►
From YouTube: GitLab 16.2 Kickoff - Verify:Runner
Description
Kickoff video for GitLab Runner Fleet, SaaS, Core
https://gitlab.com/gitlab-org/gitlab-runner/-/issues/29288
https://about.gitlab.com/direction/verify/runner_core/
https://about.gitlab.com/direction/verify/runner_fleet/
https://about.gitlab.com/direction/verify/runner_saas/
A
Hi
everyone.
This
is
the
runner
group
that,
when
we're
hosting
the
16-2
Milestone
kickoff
video
today
per
usual,
I'm,
going
to
start
off
with
the
runner
Fleet
plan
for
this
Milestone.
So
I'll
start
sharing
my
screen
and
as
a
reminder,
Runner
Fleet
really
focuses
on
those
people
who
are
managing
large
fleets
of
Runners.
A
A
So
last
time
we
talked
about
a
larger
feature
that
we
were
working
on,
which
is
our
MVC
of
a
runner
Fleet
dashboard
to
provide
you
a
higher
visibility
into
your
Fleet
of
Runners,
but
we
kind
of
had
to
Pivot
on
that
this
Milestone
and
we're
working
on
coming
up
with
the
architecture
blueprint
of
how
we're
actually
going
to
implement
those
features
that
that
we
want
to
do
we're
dealing
with
a
lot
of
data.
So
we're
gonna
try
out
some
the
plan
that
we
have
right
now,
which
is
documented
right
here.
A
If
you
want
to
look
into
it
more-
and
this
should
give
us
a
good
idea
about
a
POC
of
what
is
really
possible
here
and
then
we'll
continue
to
go
from
there,
but
we
are
still
focusing
on
the
same
features
as
we
wanted
to
before
a
big
one
being
the
runner
cues,
which
is
something
that
is
very
popular
and
ask
for
some
other
things
that
we're
working
on
right
now
on
the
front.
End
are
related
to
the
new
token
architecture,
which
we've
gotten
a
lot
of
feedback
on
that
new
flow.
A
So
far,
so
thank
you
for
providing
that
some
of
those
things
include
updating
our
runner
Details
page,
so
that
you
can
see
the
groups
of
Runners
that
are
using
the
same
configuration
and
then
some
other
smaller
things
like
summarizing.
The
upgrade
statuses
of
all
of
your
system,
IDs
that
are
in
that
group,
so
that
you
can
get
a
better
high
level
view
of
how
many
upgrades
are
available
and
then,
lastly,
in
the
ux
area,
we're
also
working
on
making
the
experience
for
that
new
token
architecture
as
good
as
we
can
so.
A
Some
things
like
telling
users
how
to
bring
their
Runner
online,
which
is
if
you
haven't,
if
you
register
a
runner
and
never
get
it
to
contact
gitlab,
then
it
will
show
up
as
never
contacted
and
there's
really
no
way
for
we
don't
tell
you
how
to
fix
that.
So
this
would
be
something
that
we
could
work
on
to
allow
you
to
have
less
of
those
stale
runners
in
your
backlog,
and
that
is
pretty
much
it
from
Fleet.
So
I'll
pass
it
to
Gabriel
to
talk
about
Runner,
sass.
B
Thank
you
so
much
Gina.
Let
me
also
share
my
screen.
B
Perfect
that
should
work
so
after
an
exciting
last
release,
where
we
have
released
the
new
Mac
and
one
Runners
for
Runner
SAS.
We
have
released
the
first
GPU
Runners
and
we
have
also
upsized
all
of
our
Linux
Runners,
which
you
can
see
here.
We
have
now
in
this
iteration
that
currently
runs
also
upsize,
our
storage
with
also
no
increase
in
cost
factors.
So
that's
super
exciting
because
we
had
some
feedback
that
some
of
you
with
larger
repositories,
ran
out
of
storage.
B
So
now
you
can
build
your
applications
up
to
100
gigs,
so
that's
exciting,
jumping
or
in
the
new
iteration
one
thing
to
Mark.
We
used
to
have
one
big
iteration
planning
issue
for
all
three
of
our
groups,
slash
categories
we
have
now
branched
out
because
they
became
quite
big.
So
for
the
first
time
we
have
our
own
planning
issue
for
the
runner
SAS
topic,
which
I'm
just
going
to
talk
through
today.
So
the
main
focus
of
us
is
still
I'm
getting
Mac
OS
into
GA.
B
We
have
narrowed
out
most
of
the
the
issues
that
we
have.
It's
currently
running,
smooth,
there's
just
some
few
background
work.
We
still
have
to
do,
for
example,
CD
config.
B
This
is
boring
for
our
users,
but
an
exciting
thing
that
comes
with
it
is
that
we
are
actually
releasing
images
more
often
so
we're
planning
to
have
one
latest
image,
for
example,
that
we
update
every
week,
which,
for
you
is
exciting,
because
if
you
use
that
image
you
will
not
have
to
run
any
lower
installation
types
report,
your
jobs,
so
overall
it
should
reduce
some
of
your
job
execution
times.
