►
From YouTube: 2021-01-19 Delivery team weekly
Description
Delivery team discussion of ideas for Q1 OKRs
A
Hello,
so
welcome
to
tuesday.
I've
got
in
a
slightly
extended
time
for
this
meeting
today,
just
so
that
we
can
get
through
a
bit
of
discussion
on
okrs.
We
may
not
need
all
the
time,
and
so
you
may
get
some
time
back.
We'll
certainly
finish
on
time,
so
this
will
be
time
box
so
that
we
have
a
little
bit
of
time
at
the
end
to
go
through
our
kind
of
social
chat
as
well.
A
So
if
we
need
more
time
afterwards,
we
can
have
that
or
we
can
go
async,
but
we
shall
see
how
we
go
so
there's
a
few
announcements
in
there.
Please
have
a
read
through
and
well.
No,
the
only
one
I'm
not
going
to
have
is
read.
Only
is
congratulations,
myra,
I'm
getting
a
bonus!
Thank
you
for
your
help
on
these
things.
A
Also,
my
slack
is
still
melting
because
there
are
so
many
emoji
responses
to
that
announcement.
So
that's
awesome
cool.
So
what
I
mostly
want
to
focus
on
today
is
okrs.
If
anyone
has
other
things
you
want
to
discuss,
please
just
add
them
in
the
agenda
we'll
get
to
those,
but
firstly
about
okrs.
A
So
thanks
for
the
feedback
in
the
retro
from
the
previous
quarter,
and
thanks
for
the
work
you've
already
done
to
get
us
the
stage
so
far,
so
a
few
little
things
around
okl's
before
we
dive
into
the
discussion,
okrs
won't
be
the
only
work
we'll
do
in
the
quarter,
but
they
are
there
to
help
us
prioritize
so
they're
there
for
us
to
be
able
to
make
decisions.
If
something
drops
in,
we
actually
know
what
we're
having
to
prioritize
that
new
piece
of
work
against.
A
Hopefully,
it
gives
you
a
bit
of
a
way
as
well
to
if
people
make
requests
of
you
to
be
able
to
explain
to
them
like
what,
where
you
have
focused
on
working
on
as
well
as
sort
of
feeling
like
you,
don't
have
to
do
everything,
so
this
is
one
way
we
hopefully
be
able
to
limit
things
a
little
bit
and
generally
with
the
okrs.
They
are
meant
to
be
ambitious
enough
for
like
to
cover
a
whole
quarter.
A
It
doesn't
have
to
be
that
a
single
key
result
covers
a
whole
quarter,
but
the
objective,
so
we
can
have
different
focuses
of
key
result
within
a
single
objective,
but
the
goal
itself
is
we're
generally
aiming
to
hit
70
percent.
A
If
we
consistently
hit
100
on
all
of
our
okrs,
then
we
need
to
be
a
little
bit
more
ambitious.
If
we're
consistently
below
on
that,
then
maybe
we
need
to
rein
in
our
ambitions
just
a
little
or
consider
like.
Are
we
prioritizing
carefully
enough
or
things
like
that
so
generally,
as
we
go
through
like
if
you're
in
your
mind,
trying
to
work
out
what
the
right
size
is
have
a
think
about
something
at
seventy
percent?
A
So
the
main
thing
I
wanted
to
cover
today
is
a
bit
of
a
discussion
around
how
we
rank
these
things,
which
we
didn't
do
at
all
last
quarter.
So
last
quarter
we
were
kind
of
at
this
stage
and
we
had
some
ideas
and
then
we
more
or
less
just
kind
of
went.
Okay,
they
look
good
on.
We
go
so
two
things
we
didn't
do.
One
was
discuss
other
options
and
two
was
discuss
the
size
of
the
things
and
were
they
clear
enough
for
us
to
to
pick
up
for
the
next
quarter?
A
Cool,
so
what
I've
done
with
all
the
scoring
is
try
to
put
these
together
into
what
is
our
team
view
so
these
categories,
some
of
them
were
very
deliberately
opinions
because
that's
kind
of
the
point
at
this
stage.
So
this
is
a
sort
of
view
of
here.
Are
our
team
opinions
of
things
so.
