►
From YouTube: 2021-06-16 Delivery team weekly APAC/EMEA
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
A
Same
here
actually
at
my
101
with
marin-
and
I
keep
pushing
my
fan
here
on
the
side-
and
my
son
unplugged
it
yesterday.
So
it's
kind
of
nothing
is
coming
out
of
us.
B
Oh
bad
luck,
we're
due
thunderstorms
tonight,
so
today's
gonna
be
the
absolute
worst
day
like
right
before
it
all
breaks.
So
it's
a
bit
hot!
It's
like
a
bit
unusual
in
june,
though,
like
I
mean
like
last
year,
it
got
hotter
than
this,
but
in
sort
of
like
late
july,
but
well
I
hope
this
isn't
gonna.
I
hope
this
isn't
gonna
stick
about
for
the
whole
summer.
B
Awesome
so
let's
begin
so
mttp
was
a
little
surprised
by
this.
Like
did
was
last
week.
Did
we
have
lots
of
incidents
towards
the
end
of
last
week?
That's
keeping
the
mttp
number
so
close
to
the
target.
A
B
Take
a
closer
look
awesome,
so
first
discussion
item
henry.
C
Yeah
I
put
this
here
and
maybe
it's
more
fitting
to
the
current
schema
where
I'm
meeting,
but
I
want
to
talk
about
this
anyway,
because
I
was
working
on
metrics
and
observability
for
kubernetes
to
improve
what
we
can
see
there,
and
one
thing
I
notice
is
that
we
are
missing
some
metrics
that
we
had
before
when
we
still
were
running
the
vm
fleet,
especially
cpu
saturation,
because
in
kubernetes
we
can't
easily
measure
this,
because
normally
we
measure
this
against
the
node,
but
in
kubernetes
we
can
only
measure
this
against
a
container
right,
but
there
are
several
ways
to
measure
cpu
saturation
in
a
container
the
way
we
used
it
until
now
is
measuring
against
limits
like.
C
If
you
set
a
limit,
then
we
can't
get
higher
on
usage
in
a
container,
but
usually
we
don't
set
limits
for
cpu
in
a
lot
of
our
deployments,
so
we
have
no
saturation
metrics
for
cpu
in
these
cases.
The
other
way
to
do
this
is
to
measure
saturation
against
the
requests
right.
So
we
try
to
reserve
cpu
requests
for
each
deployment,
so
we
are
sure
that
we
have
preserved
enough
space
for
each
container.
C
And
so
I
made
a
list
of
things
and
services
that
should
be
fixed
and
especially
sidekick
is
full
of
of
needs
to
be
fixed
there.
And
the
same
is
true
for
a
lot
of
the
memory
requests,
and
the
thing
is
here.
This
is
blocking
the
observability
work,
because
andrew
was
of
the
opinion
that
you
maybe
should
first
fix
this
and
then
implement
the
observability
item,
because
etsy
would
start
alerting.
C
But
this
is
a
lot
of
work
and
will
take
some
time.
The
other
way
would
be
to
make
a
list
of
what
needs
to
be
fixed,
which
I
did
more
or
less
now
and
anyway
start
implementing
the
monitoring
and
saturation
metrics
and
just
silence
the
alerts
and
the
places
where
we
are
over
saturation
and
know
about
it,
and
we
can
later
work
on
this
and
then
do
it
so
two
ways
to
do
it:
I'm
not
really
decided
on
which
way
we
should
go.
We
need
to
fix
it
anyway.
B
C
It's
not
working
like
this,
because
if
I
enable
the
saturation
metrics,
it
will
be
enabled
for
all
the
services
and
these
elements,
so
we
can
just
then
start
setting
silences
and
and
prometheus
for
the
single
effective
services
which
is
easy
to
do,
and
we
would
also
get
a
notification
in
the
related
issue.
Because
andrew
has
this
helicopter
automation,
he
called
it
helicopter.
C
It's
just
looking
regularly
like
every
hour
into
our
silences,
and
each
silence
needs
to
be
linked
to
an
alert
to
an
issue,
and
then
it
will
just
put
a
note
in
the
issue.
Okay,
there's
a
silence
for
this
issue
still,
so
this
is
the
way
to
kind
of
track
silences
too.
Okay,.
B
C
B
Let's
fix
this
for
these,
but
yeah,
let's,
let's
get
a
I'll,
have
to
read
through
the
issues
before
the
demo,
but
let's.