B
Also,
there's
another
thing:
that's
exciting:
we
are
planning
to
use
local
SSD
for
the
Mac
OS
Runners,
which
should
improve
the
iops
by
a
big
margin.
That's
something
that
we're
super
excited
about,
because
we
know
that
the
EBS
volume
currently
is
a
bit
slow
yeah,
and
all
of
that
should
push
us
further
into
GA
territory
with
Mac
OS.
Another
thing
that's
super
exciting
is
we
are
planning
to
offer
larger
machine
types
building,
even
on
top
of
the
ones.
B
B
Maybe
you
hear
about
it
in
CI
authoring,
if
you
watch
their
planning
video,
there's
a
concept
going
on
right
now.
It's
called
gitlab
Subs,
it's
still
in
early
phases,
but
for
that
we
are
doing
a
POC
to
see
how
we
can
execute
events
and
get
lab
steps
in
the
current
Runner
architecture.
C
Hey
thanks
to
Thanksgiving
for
the
handoff,
hey,
really
exciting
stuff
coming
in
both
of
those
areas,
so
folks,
I'll,
just
briefly
cover
run
a
core
and
I
should
start
off
with
just
a
reminder.
C
If
you
want
to
look
at
the
20
000
foot
view
of
what
we
have
directory
planned
and
want
to
call
for
this
fiscal
year,
the
direction
patients
places
to
start,
and
when
you
come
to
our
Direction
page,
you
can
see
we
have
the
one-year
plan
and
under
one
year
plan
we
have
the
male
FY,
24
color
themes
and
there
you
can
see
some
of
the
big
boulders
or
big
subcategory
of
themes.
C
If
you
will
that
we're
focused
on
for
Runner
core,
so
under
the
world-class
Dev
secups
experience
for
the
for
fy24,
the
next
one
is
open
architecture
Gina
touched
on
that
a
bit.
We
don't
have
a
lot
of
work
left
in
the
core
area
for
that
most
of
the
heavy
lifting
has
been
done
there.
They
have
the
remaining
will
this,
mostly
in
in
Gina
arms
area,
on
fleet.
The
next
big
theme
on
the
world-class
Dev
seconds,
I'll
talk
on
and
touch
one
in
today's
kickoff
is
next
on
auto
scaling
and
again
just
a
reminder.
C
This
is
auto
scaling,
git
lab
Runners
on
public
Cloud
instances,
but
more
specifically
virtual
machines.
Again,
we
have
multiple
different
solutions
for
hosting
Runners
on
various
Platforms
in
terms
of
Auto
scaling.
We
have
our
kubernetes
executor
and
we
also
have
our
Auto
scale
solution
that
enables
you
to
scale
get
lab,
Runner
worker
agents
or
or
if
you
will,
on
public
Cloud,
VMS
and
so
switching
over
to
the
1602
iteration
plan.
C
Major
Focus,
every
iteration
for
us
in
one
call
is
very
clearly
on
both
we
have
a
lot
of
surface
area.
We
cover
and
run
a
call
in
terms
of
platforms
that
we
support
in
terms
of
sub
managed
customers
in
terms
of
different
use
cases.
So
we
invest
a
lot
of
time
in
each
iteration
on
bugs
that
are
affecting
CI
workloads
all
of
these
different
use
cases
and
then,
in
terms
of
new
features,
what
we
hope
to
get
through
in
1602,
specifically
for
another
scaling
in
15
that
9
15.10
15
111.
C
We
release
the
initial
feature,
sets
related
to
run
auto
scaling.
We
release
the
pleading
plugin,
which
is
the
plugin
that
enables
Runner
Auto
scaling
on
Amazon
EC
chip
instances
in
experimental.
What
we'd
like
to
do
is
to
get
to
a
beta
version
of
that
in
16-2,
so
that
you
could
feel
much
more
comfortable.
Adopting
that
kicking
the
tires
the
goal
is.
C
We
would
like
to
get
feedback
as
soon
as
possible,
the
next
two
to
three
iterations,
so
that
we
can
determine
if
it's
the
right
time
for
us
to
switch
that
to
GA,
so
major
focus
in
terms
of
new
features
throughout
the
person.
16
2
is
the
cleaning
plugin
for
ec2.
My
goal
is
to
hopefully
get
that
over
to
Beta,
so
we
can
continue
attraction
of
getting
folks
to
start
adopting.
Then
you
run
auto
scaling
patterns
and,
if
folks
are
interested,
we
already
have.
This
is
a
quick
reminder.
C
We
already
have
a
documentation
for
the
new
auto
skating
Solutions.
If
you
go
to
the
doctor,
get
that.com
Auto
scale.
This
is
the
new
landing
page
DOC
for
the
new
run.
An
auto
scaling
solution
gives
you
a
high
level
overview
of
what
the
solution
is,
what
the
solution
is
and
then
you
can
go
into
the
various
executive
types
to
understand
how
to
use
the
new
auto
scaling
solution
so
run
a
core
main
focus
bugs
versus
apparently
well,
but
in
terms
of
new
features
again
Runner
Auto
skidding
moving
the
ball
forward.