►
Description
Issue: https://gitlab.com/gitlab-org/incubation-engineering/jamstack/meta/-/issues/37
Cloudflare Integration Epic: https://gitlab.com/groups/gitlab-org/incubation-engineering/jamstack/-/epics/12
Last Pipeline Wizard MR: https://gitlab.com/gitlab-org/gitlab/-/merge_requests/76128
A
So
this
is
my
weekly
update
for
this
week,
so
I
got
two
things:
one
is
the
good
old
pipeline
editor,
which
is
just
stringing
along
on
the
side
right
now.
A
So
that's
down
to
the
last,
the
last
mr,
at
least
for
the
view
component
that
can
be
used
and
following
that
I
need
to
actually
use
that
wizard
inside.
A
So
I'm
preparing
that
at
the
moment
this
has
gotten
its
first
review,
so
I'm
confident
that
we
get
this
merged
this
week
by
the
end
of
this
this
week,
hopefully,
which
would
be
great
other
than
that
I
have
been
working
a
lot
on
cloudflare
integration,
so
this
is
basically
just
an
abstract
on
how
how
we're
going
to
achieve
that,
and
where
do
I
want
to
go
with
that
cloudflare
integration?
A
I
think
the
performance
gain
of
using
cloud
filler
is
probably
something
that
we
should
strive
for,
because
I've
demoed
that
on
an
earlier
update
that
the
performance
for
gitlab
pages
sites
outside
the
us
isn't
that
great,
mostly
because
the
pages
is
running
on
a
us
server
and
we
don't
have
a
cdn
so
using
cloudflare's
cdn
caching
would
probably
help
a
lot
with
performance.
A
A
I
know
that
the
pages
team
has
spent
a
huge
amount
of
effort
with
protecting
against
ddos
recently
and
is
investing
a
lot
of
time
into
monitoring
and
utilizing
cloudflare's
services.
There
will
probably
be
a
benefit
as
well
as
server
cost,
because
right
now
we're
paying
all
this
bandwidth.
B
A
A
So
we
don't
have
a
service
like
that
at
gitlab,
where
we
can
run
basically
all
kinds
of
serverless
functions
and
applications
that
need
a
javascript
runtime
on
a
server
like
side,
renderings
apps
next
js,
maxjs,
remix,
the
newest
one,
but
also
astro
quasar
spelt
kit,
there's,
probably
a
huge
market
that
is
left
untouched
by.
B
A
Right
now,
so,
if
we
can
utilize
cloudflare.
B
A
A
Integration
is
probably
a
stretch
because
the
completely
different
things,
although
they
both
utilize
the
cloudflare
api
to
achieve
that,
so
one
is
option
a
is
connecting
to
cloudflare
on
an
instance
level,
and
I
think
previous
explorations
in
that
regard
have
probably
looked
mostly
at
this
option,
where
we
as
gitlab,
have
a
cloudflare
account
or
whoever
has
an
in
a
gitlab
self-hosted
instance
has
a
cloudflare
account
and
during
setup
somewhere
and
you're
configuring
hey.
This
is
my
cloudflare
account.
A
Here's
an
api
token
go
have
fun
and
whenever
a
repo
updates
a
page's
site,
this
just
goes
directly
to
that
account
and
it's
hidden
behind.
A
Really
know
or
care
that
there's
cloudflare
powering
all
this,
it's
kind
of
behind
the
scenes.
It's
just
cloudflare
does
the
delivering
cloud
letters,
the
dns,
routing
and
whatnot
service,
ssl
provisioning-
and
I
don't
know
what
else
is
all
involved.
So
it's
basically
very
very
much
behind
the
scenes.
The
upside
is
since
it's
behind
the
scenes.
It's
probably
a
very,
very
low
effort
on
on
the
side
of
the
user,
so
the
end
result
could
very
much
look.
A
It's
like
it's
very
integrated,
like
there's
just
one
single
button
in
the
gitlab
ui
that
says,
run
this
remix
app
now
click
and
somehow
it's
running.
It's
probably
not
that
easy,
but
that's
the
end
goal,
but
obviously
it
would
probably
cost
a
lot
of
money
because
whoever
maintains
that
instance
cloudflare
account
has
to
pay
for
all
those
zones
and
all
the
access
to
it.
That
said,
we're
doing
this
with
gitlab
pages
right
now,
just
paying
it
to
google
instead
of
cloudplay.
But
it's
just
a
caveat.
That's
that's
in
there.