A
So
what
I've
done
is
for
each
of
these
categories.
The
top
ranking
idea
is
slightly
darker,
green
and
also
bolded
number
and
then
the
second
and
third
option
are
slightly
lighter
green
there's
one
exception
where
we
have
three
with
the
same
score,
so
I've
just
picked
the
second
third
and
fourth.
So
what
we
can
see
here
is
quite
interesting
as
a
kind
of
personal
opinion
across
the
whole
team,
the
continuing
the
kubernetes
migration
ranks
top
and
then
we
have
quite
consistently
high
results
on
rollbacks
and
also
on
changelog
plus
security
backpacks.
A
A
Well,
let's
start
off
with
this
with
rollback
right,
so
I
picked
that
one
just
because,
as
we
haven't
got
jarvan,
I
would
dive
straight
into
kubernetes
and
also
scarbec,
knowing
you're
not
100,
so
we
maybe
loop
back
to
that
one,
but
for
the
rollbacks
is
there
anybody
who
feels
either
that
this
wouldn't
be
a
great
one
to
take
for
q1?
A
So
specifically
like
do
you
feel
we're
ready
like
not
to
say,
do
you
believe
we
should
do
it
ever
or
not
yeah
so
does
anyone
feel
like
from
their
scores
that
that
would
be
risky
or
a
bad
idea?
For
example,.
B
So
I
think
we
are
generally
ready
instead
sounds
like
abs
and
we
we
can
do
it.
I
think
the
concern
thus
far
has
always
been
like.
Oh
we've,
never
done
it.
So
you
know
what,
if
this,
what
if
that
and
yeah,
I
don't
think,
there's
a
way
we
can
sort
of
get
rid
of
that
unless
we
actually
do
it
or
at
least
try
to
do
it.
C
So
how
do
we
actually
plan
to
roll
back
like
in
a
broad
level?
How
is
that
going
to
work.
D
So
the
general
idea
is
that
there
is
kind
of
go
no
go
point
which
is
around
post
deployment
migration.
So
if
we
want
to
roll
back
prior
to
post
deployment
migration,
then
it
is
allowed
if,
after
the
job,
we
need
a
way
to
figure
out
if
there
we
actually
run
post-deployment
migration
or
not.
But
the
idea
is
that
we
roll
back
code
but
not
database.
D
D
So
it's
basically
we
are
kind
of
running
the
same
situation
with
cannery,
so
we
and
we
end
up
with
next
database,
so
schema
and
plus
one
with
version
with
code
version
n.
A
And
one
of
I
suppose,
one
of
the
big
things
as
well
is
going
to
be.
We
all
need
some
percentage
of
coordinate
deployments
completed
as
well,
so
this
ties
in
would
people
feel
comfortable
having
something
along
the
lines
of
the
objective
being
basically
like
this
isn't
the
wording,
but
something
that
basically
says
like
run
a
rollback
or
something
that
kind
of
sets
us
up
that
near
the
end
of
the
quarter.
A
Answer
these
questions
like
work
out
like
how
on
earth
would
this
work
kind
of
assuming
that
that
will
be
about
a
lot
of
the
work
and
then
maybe
sort
of
like
it's
actually
sort
of
nearer
the
end
of
the
quarter
that
we
would
actually
even
attempt
to
see
a
rollback
happening.
D
Well,
I
think
it
makes
sense
it.
We
may
also
be
able
to
run
it
earlier.
I
mean
it's,
it
really
depends,
probably
is
just
like
a
deployment
without
certain
jobs
being
run,
so
it's
kind
of
deploying
again
the
old
version
without
running
migrations,
and
things
like
that.
So
once
we
have
them
a
an
easier
pipeline,
so
the
deployer
is
easier
to
handle
because
it
doesn't
have
multi-environments
and
things
like
that,
and
we
have
more
control
over
how
we
trigger
it.
A
Okay,
that
sounds
interesting,
so
we're
saying
that
kind
of
our
objective
will
be
a
lot
of
it
will
be
around
the
actual
working
out
of
how
we
do
this.
A
Cool
okay,
great
and
that
say
rollbacks
has
we
believe
it
has
most
significant
contribution
towards
site
reliability,
which
is
great.
