►
From YouTube: 2021-07-19 Multi Large Working Group Weekly
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,
I
think
we
got
the
quorum
here
to
start.
Let's
get
started
today
is
july
19th.
This
is
the
multi-large
working
group
meeting
and,
what's
being
done,
amy.
B
Yes,
thank
you.
So
we
are
continuing
progressing
with
the
web
fleet
migration.
So
so
far
we've
got
pre
and
staging
migrated
and
then
what
we're
the
next
steps
will
be
to
prepare
to
move
over
canary
traffic.
So
we
are
seeing
a
few
extra
qa
failures
on
staging,
I
think
just
staging
so
we're
investigating
those,
and
once
we
have
all
our
all
the
other
checks
in
place,
we
can
we
can
start
planning
what
the
what
the
next
steps
look
like.
A
Are
those
issues
logged
into
the
issue
tracker
or
yeah,
or
it's
just
a
gnome?
Yes,
it's
just.
B
Known
at
the
moment,
they
actually,
we
believe
they
might
actually
be
registry
related
rather
than
this
migration,
but
I
can,
I
can
I'll,
go
through
and
check
check.
We
have
all
the
labels
on
those
at
the
moment.
A
Okay,
cool-
if
there
are
not
issues
for
this,
please
log
into
the
blog
as
a
new
issue,
so
we
can
follow
up
if
there
is
any
development
work
involved
here
I
can
help
follow.
B
All
be
at
the
moment,
as
we
understand
it,
it's
all
all
infrastructure
at
the
moment,
so
yeah
thank.
B
Yeah
the
question
said
not
at
the
moment,
so
the
main
task
we're
going
through
at
the
moment
is
evaluating
the
differences
between
the
kubernetes
and
vms,
so
I've
put
the
issue
in
there.
So
if
you
want
to
track
what
this
will
do
is
basically
give
us
like
in
an
ideal
world
they're
the
same,
and
then
we
know
it's
quite
a
quite
an
easy
step.
There
may
be
things
within
that
that
we
haven't
yet
uncovered
so
that
will
the
outcome
of
this
task
will
determine
our
next
steps.
A
Thank
you,
so
what's
next
is
basically
described
here,
continue
to
work
on
the
webinars
migrating
to
over
migrate
to
some
canary
traffics.
That's.
A
Cool
blockers.
B
Yes,
so
I
wanted
to
just
highlight
this
one,
so
we
we
don't
lose
it
so
we
we
have
a
mitigation
that
we'll
use
around
the
dependency
proxy
migration
needed,
but
it
is
just
a
intended
to
be
a
short
term.
So
we'll
take
that
back
out
again,
once
we
have
the
actual
real
fix
in
place.
A
Okay,
actually,
the
the
work
to
be
done
was
broken
into
two
sub
issues
under
the
under
that
issue.
So
I
just
the
pain
I
think
team
and
team
lizzie
and
dan
croft
so
follow
up
to
schedule
those
sub
issues
for
right.
Now
the
the
master
issue
is
scheduled
for
143.
I
will
follow
up
to
make
sure
that
that
happens
to
the
two
sub
issues.
B
Great
yes,
so
just
wanted
to
follow
up.
So
I
know
we've
got
the
goodly
should
run
well
on
kubernetes
issue
and
we've
started.
Delivery
are
going
to
start
thinking
about
perfect
and
see
what
we
can
like
what
might
be
involved
in
us
moving
those
that,
over
now
at
the
moment,
in
terms
of
what
it's
a
kind
of
a
learning
task
from
our
side
to
understand
like
how
this
might
have
work
in
kubernetes.
How
deployments
may
be
affected,
but
I
know
we've
talked
a
lot
around
performance
impact.
B
C
D
C
Okay,
it
doesn't
sorry
super
weird
anyway
yeah,
so
when
we
rolled
out
prefecture
production,
we
put
it
on
the
gita
box,
which
was
insane
in
terms
of
costs
of
what
it
was
capable
of
and
we've
put
it
on
a
diet
twice
and
twice.
No
one
noticed,
but
at
the
same
time
no
one
felt
it
was
important
enough
to
put
it
on
a
diet
and
diet
another
time
so
right
now
we
have
no
formal.
C
Well,
we
have
a
minimal
bounce,
but
we
don't
have
a
practical
bound
for
bigger
rollouts
as
production.
The
only
bounds
we
do
have
are
through
synthetic
loads
through
like
qa
and
stuff.
C
A
Okay,
amy:
does
that
give
you
more
context
here.
B
Maybe
so
I
think
it's
a
I
suppose.
It's
a.
B
B
C
Yeah,
so
I
spoke
with
scarbeck
this
week.
It's
stateless
like
as
long
as
italy
is
deployed.
First,
we
do
not
care
about
prefect
from
a
giveaway
perspective.
Nor
does
rails
know
there
is
even
a
git
prefect,
so
it
really
is
stateless
and
the
only
dependency
it
has
is
that
there
is
a
giveaway
behind
it,
which.