A
That's
an
option
where
user
can
connect
their
existing
cloudflare
account
to
their
project,
so
every
project
might
have
their
own
cloudflare
account
can
create
it
and
they
can
obtain
the
api
token
and
they
use
the
gitlab
user
interface
to
save
that
api
and
then
during
build,
deploy.
Gitlab
communicates.
A
To
configure
a
few
settings
in
the
end,
it's
not
that
much
different
from
doing
right
now,
you
can
do
this
manually
using.
A
A
So
I'm
not
quite
sure
what
how
much
more
benefit
I
can
do
here
other
than
helping
the
users
set
up
their
their
pipeline
in
this,
the
upside
of
this
option
is
configurability,
which
is
something
I
have
been
saying
a
long
time
that
is
the
usp
of
gitlab,
where
users
can
have
their
own
interface
to
cloudflare.
They
can
interact
with
the
monitoring
themselves.
A
A
A
So,
in
the
end,
option
b
would
look
very
much
like
what
shree
is
doing
with
the
google
cloud.
Like
one
click,
integration
gets
set
up
quickly.
A
But
from
then
on,
you're
you're,
basically
a
bit
more
on
your
own,
because
we
helped
you
set
it
up,
but
now
go
ahead
and
deal
with
everything
yourself.
I
think.
In
the
end,
we
should
do
probably
both
solutions
because
they
can
easily
coexist.
A
The
question
is
what's
more
important
to
focus
on
right
now,
and
I
don't
know
that
yet
I'm
up
to
feedback
on
that.
I'm
currently
exploring
both
options.
So
I'm
having
a
play
around
with
the
with
the
api
of
cloudflare
trying
to
figure
out
what
can
be
done.
A
For
example,
I
know
there's
a
little
bit
difficulties
with
connecting
domains
so,
for
example,
c
naming
domains
to
a
cloudflare
instance
to
a
cloudflare
zone
is
an
enterprise
option
and
whatever
I
do,
I
would
like
to
stay
in
cloudflare's
three-tier
interface,
so
that
I
don't
know
someone
that's
self-hosting
on
cloudflare
can
get
started
easily
and
don't
doesn't
straight
away
need
to
buy
an
enterprise
subscription.
B
A
Cloudflare
just
to
make
gitlab
work
with
cloud.
I
think
that
would
just
be
an
entrance
barrier.
That's
that's!
Not
that's
not
helpful,
so
yeah.
I
would
like
to
stay
in
the
free
tier,
so
any
api
functionality
should
stay
within
one
of
those,
but
yeah
I'll
have
to
figure
the
details
out.
What
is
possible
right
now,
what
is
possible
within
the
free
tier?
What
are
the
caveats?
So
that's
my
work
that
I've
started
doing
this
week
and
will
probably
do
a
lot
more
of
in
the
next
couple
of
weeks.
B
Yeah,
I
didn't
want
to
interrupt
what
you
were
presenting
so,
and
you
actually
answered
a
bunch
of
questions
that
I
had
along
the
way
too,
regarding
like
custom
domains
and
things
like
that,
the
you
know
you,
you
mentioned
the
the
sort
of
parallel
to
the
five-minute
production
app
and
the
other
sort
of
parallel
that
I
was
thinking
of
as
you
were
talking
through.
This
is
the
way
that
we
have
our
kubernetes
integration
set
up.
B
So
it's
a
similar
process
in
which
you
go
to
create
a
cluster
and
you
in
an
art
and
now
on
gitlab.
It
didn't
used
to
be
this
way,
but
now
you
can
use
both
amazon,
eks
and
google
gkd
so
like
the
the
option
b
is
actually
kind
of
a
nice
way
to
think
about
it.
If
you
can
decouple
the
the
like
provider
from
the
integration,
so
that,
like
you
know,
say,
there's
some
other.
B
You
know
cloud
cloudflare
competitor
that
we
want
to
integrate
with
like
there.
There
could
actually
be.
You
know
like
an
opportunity
to
make
that
a
an
option
that
users
could
choose.
So
I
mean,
I
think,
like
option
b
from
you
know
from
my
less
informed
perspective
seems
like
at
least
a
good
first
iteration.
A
B
Versus
adoption,
a
which
is
kind
of
like
trying
to
almost
trying
to
boil
the
ocean
and
do
all
the
things
and-
and
maybe
that
kind
of
prevents
us
from
being
able
to
like
iterate
more
quickly
and
get
kind
of
feedback
that
way.
So
those
are
those
are
kind
of
my
first
reactions.
A
Thinking
as
well,
yeah
initially
but
right
now,
I'm
in
a
position
where
I
think
this
is
as
the
first
iteration
actually
a
bit
easier,
maybe
because
it
needs
less
less
work
to
to
achieve,
but
I'm
still
not
decided.