It's
going
to
be
a
huge
focus,
continues
to
be
a
huge
focus
for
the
whole
department
and
then
slightly
lesser,
like
I
guess
more
as
we
become
comfortable
with
how
it
works,
it
will
contribute
to
mttp
and
also
make
release
management
easier.
So
nice
does
anyone
want
to
add
anything
else
to
rollbacks
before
we
move
on.
A
So
let's
discuss
the
changelog
because
I
think
that
one
has
is
our
top
dog
fooding
effort
and
it's
also
our
top
useful
for
the
community
and
it's
in
progress
so
all
good
stuff.
A
B
I
would
imagine
it's
probably
more
leaning
towards
like
half
a
quarter,
so
one
one
and
a
half,
maybe
two
months,
depending
on
how
it
goes
with
this
quarter.
B
So
my
current
goal
is
basically
to
get
a
api,
and
some
state
functioning
by
this
quarter
probably
still
behind
a
feature
flick,
but
at
least
it
will
be
there
and
then,
roughly
speaking,
I
would
say
the
first
month
of
the
second
quarter.
We
would
spend
testing
it
on
our
own
projects,
starting
with
release
tools.
B
B
A
Cool
okay,
so
dog
fooding
is
gonna,
be
like
I
mean
it's
been
a
focus,
I'm
sure
you've
all
noticed
for
the
last
few
months.
So
if
we
get
asked
about
it
quite
frequently,
it's
going
to
become
increasingly
so
so
would
it
make
sense
considering
any
of
our
other
dog
fooding
items.
So
we
have
these
three
around
like
it
as
a
kind
of
in
addition,
since
this
maybe
isn't
a
full
quarter
just
to
make
it
clear,
we
have
three
which
rank
quite
high
as
dog
fooding.
A
D
D
D
Should
not
focus
on
that?
The
other
two
are
doing
the
same
thing
in
different
ways.
I
am
kind
of
leaning
toward
not
using
the
feature
flag,
so
the
thing
is
using
fissure
flag.
We
are
dog
fooding,
but
we
are
not
adding
a
new
valuable
feature
with
the
product
and
there
is
something
that
I
don't
like
it
is
that,
depending
on
how
we
implement
it,
we
may
end
up
having
feature
flipping
during
ci
job
run,
which
is
not
what
happened
with
variables
with
variables.
D
A
A
How
would
people
feel
about
us
having
something
a
little
bit
more?
A
I
guess
almost
like
investigation,
because
I
think
one
of
the
things
that
we're
coming
up
across
a
little
bit
at
the
moment
is
the
constant
decision
of
do.
We
build
this
in
like
release
tools
or
add
our
own
code,
or
is
there
a
feature
already
so
how
about
we
had
something
where
a
key
result
was
either.
A
A
I
mean
we
could
do
that
separately.
I
guess
one
of
the
things
that's
in
my
mind
is
it's
been
interesting
as
we've
gone
through
the
changelog
work,
particularly
as
we've
been
trying
to
work
out,
the
edit
commit
message
feature
there's
a
bit
of
a
blur
about
who
does
the
work
and
who
we
need
to
get
on
board
with
the
ideas.
A
So
it
does
seem
that
there's
probably
some
upfront
work
we'll
need
to
do
for
future
dog
fooding
efforts,
so
we
could
maybe
focus
on
that,
like
not
necessarily
as
an
okr
and
just
you
know,
make
our
kind
of
code
efforts
around
finishing
off
change,
logs
and
roll
backs
plus
other
stuff.
E
And
I
think
ci
labels
is
one
of
the
ones
that
I
feel
like
falls
into
that
category.
But
it's
horribly
useful
for
everyone
else
who
searches,
ci
pipelines
a
lot.
A
Yeah,
that
might
be
a
good
idea
actually
like
again,
I
wasn't
really
sure
where
that
f,
whether
that's
in
a
backlog
already,
but
it
certainly
doesn't
seem
like
it's
necessarily
something
that
is
controversial
enough,
that
you
know
we
wouldn't
be
able
to
work
that
out
in
a
quarter
so
yeah.
That
might
be
a
good
idea.