B
B
Cool
cool
and
then
on
the
environment
one
I
just
wanted
to
make
sure
we
were
kind
of
all
knew
what's
going
on
there.
So
as
I
understand
it,
we
won't
be
able
to
deploy
the
rc,
the
14.0
and
40
rc
to
anything
that's
running,
ubuntu
16.04,
that's
end
of
life
or
on
pg12
or
pg
11,
because
that
is
currently
also
end
of
life.
B
There
is
an
issue
discussing
that,
but
I
haven't
caught
up
with
what
options
were
taken
up,
but
as
it
stands
right
now,
that
will
fail
so
pre
and
ops
are
currently
not
going
to
get
a
clean,
deploy.
B
B
C
A
Regarding
oops,
given
that
I'm
speaking
regarding
oops
ops
is
no
longer
a
problem,
because
yesterday
we
failed
the
upgrade
we
we,
we
had
a
maintenance
window
in
the
morning
to
shut
down
oops
to
run
the
app
pg
upgrade
it
failed.
So
I
just
asked
kindly
to
postpone
this
after
the
release,
because
we
can't
waste
time
because
during
that
deployment
we
we
had
no
auto
apply
no
ability
to
deploy.
A
B
Yeah,
okay,
so
I
think
on
the
pre-one
I'll
follow
up
where
we
are
with
that.
If
the
ops
one
is
pushed
out,
then
I
hope
that
we'll
get
back
onto
prey,
so
they
prioritize
ops
over
pre.
If
you
want
to
go
ahead,
I
assume
that
you
want
to
like
test
the
tagging
for
the
rc.
B
If
you
want
to
go
ahead
and
do
that
go
for
it
like
just
know
that
the
deployed
depre
will
fail,
so
we
won't
be
able
to
run
the
qa
until
this
is
done,
but
I'll
follow
up
on
those
things
today.
B
Ops
will
break,
though
right
so
is
it
intended
that
ops
will
break
overnight
on
the
22nd
because
it
will
get
the
auto.
I
don't.
B
D
D
A
A
A
A
D
C
B
C
B
D
It
also
has
so
it's
got
two
ubuntu
package
repositories:
git
lab
ee
and
gitlab
unstable
packages.getlab.com,
gitlab
unstable.
B
That's
a
good
point
yeah,
so
the
rc.
When
we
tag
the
rc,
I
think
maybe
it
goes
into
unstable,
but
it
shouldn't
use
it
like
from
when
we
saw
the
I
think
they're
just
available
there.
D
Yeah
but
the
problem
is
it
just
the
cron
job
will
just
do
an
apt
update
and
if
the
repo
is
on
the
box
and
it's
a
newer
version
number,
it
will
try
and
install
it.
That's
it.
Okay.
B
Let
me
catch
up
on
that
issue.
Then
alessia
before
you
do
the
rc
tag.
Let's
get
pre
and
ops
into
a
known
state
for
that.
B
And
then
just
so
that
we
all
kind
of
also
know
what's
happening,
so
it's
quite
a
few
things
going
on
with
pre
at
the
moment.
So
I
so
it
sounds
like
maybe
from
today,
possibly
so
yeah
so
just
wanted
to
give
heads
up
like
grandma
chatting
earlier
about
when
my
web
fleet
migration
changes
will
be
going
on
to
pre
like
after
22nd
and
then
one
for
you,
graeme
just
to
be
aware
of
that
registry
is
set
up
on
pre
in
its
special
case
as
well.
D
B
New
tests-
so
it's
probably
not
unusual
like
so
registering
on
pre,
is
in
a
special
state.
I
think
you
can.
B
A
I
mean
I
I
don't
know
I
mean
pre
gets
the
rcs
right,
so
we
install
earth
season
pre.
So
if
we
can
have
it
in
a
working
state
and
not
touching
it,
it
would
be
better
because,
basically,
the
the
idea
is
that
we
want
to
tag
earlier
and
maybe
big
fixes
and
so
doing
more
rc's,
but
removing
extra
changes
from
the
release,
because
it's
going
to
be
a
has
been
a
bumpy
set
of
weeks,
not
just
the
weekend.
B
D
Yeah,
that's
fine,
I
look
getting
it
getting
it
to
a
point
where
I'm
something
will
be
in
pre
before
the
22nd
is
probably
optimistic
on
my
part
anyway,
so
that
that's
fine-
and
I
assume
even
once
pre-
is
see-
there's
two
parts
to
me
personally.
