►
From YouTube: Delivery Fast Boot: Day 3 discussion with Product and Development representatives: Part 2
Description
Delivery team fast boot Day 3: Discussing the process from development perspective with Jason Lenny from Product team and Douwe Maan from Create team.
The issues listing the work done are contained in epic https://gitlab.com/groups/gitlab-org/release/-/epics/17
A
Half
an
hour,
and
that's
it
it's
weird
so
yeah
when
I
was
I,
was
trying
to
see
you
like
what.
If
we
approach
it
a
bit
differently
and
say
our
self-managed
customers
are
going
to
get
the
version
of
the
club
that
we
deployed
at
some
point
in
time
that
month
and
that
is
going
to
be
called
1111,
for
example,
and
this
this
actually
buys
us
time.
A
A
B
B
A
Theoretically,
though,
ya
know,
we
would
then
have
to
test
that
snapshot
as
well,
somehow
right
and
have
the
stable
branch
created
off
of
that
package.
It
up
and
then
have
a
way
of
ensuring
that
everything
that
we
want
to
ship
is
not
already
fixed
and
if
it
is
pick
it
into
stable
and
then
we
run
into,
but.
B
C
Yeah
and
we
sort
of
have
another
version
of
the
same
problem
when
we
get
to
the
point
where
we're
doing,
because,
in
theory,
we'll
get
to
a
point
Sunday
where
the
first
RC
deployment
of
the
next
release
goes
out
the
day
after
the
last
release
was
cut
yeah,
and
so
at
that
point
yeah,
what's
the
on
from
experience
it
then
just
it's
eventually
going
to
become
one.
We
had
the
other,
an
arbitrary
cutoff
date,
so
you
might
as
well
pick
an
arbitrary
cutoff
date
and
track
to
that
and.
A
C
More
than
interesting
things
we're
working
with
Melissa
on
some
like
compliance
requirements
Risa's,
so
maybe
what
defines
that
becomes
like
a
queueing
sign
of
flight
or
a
compliant
sign-off.
That
says
this
is
the
one
and
then
that
can
be
entered
into
the
system
and
then
resume
the
automation.
So
it
is
still
a
bangle
intervention,
but
we're
going
to
put
something
into
the
process
at
some
point,
and
so,
if
you
can
align
this
to
that,
we're
not
creating
multiple
manual
intervention.
A
C
Either
have
to
have
someone,
he
says
like
this
human
or
solution
once
we
confirm
the
quality
is
good
at
this
or
you
just
all
ship.
Today
you
chose
and
hope
for
the
best
and
no
legal
patches
like
if
something
is
discovered
just
but
with
self-manage.
That's
a
lot
harder
than
it
might
sound
scary
to
talk
on
website,
but
we
can
just
press.
B
A
Now
this
is
going
to
be
complicated,
so
I
have
pyramid
me
what,
if
we
create
like
Yogi's
working
on
this
thing
with
the
QA
issue
and
all
of
that
stuff,
but
one
of
the
things
that
he's
adding
there
is.
We
are
going
to
have
an
overview
of
what's
actually
deployed
to
which
environments
and
what
we
could
do
in
theory
is
say
that
we
still
have
a
20-second
cutoff
and
we
are
going
to
have
this
JSON
file
that
is
going
to
have.
Like
you
know,
environments,
it's
going
to
have.
A
B
Premises
on
premises
of
London
service
packages,
we
keep
the
7th
as
the
same
type
of
people.
Please
we
have
today,
then
you
would
say
on
the
7th
we
always
create
because
expect
the
branch
is
called.
Leonard
cannot
employ
twice
version
religion
in
there.
If
we're
talking
about
tooth
comb
where
he
doesn't
so
if
we.
B
A
B
A
A
B
A
D
B
A
Here's
a
question
imagism
for
you
as
well:
we're
never
gonna
win
as
in
we're.
Never
gonna
have
like
a
hundred
percent
spotless
release,
as
in
there's
always
going
to
be
a
buck
somewhere,
so
we
might
need
to
actually
find
what's
our
acceptable
failure
rate
as
in
if,
for
example,
you
released
something
on
the
22nd
and
we
realize
this
didn't
end
up
there,
maybe
it's
okay
to
just
edit
the
blog
post
and
announce
you
know
edit.
C
A
B
A
A
A
B
B
E
E
Because
we
I
mean
like
I,
guess
for
back
ports
for,
like
that
footage
map.
We
wouldn't
have
that
ports
because
we
would
always
be
moving
forward
and
that's
not
possible.
September
customs
can
but
yeah,
but
you
ignore
back
ports
and
just
the
same
I
think
the
exception
request
stuff
and
all
this
picking
stuff.
It
would
still
be
the
same
well
I'm
being
followed
by
some.
E
B
D
A
Not
the
problem
for
them
to
do
right
now
is
just
that
that
instability
that
is
implied
in
the
package
is
making
it
like
something
that
people
don't
want
to
do
below.
