►
From YouTube: GitLab 13.1 Kickoff - Release:Progressive Delivery
A
A
The
first
one
is
offer
in
product
guidance
for
deploying
to
aws
and
in
the
last
few
milestones,
we've
been
doing
a
lot
of
work,
helping
you
deploy
to
aws
and
what
we
want
to
do
here
is
in
the
ci
cd
variables
area,
which
we
have
already
changed
to
make
it
easier
to
deploy
to
aws.
If
we
detect
that
you're
using
aws,
we
will
help
you
out
with
a
few
links
to
our
aws
deployment
manuals
and
to
our
different
cloud,
deploy
images
and
a
ci
cd
template
that
helps
you
deploy
to
ecs
to
aws
ecs.
A
A
A
So
the
idea
is
that
we
want
to
expand
our
current
canary
capabilities
for
kubernetes
to
allow
you
to
to
configure
a
real
blue
green
deployment
and
the
way
that
we
do,
that
is
by
allowing
to
set
weights,
and
then
you
can
set
the
deployment
to
different
weights
to
different
environments.
So
now
you
have
a
blue
environment
and
a
written
environment,
and
you
can
start
rolling
out
a
canary
pods
based
on
the
weight
that
you
set
and
then
you
can
gradually
switch
over
to
the
secondary
environment,
and
then
you
can
switch
back
and
forth.
A
So
this
is
a
rather
large
issue
again
expanding
our
current
canary
solution
and
for
this
specific
iteration.
What
we're
going
to
do
is
we
are
setting
up
all
the
things
we
need
behind
the
scenes
in
order
to
allow
our
canary
to
support
this
new
configuration
that
will
support,
ultimately
support
two
environments.
A
So
that's
what
we're
doing
now
backstage
work
in
the
next
few
milestones.
I
I
will
talk
about
what
we're
going
to
do
so
we're
going
to
start
with
api
and
later
ui,
but
this
is
the
specific
milestone
that
we're
just
setting
up
the
things
behind
the
scenes.
A
Next
up
is
show
error
rate
threshold
exceeded
on
the
playboard
okay,
so
we're
experimenting
on
helping
understanding
how
our
deployment
is
behaving
post
deployment
and
for
the
very
first
iteration.
What
we
want
to
do
here
is
before
even
attempting
to
stop
or
notify
or
anything
about
how
your
deployment
is
behaving.
A
We
want
to
show
that
there's
an
error
on
the
deploy
board
in
case
that
happens,
so
the
way
that
it
works
is
we're
using
prometheus's,
pre-existing
defined
error
rates,
throughput
latency
http
error
rate
and
if
such
an
event
happens,
that
one
of
the
thresholds
is
crossed
on
the
deploy
board
itself,
you
will
see
something
like
deploy.
A
The
final
issue
that
I
want
to
talk
about
and
one
that
I'm
very
excited
about,
since
we
had
a
lot
of
community
conversation
and
contributions
to
this
issue,
is
allow
parents
project
the
developers
to
create
pipelines
for
merger
requests
for
forked.
A
Mrs
and
there
used
to
be
a
larger
issue
that
this
was
broken
off
from,
and
so
after
having
lots
of
discussions
with
with
the
community
on
this
issue,
we
have
decided
that
this
is
the
first
approach
and
it
may
it
matches
a
lot
of
the
different
use
cases
that
we
heard
from
users,
but
we
do
plan
to
expand
on
it
later.
A
So
the
idea
here
is,
if
you
have
you're
an
open
source
project
maintainer,
and
you
have
a
specific
fork
demar
that
you
want
to
create
in
the
pipeline
of
the
parent,
and
you
don't
really.
If
you
don't
have
permissions
in
the
parent
project,
you
can't
really
run
it
and
test
it
and
see
if
everything's,
fine.
This
is
even
gonna,
help
us
internally
and
get
lab
with
your
community
contributions.
A
So
the
way
that
that
it's
going
to
work
in
consideration
is
that
you
can
request
for
someone
that
has
appropriate
permissions
in
the
parent
pipeline
to
run
it
for
you,
and
that
way
you
get
an
added
bonus.
You
get
a
code
review
done
out,
while
that
happens,
and
and
that
way
you
can
see
how
how
it
behaves
again.
A
If
you
don't
have
the
proper
permissions,
you
will
not
be
able
to
run
the
project,
but
you
can
ask
for
someone
who
has
developer
permission
or
up
in
the
parent
pipeline
and
they
will
run
it
and
check
that
everything's.
Okay,
so
super
excited
about
this.
This
is
one
of
our
most
popular
issues
in
release
and
specifically
for
progressive
delivery,
and
I'm
really
happy
that
we
can
finally
work
on
it.
This
has
been
pushed
for
several
milestones
because
we
were
debating
what
was
the
right
solution
and
really
with
the
help
of
the
community.