A
D
The
merge
requested
marin
is
working
on
and
I
am
slightly
concerned
by
timing,
because
this
is
gonna,
be
we
are
talking
about
quarter
one
fiscal
year,
2022,
okay,
so,
in
the
current
shape
of
that
merge
request,
we
have
several
things
that
we
want
to
ship
by
the
end
of
quarter,
2
and
just
namely
complete
kubernetes
migration
of
stateless
services
and
what
it
is
defining
deployment
as
a
low,
and
no.
D
This
is
just
these
are
the
only
two
that
we
have
for
quarter
two
so
which
means
that
for
doing
this
we
have
this
quarter
and
the
next
one,
and
then
for
quarter
three
we
have
rollback,
completed
and
working
and
for
quarter
four,
we
have
stateful
services
migrated
to
kubernetes,
plus
converting
for
release
workload,
2
features
so
dogs
for
dog
fooding
effort.
D
This
seems
a
bit
too
much
because
I
mean
it's
it's
four
quarters,
so
it
means
that
every
quarter
we
need
to
ship
dog.
We
need
to
dog
food,
something
plus
we
have
two
quarter
two
going
from
not
having
rollback
to
having
kind
of
stable
and
actually
enjoyable
rollback
system,
plus
kubernetes
and
defining
deployments
low
as
hello,
which
is
kind
of
yeah,
a
lot
of
work,
even
in
terms
of
understanding,
measuring
and
figuring
out
what
we
want
to
do
so
this
was
kind
of
driving
my
scoring
here
to
try
to
figure
out.
D
A
Yes,
I
agree,
I
think
I
mean
particularly
on
rollbacks
and
kubernetes,
I
think
they're,
the
ones
the
scoping
is
harder
to
flex
on
for
the
dog
feeding
one.
A
I
think
we
can.
I
think,
there'll
be
some
things
where
product
teams
will
come
up
with
features
that
we
want
to
adopt.
There'll
be
other
things
which
or
maybe
reasonably
small-
they
wouldn't
be
a
quarter.
A
I
wouldn't
think
it'll
be
four
things
as
big
as
the
changelog,
because
yeah
that
would
be
pretty
ambitious,
but
I
think
it'll
be
a
case
of
us.
I'm
actually
wondering
whether
the
number
may
end
up
being
higher
but
they'll
be
a
lot
smaller.
A
So
we
might
need
to
just
tweak
on
that
a
little
bit,
but
I
agree
on
the
rollbacks
I
mean
the
other
thing
to
contribute
towards
rollbacks
is
our
current
mttp
target
is
12
hours
and
it
won't
be
lowered
until
we've
been
hitting
that
for
three
months,
which
actually
we're
getting
very
close
quite
frequently,
but
it
will
be
that
plus
rollback.
So
actually
we
can't
make
much
more
impact
on
mttp
without
having
a
rollback,
so
okay.
A
So
what
I'm
hearing
here
is
that
we
don't
want
to
add
anything
more
to
dog
fooding
at
this
stage.
Is
that
correct.
D
Yeah
but
I
second
your
idea
of
kind
of
having
some
kind
of
pre-work
state
for
because
especially
for
the
things
that
requires
that
requires
pms
right,
because
if
we
are
adopting
something,
then
it's
just
up
to
us.
But
when
we
are
actually
coding
the
feature
ourselves,
then
he
needs
interaction
with
pms
and
things
like
that.
So
it
may
take
some
work
up
front.
Otherwise
we
just
don't
have
time
to.
A
Yeah,
okay,
that
makes
total
sense
like
I.
I
fully
intend
to
kind
of
kick
off
on
that
stuff
and
pull
you
in
as
needed,
but
yeah
I
agree
like
it
doesn't
have
to.
I
think
we
can
use
it
to
set
ourselves
up
for
future
quarters
nicely:
okay,
great,
okay
and
then
the
other
big
one.
I
think
that
that
makes
worth
discussing
is
security,
automating
security,
so.
A
So
I
suppose
two
things
one
is:
how
does
it
feel
if
we
just
don't
do
this.
C
D
A
Yeah,
I
know
I
mean
I
think
that
it's
I
could
certainly
having
been
through
it
recently
like.
I
can
definitely
see
the
benefits
like
after
having
it,
but
I
still
think
people
are
remember
the
recent
security
process.