So
there's
there's
obviously
like
the
upgrades
themselves
going
well
and
things
going
rolling
out
and
then
it's
like
if
we
start
testing
stuff
in
pre,
not
not
that
we're
really
doing
heavily
testing.
But
it's
what
am
I
trying
to
say?
D
How
do
I
figure
out
what
is
broken
because
it's
been
it's
web
migrated
to
kubernetes
versus
it's
broken,
because
this
is
a
14.0
bug
and
that
becomes
a
bit
trickier.
I
don't
even
have
an
answer
for
that
I
mean.
D
Maybe
it
doesn't
matter
because
as
long
as
gitlab.com
like
gitlab.com
is
production's
going
to
be
the
main
place,
people
will
notice
things
broken.
So
if
I'm
only
doing
stuff
in
pre
then
and
something
breaks,
we
should
be
confident
it's
more
likely
the
kubernetes
migration,
because
if
it's
not
broken
in
production,
then
I
guess
maybe
that's
the
heuristic.
We
use.
D
A
Pre,
so
I'm
looking
at
the
schedule,
so
in
theory
we
will
tag
14.0.0
monday
21st,
which
means
that
pre
will
be
in
active
usage
from
let's
say
today.
If
we
attack
today
rc42
up
until
the
21st,
because
every
if
we
need
to
pick
something
if
we
need
to
add
some
breaking
change,
that
missed
the
deadline
or
things
like
that,
we
are
going
to
cut
a
newer
c
or
if
we
spot
some
big
problem
once
we
have
14.0.0.
A
A
You
so
I
did
by
that
point
so
21st
we
no
longer
care
about
pre.
D
B
It
you'll
be
able
to
trust
the
qa
tests
at
that
point,
because
they'll
have
already
passed,
enable
for
us
to
deploy
14.0
so
they'll
be
a
good
one
as
well
awesome,
so
yeah
just
kind
of
to
wrap
up
again
on
your
point
there
and
seo
like
yeah.
If
pre's
not
running
on
pg
12,
it
will
fail
I'll
I'll
sort
that
out
today,
so
that
we
can
move
ahead
with
tagging.
C
B
It
will
fail
yeah
cool,
so
alessio.
A
Okay,
so
I
draft
kind
of
a
communication
plan
I
mean
I
just
quickly
wrote
it
before
joining
the
call.
The
idea
here
is
that,
especially
if
we
want
to
talk
earlier,
we
want
to
be
informed
if
there
are
breaking
changes
that
are
not
yet
on
gitlab.com.
A
So
is
one
option
here
is
that
we
do
rc
today's,
and
then
we
pick
missing
breaking
changes,
or
at
least
we
need
to
be
aware,
because
these
are
something
that
either
products
or
us
or
someone
have
to
deal
with
so,
and
there
was
kind
of
confusion
around
already
happening,
like
every
milestone,
things
being
re-milestone
labeled
and
things
yeah
15
has
gone
so
we
have
lost
the
window
for
14.00,
which
is
not
entirely
true.
A
Yeah,
let's,
if
we
can
review
this
and
then
we
can
broadcast
later
today,
so
that
we
just
to
reach
the
wider
audience
possible.
Yeah.
B
Perfect,
that
sounds
that
sounds
great
yeah
I'll.
Take
a
look
over
your
comment
and
just
I
might
just
tweak
a
word
or
two
but
yeah
great
idea.
Let's
do
that.
Okay,.
B
Great,
is
there
anything
else
to
bring
up
on
the
recording.
D
B
Follow
up
on
that
because
there's
already
an
issue
I'll
find
out
who's
actually
working
on
ops
and
what
what
the
plan
is
on
that
so
yeah,
if
it's
probably
beyonce,
who
does
the
actual
task
but
I'll
follow
up
on
that.
One
and
yeah.
D
D
And
I'll
make
a
note
tomorrow
morning,
my
time,
which
will
be
the
next
time
it
upgrades
so
it
upgrades
during
the
middle
of
my
day
I'll
make
a
note
to
come
in
the
morning
and
check.
I
should
be
able
to
determine
that.
It's
not
doing
that
and
if,
if
I
still
see
it
going
to
upgrade
tomorrow,
I
will
step
in
and
stop
it.
Okay,
even
just
as
a
like
a
break
fit
as
a
as
a
break.