in
the
get
up
chart?
B
A
Cool
any
any
concerns
of
of
using
vietnamese
charts,
stephen
or
jason
jason
here.
C
C
And
no,
I
have
no
concerns.
I
actually
recommended
that
myself.
C
That
is
actually
how
we
did
the
implementation
for
multiple
redis
when
we
did
the
actual
support
for
it,
I'm
still
trying
to
figure
out
the
right
safe
way
to
do
it
in
terms
of
how
to
have
multiple
instances
of
it
actually
within
the
primary
chart
is
going
to
be
some
convoluted
logic
to
make
that
work.
B
Is
there
a
is
there
a
kind
of
a
plan
to
get
multiple
reddits
running
in
the
gitlab
chart
like?
Is
that
something
on
the
on
the
roadmap.
C
So
the
functionality
to
configure
it
is
a
part
of
the
chart,
but
not
to
deploy
it
right.
Now,
we've
been
holding
off
on
that
largely
because
we
actually
recommend
to
not
run
redis
in
kubernetes,
so
we've
just
kind
of
shied
away
from
actually
implementing
that
functionality
within
the
chart
itself.
Strictly
because
we
say
run
your
state
outside.
B
Cool
okay,
great,
okay,
so,
hopefully
yeah.
We
hope
to
learn
more
about
about
how
we
run
redis,
particularly
when
we
go
to
the
slightly
more
involved
ones.
And
then
we
can
hopefully
feed
that
in.
A
Cool,
we
don't
have
bloggers
and
the
discussion
item.
I
mean
you
want
to
verbalize
that
item.
B
Yes,
I
was
just
wondering
I
was
just
wondering
with
the
recent
giveaway
event
whether
this
has
altered
our
timeline
either
brought
it
forwards
or
pushed
it
back
around
the
sort
of
the
testing
and,
I
suppose,
like
motivation
to
migrate,
get
lead
to
kubernetes.
E
Good
question:
with
all
that's
going
on
with
italy:
right
now
I
mean
we've
had
a
large
push
to
get
testing
done
for
a
couple
of
our
larger
customers,
we're
trying
really
hard
to
ensure
stability
for
our
customers.
I
think
we've
made
huge
strides
there.
We've
done
some
randomization
testing
tanya
and
her
team
have
done
a
great
job
of
helping
with
that.
The
team's
really
work
together,
but
that
does
mean
yes,
it's
come
at
the
expense
of
not
working
on
features
this
month
and
minimal
progress
in
other
areas.
E
Just
because
we've
had
some
some
high
priority
incidents
that
needed
to
come
first,
so
I'm
hoping
to
get
that
work
back
on
track
here
come
the
new
year.
We
do
have
an
engineering
manager
starting
as
well,
so
there
will
be
additional
support
there,
which
will
be
great
but
yeah
in
the
last
month,
we've
probably
done
very
little
in
the
way
of
kubernetes,
just
because
of
the
incidents
that
are
going
on
in
the
testing.
D
B
Oh
so
far,
lovely
to
hear
that
yeah
brilliant
do
we
know,
like
is
the
kind
of
expectation
that
having
italy
on
kubernetes
would
really
help
with
stability
like
is
there?
Is
there
more
urgency
around
this.
E
I
don't
think
we've
tied
the
ability
to
be
running
on
kubernetes
to
stability
or
not.
I
think
the
goal
of
kubernetes
is
primarily
that
it
will
allow
us
to
have
more
deployment.
Hopefully
within
gitlab
ourselves.
We
can
dog
food
things
easier
and
that
in
and
of
itself
is
a
very
large
attraction
to
us,
because
a
lot
of
these
issues
we're
having,
if
we're
not
dog
food
against
scale,
that
some
of
our
customers
are
at,
which
is
an
unfortunate
situation,
so
we're
finding
scalability
issues
at
customer
sites.
E
So
we
would
love
to
get
done
because
it
does
mean.
Hopefully
we
can
start
scaling
larger,
get
lab
and
we'll
be
catching
things.
So
there's
that
motivation,
which
is
why
it
is
a
top
priority
for
us
when
we
don't
have
a
customer
issue
ongoing.
So
I
don't
think
we're
expecting
better
stability
with
kubernetes.