So
in
a
way,
it's
almost
still
in
a
it's
a
positive
we've
made
big
impacts
on
this.
I
don't
know
how
much
additional
work
will
will
add
a
much
more
value
onto
that
one.
D
I
remember,
and
it
is
done
I
mean
it's
just
a
matter
of
babysitting
pipelines
and
things
like
that,
but
pipelines
will
still
be
there,
so
all
the
edge
cases,
all
the
fact
that
we
want
to,
if
I
remember
correctly,
there's
this
idea
of
trying
to
to
cherry
peak
ahead
of
time
so
that
we
know
if
the
merge
request
is
still
say
compatible
with
the
security
branch
or
with
the
the
target
branch,
the
backboard
branch
or
not.
So
this
guy.
D
All
these
kind
of
things
are
super
nice
to
have,
but
it's
a
lot
of
stuff
built
on
top
of
our
tooling
system.
It's
a
lot
of
code
to
manage,
and
it's
just
really
how
much
time
is
it
saving
for
the
engineers
involved
into
this,
because
when
you
are
working
on
a
security
fixes,
usually
this
is
your
main
goal
right.
So
you
have
as
an
engineer
you
have
all
the
your
top.
D
Priority
is
working
on
that
you
put
all
the
attention
on
that
fix
and
this
kind
of
yeah
it's
a
minor
part
of
the
of
the
process,
the
providing
the
the
backboards
and
it's
it's
a
problem
when
the
the
peak
doesn't
well
when
you
can't
pick
it,
but
in
that
case
no
tooling
can
help
you.
It's
just
up
to
you
to
figure
out
how
to
code
that
fix
for
the
old
version.
C
C
It
is
hard
for
them
to
find
what
stable
version
they
need
to
target.
It
is
also
hard
to
have
them
approved
by
the
maintainer
and
then
assign
them
to
the
bot.
I
am
not
saying
it
is
difficult
itself,
but
I
think
it
is
tiresome
because
they
have
to
do
that
three
times
but
yeah.
I
I
think
it
is
fine
not
to
do
this.
This
quarter,
I
mean
we
have
been
living
like
this
since
forever.
C
We
just
wanted
to
make
clear
that
it's
also
not.
That
is
very
easy
for
developers
to
actually
create
backboards.
A
Yeah,
I
think
it's
a
great
point,
and
actually,
I
think
one
other
thing
that
makes
me
want
to
defer
this
actually
is
that
if
we
are
possibly
saying
that
this
might
change
at
some
point
this
year,
and
actually
the
number
of
releases
we
backport
to
theoretically
could
change
that
might
add,
or
also
remove
the
need
to
be
automating
these
back
ports
right.
If
we're
suddenly
going
back
a
lot
more
versions,
maybe
automating.
It
is
the
wrong
approach,
because
we
won't
get
so
much
benefit.
A
If
there
are
small
things,
though,
that
we
can
do
to
make
it
easier
for
people
to
find
the
right
branches
and
know
what
they
need
to
do,
particularly
like,
if
they're
trying
to
like
create
them
ahead
of
us,
cutting
the
monthly
release
like
there's
nothing
to
stop
us
doing
that
stuff
right,
that's
not
an
okr
sized
piece
of
work,
so
we
can
still
think
about
whether
there's
like
smaller
improvements
we
can
do
before
we
get
to
this
work.
A
Great
and
then
so,
the
only
other
one
which
we
have
as
a
kind
of
big
question
mark,
is
around
observability
for
kubernetes,
so
scobic,
I
know
you're
not
feeling
100.
So
what
I
was
going
to
briefly
propose
is,
I
would
like
to
see
if
we
can
think
of
a
way
to
wrap
this
up
into
our
main
kubernetes
okr.
E
It
sounds
reasonable,
I
think
you
and
I
could
work
on
that.
Async
yeah.
E
Out
what
needs
to
happen?
What
I'm
worried
about
is
that
we've
got
a
few
things
that
have
come
up
in
the
last
few
weeks
that
I
think
are
going
to
hold
us
back
even
further
the
kubernetes
upgrade
and,
and
then
this
blocker.
That
is
whatever
the
thing
is,
I
pulled
last
week,
oh.