We
have
been
smoking
with
some
customers
that
are
more
than
receptive
of
doing
some
continuous
deployment
appeal
of
itself.
If
there
are
some
guarantees,
that's
dead,
stable
enough
I,
don't
wanna
mention
a
name
because
we're
still
recording
beds.
They
were
actually
specifically
interested
in
like
oh,
if
you
can
make
it
hands-off
for
us
and
team.
A
B
D
A
C
This
would
potentially
be
around
the
22nd.
They
would
probably
mostly
be
whatever
difference
we
decided
apart.
They
are
like
if
we
waited
three
days
for
the
latest
attack
to
be
in
production
and
then
that
automatically
gets
the
stable
tag
unless
there's
been
a
production
incident
related
to
the
code
and
then
the
self-manage
installer
is
built
off
of
the
stable
branch
yeah.
C
A
C
B
D
A
B
A
A
A
And
it
does
the
same
thing
so
the
idea
here
was,
if
we
do
this,
we
can
always
deploy
from
this
branch,
can
always
propagate
through
environments
and
master
always
remains
the
source
of
truth
and
that's
the
whole
process
right.
People
speaking
to
the
or
system
picks
into
the
branches
automatically,
and
we
discard
those
branches
right
now.
What
you're
kind
of
proposing
here
is
master
and
main
line,
and.
C
There's
a
couple
ways
you
can
do
this
and
one
way
you
can
do
it
is
just
with
a
tag
and
say
the
this
was
known
to
be
stable.
We
don't
have
the
ability
would
like
we're,
not
bridged
we're,
not
creating
a
bread
and
they're
developing
in
that
branch
separately.
For
this,
so
we
would
just
say
like
when
that
last
mater
no
Potter
deployed
was
so
into
production
like
three
days
later
gets
attacked
stable.
So.
A
A
Separately
so
you're
saying,
but
you
need
backwards
as
well.
That's
the
problem,
that's
the
main
problem
that
you.
This
is
only
one
dimension
and
we
have
customers
that
use
tree
backwards
or
whatever.
So
you
constantly
need
to
have
a
branch
where
you
would
want
a
back,
but
because
we
have
security
releases.
We
have
all
that
stuff
that
requires
stable,
yeah.
E
D
E
E
C
A
Is
fine
if
you
have
one
dimension,
but
we
don't.
We
have
at
minimum
three
other
stable
branches
that
you
have,
but
Howard
those
automatically
maintain
and
so
right
it
becomes
a
problem
because,
in
order
for
your
security
face
to
end
up
in
any
of
these
branches,
you
need
to
then
play
once
this
got
the
scar
that
you
need
to
be
quick
to
react.
But
if
you
help,
if
this
was
a
security
severity,
one
issue
right,
you
picked
it
and
deployed
it
on
get
level
calm,
but
you
can't
wait
anymore.
A
D
A
But
just
now
that
you
understand
what
kind
of
levels
of
challenges
we
have
here,
the
team
that
we
proposed
here
with
this
switch
and
how
we
are
thinking
about
weekly,
deploys
and
any
other.
But
we
know
that
there
is
going
to
be
a
problem
of
announcing
things
right
and
timing.
What
is
going
to
end
up
in
a
self-managed
release,
because
during
that.
C
A
C
C
A
Who's
gonna
install
it
that's
a
different
question:
how
are
you
gonna
support
your
customers?
That's
that's
where
the
problem
comes
from,
because
we
are
already
having
problems
of
this
one
one
month
cycle
being
major
problem
for
a
lot
of
our
customers
right,
hello,
well,
you're
not
forced
to
upgrade,
but
when
they
do
without
many
names
again,
you
know
this.
We
have
some
customers
that,
just
recently
to
jump
to
11.5,
for
example,
and
because
we
already
moved
away,
we
already
like
progress
further
their
regressions.
A
Their
problems
are
something
that
we
didn't
see
because
we
moved
on
and
they
upgraded
what
eight
nine
months
later
so,
but
the
amount
of
changes
that
accumulates
you,
you
have
a
very
difficult
problem
of
controlling
once
it
actually
by
customer,
actually
upgrades
controlling
what
that
actually
brings
in
their
environments,
yeah
and
then
almost
suddenly
have
to
talk
about
like
hey.
Maybe
you
know
we
can
have
all
the
the
work
related
to
feature
developing
actually
also
have
to
be.
A
B
A
A
I'm
gonna
assume
that
this
is
okay
for
now.
I
already
know
that
there
are
going
to
be
certain
problems
in
my
communication
and
all
that,
but
we'll
both
survive.
It's
going
to
be
a
bit
confusing
just
know
like
the
questions.
I
have
are
more
for
you
on
what
we
can
actually
do
and
use
of
good
luck
for
some
of
these
things.