►
From YouTube: Delivery Fast Boot: Kickoff Day 1
Description
Delivery team fast boot Day 1 kickoff, overview of existing state and goals for the fast boot week.
The issues listing the work done are contained in epic https://gitlab.com/groups/gitlab-org/release/-/epics/17
A
A
So
everything
that
cumulus
up
until
this
point
right
like
we
know
that
in
the
run
up
to
this
date,
people
modular
a
lot
of
things
and
once
that
is
done
this
period,
we
call
the
the
free
spirit,
or
this
is
where
we
create
the
stable
branches
right.
So
from
here
onwards,
no
new
feature
is
being
emerging,
the
branches
and
also
we
only
pick
the
items
that
are
broken
or
that
we
notice
that
something
is
not
working
correctly
and
out
of
these
stable
branches.
A
We
create
our
sieves
and
we
create,
as
many
artists
is
necessary
and
go
with
the
dance
of
picking
two
stable.
Then
once
the
stable
branch
is
ready
tag
once
we
tag
we
deploy
two
staging
and
then
we
do
QA
and
then,
when
the
QA
is
done,
we
move
to
canary
and
then
finally,
if
everything
is
correct,
go
to
production-
and
this
gets
repeated
over
and
over
again
up
until
the
22nd
right.
A
B
A
A
C
A
A
A
B
A
Up
until
the
7th
we're
going
to
use
the
same
processes
we
have
been
doing
so
far,
but
on
this
date
we
are
going
to
turn
on
up
to
deploys.
So,
starting
from
this
day.
Anything
that
you
created
up
until
7th
is
accumulated
is
going
to
be
deployed
in
one
of
the
first
are
seas
or
whatever
else
is
like,
let's
say
our
c1,
but
then,
following
from
this
day
onward
we're
going
to
go
to
the
process
of
how
to
deploy
branches.
C
A
B
A
Basically,
going
for
the
same
model
with
the
stable
branches,
but
only
with
a
bit
faster
cycles
so
per
week
to
ensure
that
we
can
like
iron
out
all
the
things
so
basically
on
Sunday.
This
is
going
to
be
automatically
created
and
this
branch
is
going
to
go
for
a
green
commit,
so
lots
green,
commit.
That
is
a
master.
A
C
A
To
have
people
check
staging
as
part
of
the
regular
flow
of
flows,
I
think
we,
we
will
probably
want
to
also
change
the
definition
of
dumb.
Our
definition
of
done
in
development
currently
does
not
state
that
you
need
to
see
through
your
changes
through
the
environments
yeah
it
sort
of
implicitly
is
where
it
says
something
like
definition
or
something
does
not
include
appropriate
features,
but
it's
implicit
I
brought
this
up
in
the
past
and
also
responds
I.
Well,
I'm
gonna
make
it
explicit.
No,
no
because
now.
A
A
Whatever
the
release
number
is,
and
this
is
going
to
be
automatically
pets.
Rights
like
we
are
building
an
automated
system
where
every
hour
we're
going
to
Lexi
check
any
merge,
requests
that
have
this
label
they're
going
to
add
that
to
the
auto
deploy
branch,
we're
going
to
try
and
tag
that
and
deploy
it
within
the
next
six
hours,
alright,
and
go
again
directly
to
staging.
A
A
So
at
any
point
we
can
just
kill
it
and
say
alright
nobody's
using
it,
but
what
we
wanna
do
is
we
want
to
make
sure
that
anything
that
goes
on
canary
is
fixed,
quick,
because
we
want
to
have
canary
enabled
as
much
as
possible
right,
because
if
you're
change
broke
something
there
are
ten
other
changes
that
people
are
bringing
into
the
environment.
We
want
to
verify
that
their
changes
are
also
working,
and
anything
we
notice
in
canary
that
is
broken
will
follow
the
same
process.
A
A
B
A
As
certain
as
you
can
be
right,
that
thing
is
fine.
We
are
going
to
deploy
to
production
now.
The
question
is:
what
happens
when
we
deploy
to
production
and
we
see
that
something
is
broken
well.
First
of
all,
we
are
going
to
start
using
the
hot
batcher
a
bit
more
frequently
if
it's,
if
it
really
arrived
to
production-
and
it
really
impacts
a
lot
of
users.
Hot
patcher
is
one
thing.
A
A
A
A
A
C
A
A
A
A
C
B
C
A
All
right
sounds
good.
Basically,
we
need
to
create
an
automation
that
will
create
these
branches
automatically
gets
the
green
pipeline
from
master
so
that
it
can
actually
create
that
branch
from
and
the
auto
picker.
That
will
do
this
everything
and
connect
all
of
those
things,
so
that
this
flow
goes
automatically
and.
A
This
is
what
we've
been
discussing,
how
this
is
going
to
look
in
details
so
in
the
release
tools,
project
will
have
the
auto
deploy
task
that
will
create
the
branches
every
week.
The
branches
are
going
to
be
named
this
way
and
they're
going
to
be
synced
across
ie,
omnibus
and
the
deployer.
The
reason
for
that
is
omnibus
needs
to
be
able
to
use
the
same
process
as
good
luck.
Ie.
They
also
depend
on
each
other,
with
assets
and
so
on.
A
The
deployer
means
the
same
thing,
because
we
might
need
changes
inside
of
the
deployer
/
deployment
cycle
as
well.
The
picker
is
going
to
look
for
a
specific
label
and
that
label
is
going
to
be
named.
How
did
you
name
it
sheriff?
Where
was
it
beep
underscore
branch?
No,
we
decided
to
call
it.
Auto
deploy
English,
also
deploy
branch.
A
A
All
right,
the
the
picker
will
automatically
know
the
auto
deploy,
will
automatically
look
for
green
pipeline
and
then
trigger
the
building
omnibus,
which
will
transfer
the
version
right.
Click
start
creating
the
package
once
the
package
is
done,
is
going
to
ship
the
version
as
well
to
the
deployer
pipeline,
and
we
are
going
to
answer
the
deployer
pipeline
pipeline.
A
check
for
what
was
previously
the
polar
deployed,
the
old
version,
and
we
have
the
new
version
because
we
are
just
adding
it.
A
Deployer
is
going
to
deploy
it
to
staging
and
because
we
have
these
two
versions
at
whom
we
can
transfer.
We
are
going
to
automatically
create
QA
issue
that
we
currently
have
as
well,
and
this
allows
us
to
also
like
achieve
some
compliance
as
well,
where
all
changes
need
to
be
down
and
verified
before
they
go
to
production.
A
B
I
was
raising
their
plan,
that
is
going
yellow
or
the
sort
of
problems,
because
we
have
a
bunch
of
jobs
in
this
pipeline.
The
idea
we
have
that
we
expose
the
version
as
a
CI
artifacts
all
right,
and
you
can
grab
the
promise
that
that,
in
every
job
in
between
yet
and
the
path,
artifact
us
its
dependence,
are
you
get
some
like
twenty
jobs
as
it
would.