A
Cool
yeah.
No,
I
totally
agree.
Let's,
like
I
actually
think,
there's
a
great
benefit
in
these
not
being
separate,
because
they've
reached
a
point
where
we
go
too
far
into
the
migration
and
we're
not
comfortable
with
the
observability.
So,
let's
tie
those
together
into
a
single
okr.
E
A
Yeah,
oh
great,
okay,
cool
yeah.
Let's
do
that
when
I
saying
kinda
we
can
always
use
some
time
in
the
demo
on
thursday
to
debate
if
we
need
to
cool
okay,
so
the
only
other
one
I
had
as
a
question
mark
then
was
this
slo
one.
A
Now
we
do
need
to
reach
nestle;
it
probably
has
very
minimal
value
for
now.
So
the
goal
behind
this
slo
is
to
separate
out
the
different
kind
of
stages
of
our
pipelines,
and
we
we
don't
all
equally
share
responsibility
for
these
things
right,
so
we'll
be
able
to
use
this
to
identify
like.
Is
it
something
that
delivery
needs
to
focus
on
or
actually
do
we
need
to
be
spending
way
more
time
with
quality
or
or
development
or
whatever,
whoever
it
is
to
help
move
things
forwards?
A
So
I
think
probably
the
way
I'm
seeing
this
is
the
first
bit
is
just
literally
work
out
what
we
need
to
do
for
that.
The
second
bit,
then,
is
like
probably
there'll,
be
some
gathering
of
data
and
analyzing
things,
and
then,
after
that,
it'll
be
potentially
some
work.
A
I'm
gonna
propose
we
possibly
don't
have
this
as
an
okr,
unless
people
feel
strongly,
and
my
reason
for
saying
that
is
purely
because
we
already
have
three
if
we're
saying
that
we'll
do
something
with
rollbacks
something
with
the
dog
fooding
and
change
logs
and
kubernetes,
which
already
feels
like
quite
a
lot
given.
There's
only
seven
of
us.
A
If
we
do
have
one
I'll
I'll
ask
aaron
if
he
agrees
with
that,
if
we
do
have
one,
it
will
be
incredibly
lightweight
and
you
know
I'll
focus
on
doing
the
bulk
of
the
work
so
that
it's
not
four
for
everyone
in
the
team.
So
unless
people
strongly
want
to
particularly
define
an
okr
around
slos
I'd
say
for
q1,
let's
maybe
use
the
quarter
to
define
that.
E
E
A
E
A
Yeah,
no,
it's
a
really
great
point
yeah
and
I
think
for
all
of
them.
It's
it's
going
to
be
all
of
these
okrs.
It's
going
to
be
scoping
these
correctly,
but
yeah.
I
agree.
Let's,
let's
frame
this
as
let's
start
with
web
and
api
put
in
the
observability
things
and
then
to
see,
we
may
still
need
to
refine
it
down
a
little
based
off
this
last
quarter.
D
E
Sidekick
is
stateless.
Well,
it's
supposed.
D
E
D
So
it's
it's
comp
yeah
talking
about
sidekick
because
we
have
the
special.
D
A
Okay,
cool
great,
yes,
okay.
What
I
think
I
will
do
is
pop
these
ideas
down
and
we
can
async
I'll
put
it
on
a
suggestion
for
scope,
and
then
we
can
async
work
out
like
what
the
right
size
is
for
these
things.
A
Yep,
perfect,
okay,
great
thanks.
So
much
for
all
your
input
in
this,
hopefully
we're
going
to
end
up
with
very
useful
and
motivational
okrs,
so
rough
timeline
for
this
stuff
is
this
week
is
really
about
agreeing
on
the
ideas.
A
Okay,
our
issue
and
ping,
you
all
like
the
ideas
we've
got
and
then
we
can
think
about
them
for
like
a
day
or
two
scope
them
down
a
bit,
get
some
input
from
maron
and
then,
by
the
end
of
this
week,
be
fairly
happy
that
we
roughly
know
like
what
we're
going
to
try
and
achieve,
and
then
we
have
next
week
to
actually
do
the
wording
and
try
and
work
out
what
are
the
right
key
results
and
also
any
updates
on
working
epics.