Yet
it's
it's
a
first
feeling
that
I
have
and
I'm
I
think,
while
playing
around
with
the
api,
I'm
I'm
probably
arriving
at
a
solution
that
tells
me
which
one
is
easier
to
have
as
a
first
iteration,
as
I
said
ultimately
would
probably
we
might
have
both
options
to
exist,
but
I
don't.
B
Know
I
could
see
I
could
see
situations
where,
depending
on
a
customer,
they
would
want
to
have
have
it
in
their
cloudflare
account
versus
like
a
globally
shared
one.
For
some
reason,
but
you
know
that's
that's
not
a
reason
not
to
to
do
the
other
way.
If
the
other
way
is
a
faster
path
to
iteration.
A
B
A
Probably
80
percent
of
of
users
probably
won't
do
that
at
all
the
whole
long
tail
of
pages
will
probably
not
not
yeah.
That's
whole
integration
thing,
so
it's
probably
just
for
pro
users,
whereas
if,
if
we
want
the
benefits
of
ddos
protection
and
server
costs
and
whatnot
we'd
probably
want
as
many
projects
on
cloud
flare.
A
So
that
would
speak
to
option
a,
but
that's
that's
actually
not
even
my
the
incubation
engineering
way
to
think
about
that.
That's
probably
much
more
infrastructure
sure
an
infrastructure
question.
My
question
is
actually
which
one
is
easier
to
implement
right
now,.
B
Okay,
yeah,
so
the
other
other
question.
I
guess
that
came
to
mind
and
I've
not
looked
into
this
in
quite
some
time.
But
what
is
the
kind
of
competitive
landscape
look
like
in
terms
of
other
providers
in
the
space
and
like
what
what
their
cdn
capabilities
are?
Is
it
kind
of
you
know,
I
I
guess,
do
they
sort
of
connect
and
integrate
with
cloudflare
and
things
like
that
or
or
do
they
kind
of
have
their
own
or
is
it
is
it
you
know
so.
A
Never
exposes
their
cdn
thing
like.
If
you
go
to
netlify,
you
won't
find,
I
don't
know
if
they're
running
their
own
service,
for
example,
or
if
they're
using
fastly
in
the
background,
I'm
not
sure,
but
you
don't
interact
with
that
at
all
you're,
interacting
with
okay
same
with
versa,
I
mean
one
of
the
competitors
in
that
regard
is
cloudflare
itself.
A
So
yeah
I'd
say:
if
we're
going
the
exposing
the
cloud
flare
thing,
we
go
a
very
special
way,
but
it's
always
one
that
I
think
gitlab
has
been
very
strong
at
the
providing
like
giving
the
users
control
over
very,
very
fine-grained
control
over
the
whole
setup
and
just
bringing
it
all
together
and
have
it
played
well
with
each
other.
A
B
Okay,
so
maybe
maybe
the
so,
if
I
think
it's
just
so,
I
understand
like
if,
if
I
look
at
the
like
netlify
and
others
like
as
a
user,
like
the
expectation
there
is,
I
don't
have
to
fiddle
with
my
cdn
it's
just
built
in.
I
can
just
expect
that
it's
going
to
be
globally
distributed
and
there's
no
configuration.
It's
like
it's
totally
like
opaque.
To
me,
like
I
don't
see.
Any
of
that,
it
just
happens.
Is
that
true,
yeah
pretty.
A
B
Yeah,
okay,
okay,
I
mean
so
that's
that's
an
interesting
way
to
look
at
this
too,
like
the
kind
of
the
user's
expectation
from
other
products
is
like,
I
don't
have
to
worry
about
it,
so
I
just
get
it
you
know,
and
so
maybe
maybe
that
is
maybe
it
makes
it
easier
to
compare
when
when
someone
is
looking
at
different
products,
but
to
your
point
maybe
the
configurability
is
a
is
a
feature
too
that
yeah.
B
B
B
I
don't
think
I
have
any
other
questions.
I
think
this
is
a
really
really
great
description
and
detailed
breakdown
of
of
these
different
approaches,
so.
A
Yeah
I
mean
this
is
a
whole
epic,
so
there's
probably
a
lot
of
issues
that
are
gonna
spawn
from
this
right
now,
I'm
in
that
exploration
phase
and
let's
see
how
it
goes
probably
more
on
this
from
next
week-
cool
all
right,
all
right,
well,
ben
thanks
for
joining
for
the
first
live
recorded,
update.