Necessarily,
I
don't
think
kubernetes
lends
additional
stability
to
the
gitly
product,
but
what
it
does
do
is
allow
us
to
deploy
at
scale
and
that
will
help
us
find
issues
and
increase
stability.
That
way.
C
E
E
Well,
we
can't
have
it
happen
for
our
large
customers
either,
so
we're
in
a
bind
where
we
have
to
give
somewhere-
and
I
I
understand
your
concerns-
I
completely
agree
with
them,
but
we
can't
have
our
customers
raising
priority
once
every
two
weeks
either.
That's
an
unacceptable
thing
for
our
business
as
a
whole.
It
hurts
our
reputation
immensely
when
our
reference
customers
are
raising
at
p1s
all
the
time.
So
we
have
a
difficult
proposition
here.
We
just
need
to
figure
out
the
balance.
E
C
C
Second
is
actually
iops.
Okay,
you're
competing
with
everything
on
that
node
for
iops,
against
the
desk
that
you're
trying
to
get
to
and
no
you
can
get
dedicated,
nvme,
backed
storage
that
that
has
guaranteed
iops.
But
the
nodes
kernel
still
has
an
upper
limit
of
what
it
can
actually
get
done.
Even
if
you
have
a
large
allocation
on
the
remote
storage
or
even
the
local
disk
storage.
There's
still
a
limit
to
how
much
a
noisy
neighbor
can
impact
that.
E
I
just
want
to
make
sure
we're
not
trying
to
put
a
square
peg
in
around
here
kubernetes,
fantastic
for
lots
of
things.
If
gita
is
not
one
of
them,
then
I
don't
see.
I
don't
want
to
waste
a
ton
of
time,
finding
that
out,
I'd
rather
find
that
out
and
then
stop
so
as
long
as
we're
aware
of
that
moving
forward,
and
if
we
run
into
a
situation
we
say
this
is
just
a
bad
fit.
E
E
E
E
Yeah,
the
customer
incidents
wouldn't
have
been
fixed
in
kubernetes
or
not.
They
were
not
deployment,
specific
problems
per
se.
They
were
scalability
issues
that
had
to
do
with
lots
of
things.
I
don't
think
having
us
in
kubernetes
or
not
would
have
been
affected
by
those,
but
I
do
agree
that
the
testing
kubernetes
is
important.
A
Cool,
I
think,
amy.
I
think
that
now
the
testing
is
really
prerequisite
to
for
moving
deeply
over
kubernetes.
If
I
turn
out
to
be
a
no-go,
we
may
want
to
revisit
this
topic
further
after
the
testing.
A
Okay,
cool!
That's
all
about
this
topic.
A
The
last
one
is
follow-up
item,
so
the
listen
proxy
for
the
pages
that
issue
and
the
mr
what's
up
thanks
for
john
scarbeck,
so
I
think
his
last
comment
was
this
is
coming
shortly,
so
we
can
wait
for.
B
That
yeah,
I
expect
we'll.
I
expect
we're
not
far
away
he's,
probably
prioritizing
around
release
management
things,
but
yeah
we're
probably
hopefully
later
this
week
or
early
next
week,
be
able
to
start
this.
A
Yeah
and
my
request
to
the
distribution
team
is
to
how
to
review
the
mr
getting
merged
when
it's
ready.
C
Jason,
we
have
reviewed
it,
the
mr
does
what
it
says
it
does.
There
is,
however,
a
bug
in
pages,
so
we
have
a
bit
of
a
problem
in
that
the
configuration
behaves
as
designed,
and
we
can
prove
that
it
does
as
designed.
However,
if
you
turn
it
on,
it
doesn't
work
so
we're
in
a
bit
of
a
bind
in
that
regard.
C
B
Yeah,
like
I
mean
totally
fine,
to
pick
that
back
jason,
like
so
vlad
isn't
here
today,
but
we've
been
working
with
directly
with
flood,
but
also,
I
think,
other
people
from
who
know
pages,
who,
I
guess,
we'll
have
to
go
back
and
ask
for
a
bug
fix
before
we
can
enable
this.
C
B
A
Yeah
get
the
issue
up
and
I
can
also
coordinate
the
work
with
the
pages
of
biggest
team.