C
Correct
nope:
well
like
don't
put
it
on
the
in
venezuela
or
something
which,
with
that
internet
but
yeah.
If
it's
the
same
database,
it
should
be
fine.
E
B
Right:
okay,
okay,
cool,
okay,
well,
we'll
see
we'll
see
how
this
goes
then,
so,
we'll
just
try
and
get
it
stood
up
somewhere
and
see
what
what
plays.
A
Out
and
my
question
to
cj
is
how
has
the
the
testing
going
I've
been
going?
We
have
been
talking
about
it
for
a
few
months,
so
just
the
touch
base,
how
the
testing
looks
like.
C
Yeah,
so
I
I
run
it
locally,
but
like
falsifying
that
something
doesn't
work,
I
I
think
it's
rather
hard.
Like
the
obvious
failures,
I
did
like
what.
If
it
runs
out
of
memory,
it
starts
to
drop
a
lot
of
packages.
I
I
felt
that
was
obvious,
but
I
so
far
I
haven't
seen
reasons
it
should
not
like
prefect
should
not
work
in
a
kubernetes
environment.
C
I
did
see
an
issue
yesterday
know
today
from
jason
well
actually
from
a
customer
about
routing
and
dns.
I
don't
think
this
is
kubernetes
specific,
just
the
way
grpc
works
and
from
what
I've
seen
gitlab.com
is
a
much
better
heuristic.
If
it
works
in
kubernetes,
then
anything
else
would
be
like
the
review.
Apps
right
now
seem
to
work.
Well,
there
are
some
flakiness
issues,
but
it's
really
hard
to
get
data
out
of,
but
otherwise
I'm
not
aware
of
big
blockers
for
either
prefect
or
italy.
In
kubernetes.
C
I
did
talk
to
scarbeck.
Like
I
just
mentioned.
Both
of
us
were
like
if
we're
gonna
do
this
on
dot
com.
Let's
start
with
a
stateless
surface
like
prefect
over
italy
and
yeah,
it
seems
much
easier
and
less
costly
to
learn
our
things
there
over
italy,
where
we
could
actually
lose
data.
A
That's
cool:
do
you
actually
need
some
kind
of
environment
seeming
another,
your
local
environment,
but
it's
like
a
similar
to
our
staging
or
or
production
environment,
to
test
that
or
local
testing
is
sufficient
to
proceed
to
next
steps.
C
For
me,
so
far,
local
testing
has
been
sufficient.
I
haven't
tested
under
load,
I
think
maybe
grant
has.
I
don't
know.
E
E
C
In
this,
but
I
would
just
download
this
to
production
and
and
the
term
yolo
is
very
sorry.
I
shouldn't
have
used
that,
but
I
I
am
confident,
let
me
put
it
that
way.
F
We've
not
tested
it
in
the
charts
because
in
the
charts
they're
still
listed
as
alpha.
If
you
want
to
move
that
forward,
then
certainly
we
can
look
into
that,
but
we
have
tested
it
extensively
on
the
bus
and
we
found
it
to
be
quite
a
performant
little
component.
Really,
I
agree
with
cj.
I've
got
no
major
concerns
about
running
in
kubernetes.
Obviously
you
want
a
little
bit
of
buffer,
but
it's
it
doesn't
go
crazy
when
it's
memory
or
cpu.
A
Sounds
good,
so
our
looks
like
our
next
step
will
be
testings
in
production
in
some
way.
B
We
can
use,
we
use
a
staging
or
pre
or
one
of
our
test
environments
to
see
what
this
looks
like.
We
need
to
make
sure
that
we
can
deploy
it
and
run
it,
and
also
that
we
kind
of
can
handle
backwards
and
force
compatibility.
You
know
like
italy
and
things
being
on
different
versions
as
well,
so
we'll
get
that
stood
up
somewhere.
B
Is
there
is
there
somewhere
zj
that
we
can
actually
see
the
tests
and
the
sort
of
things
like
is?
Are
these
comments
in
an
issue
anywhere.
C
No,
I
just
spun
up
a
kubernetes
cluster
and
tried
to
charge
and
just
hacked
myself
into
I'm
happy
to
share
config
what
I
tried,
but
I
just
tried
to
use
whatever
jason
has
been
building
over
the
past
two
or
three
years.
Yeah.
B
A
D
Yeah
we're
right
at
the
beginning
of
14
2,
and
so
I
wanted
to
just
give
a
call
out
that
we
don't
have
any
scheduled
item
issues
for
specific
to
the
multi-large
working
group.
There
is
a
single
epic
open.
That's
for
scheduling!
We
have
not
scheduled
any
of
the
work
in
that
again,
no
real
discussion
here,
but
I
just
wanted
to
raise
awareness
of
what
distributions
focused
on
this
coming
release.
A
Thank
you.
Okay,
any
questions.