►
From YouTube: GitLab Demo Systems Discussion
Description
Jeff Martin, Senior Demo Systems Engineer, answers questions from the Customer Success Team about GitLab Demo Systems
B
Yeah
sure
well,
I,
think
Jeff
already
mentioned
my
question.
The
question
was
well
Cristobal
great
job.
As
always,
the
whole
demo
system
is
shaping
up
really
fast.
Really
great,
so
I
was
looking
at
the
demo
data
design
and
my
question
was:
do
you
think?
Is
there
any
value
and
are
there
any
plans
to
also
provide
them
a
data
escort
so
that
that
would
be
worshipped,
soared
and
Russian
controlled,
mega
travel,
I.
C
Think
it's
certainly
a
direction.
We
could
go
the
biggest
thing
that
I
I
think
it's
a
great
direction.
I
do
see
some
challenges
that
we
want
to
avoid,
and
those
challenges
are
mostly
with
codes
for
all,
so
everyone
can
create
their
own
little
scripts.
But
how
reusable
is
it
when
you
have
a
library
of
a
hundred
things
to
sift
through
what
I
prefer
to
do
is
basically
use
our
catalog
approach.
C
D
Yeah
thanks
so
much
for
a
great
presentation,
so
kind
of
putting
my
you
know,
demo
demo
hat
on
if
I'm,
using
sky,
tap
or
VMware
I
would
give
a
demo,
and
then
you
know
afterwards
reset
it
back
to
a
snapshot.
Obviously,
that's
not
quite
what
we're
doing
here.
What's
it?
What's
your
recommendation
on
being
able
to
quickly
revert
back
to
some
known
state?
Is
it
creating
a
template,
instantiating,
a
project
from
that
and
then
just
removing
the
project
afterwards,
or
is
there
a
more
efficient
or
granular
way
to
do
the.
C
Great
question:
it's
a
multi-headed
answer
right
yeah,
so
the
project
template
approach
is
what
I
would
say
for
right
now,
right
and
so
what
that
entails.
Is
you
build
a
master
template
of
what
you
want
to
call
it
right?
It
that's
issue
tracking.
If
that's
you
know
what
you're
trying
to
do
for
security
ops
trying
to
do
it
I
see
whatever
you're
trying
to
showcase
you
create
a
project
and
then
essentially
you
will
use
the
project
import
tool
to
import
from
template
or
the
problem
of
the
import
is.
C
It
does
not
include
some
of
the
extra
features
of
projects,
so
you
may
need
to
export
that
project,
then
import
it
back
in
as
a
new
project,
so
excellent
Satar
bring
it
out
and
then
bring
it
in
or
you
can
do
the
URL
approach,
so
the
export
import
and
then
throwing
away
the
new
project
you
created
for
that
demo
at
the
end
is
the
right
way
to
do
it.
One
of
the
things
that
I
heard
about
the
other
day
that
I
thought
was
very
smart.
C
Was
a
customer
had
asked
hey
I
love
what
the
demo
can
I
keep
that
can
I
see
that
long-term?
Is
you
can
export
that
ephemeral
demo
and
just
simply
give
them
the
tar
file
to
go
into
their
own
get
Leben's
to
play
with
was
one
of
the
examples,
my
that's
pretty
smart
as
long
as
there's,
no,
you
know
red
or
orange
data
in
it.
That's
okay!
C
In
terms
of
the
long-term
you
know,
post
them
will
clean
up.
That's
where
we're
looking
at
those
different
ephemeral
approaches
and
we're
trying
to
make
sure
that
we
do
it
right
so
for
the
u.s.
region,
you
do
whatever
you
need
to
do
from
project
import
of
your
group.
You
do
your
own
thing
on
the
EU
region
as
we
bring
this
demo
data
in
play,
books
things.
This
piece
is
in
what
we
like
to
do
is
have
a
kind
of
a
nightly
reset
to
snapshot
and
then
re-import
of
the
data.
C
How
do
we
focus
on
keeping
it
clean
and
the
best
approach
that
I've
been
able
to
come
up
with,
because
snapshots
revert
to
the
sink
you
took
the
snapshot,
yeah
is
if
the
demo
designer
can
use
relative
dates,
we
can
go
back
to
a
base
level.
Omnibus
configuration
have
that
ready
and
that
that
snapshot
available.
That's
nightly
if
that's
weekly,
if
it's
monthly,
whatever
the
numbers,
are
array
and
then
use
the
demo
designer
to
re-import
that
data
with
relative
and
current
dates,
as
well
as
avoiding
the
eye
click.
C
Some
configuration
setting
that
I
want
to
have
to
undo
that
work
or
I
can't
undo
that
commit
or
whatever
the
case
may
be,
so
we're
working
through
those
technical
discovery
pieces,
but
ultimately
for
right.
Now
the
project,
import
and
export
is
the
best
recommendation
that
I
have
and
if
anyone
else
in
the
team
has
other
ideas
or
other
things
that
you've
found.
That
work
feel
free
to
add
in
the
comments.
But
that's
what
I
would
suggest
brilliant.
E
C
C
So
the
current
state
is
everyone
who
wants
to
add
a
group
level
cluster.
You
have
access
to
your
cluster
in
groups,
yes,
and
you
basically
follow
the
normal
gait
lab
documentation
for
adding
a
cluster
to
your
group
and
you'll
have
to
go
into
either
cube
control
or
the
GCP
console
and
get
the
variables
that
you
need.
Usually
that's
your
commander's
endpoint.
Your
API
key
is
so
on
and
so
forth.
To
add
that
in.
G
Hey
great
job,
Jeff,
good
stuff,
I've
got
a
bad
connection,
so
I
may
go
and
out
I've
got
a
you
know.
Two
major
projects,
I
use
for
demoing
I,
bring
them
over
from
comm.
They
just
hang
and
just
do
weird.
Just
weird
behavior
I
know
I've
sent
it
to
you
I,
do
it
on
comm,
it
runs
green
all
the
time.
Is
it
something
in
my
infrastructure
or
something
on
your
side?
What
any
ideas?
It's.
C
G
C
Are
some
there
are
some
known
challenges
we
are
seeing
it
pretty
widespread,
it's
intermittent,
so
my
best
advice
is
pretty
simple
right
now
we're
the
first
is
try
to
rerun
the
job
see
if
you
can
grab
a
new
runner,
see
if
it'll
run
and
get
a
success.
So
if
a
job
fails,
try
to
rerun
it
first
Christian
and
I
are
working
through
a
lot
of
experimentation
to
try
to
basically
iron
that
out
and
make
it
more
stable
right.
C
Now,
it's
more
of
a
discovery
than
it
is
a
solution,
we're
adding
more
logging
to
or
adding
more
event
management,
just
to
kind
of
see
where
it's
going
wrong
because
anyone's
worked
with
kubernetes
knows,
there's
some
cryptic,
nough
stew
it
and
even
those
who
are
very
proficient
with
kubernetes
I
admit
it's
not
my
strong
suit,
but
those
who
are
looking
at
and
like
I
can't
see
what's
going
on
here.
So
right
now
we're
trying
to
understand
more
as
a
temporary
workaround.
We
we
first
fund
the
demo
cut
up.
C
We
had
Linux
runners
that
we're
running
instead
of
true
vanities
runners
and
those
were
pretty
stable.
We're
gonna
re
spin.
Those
up
today
to
provide
the
stability
at
the
runner
level
that
may
or
may
not
solve
the
review
app
challenges,
but
at
least
it
will
take
one
piece
out
of
the
equation
until
we
understand
what
more
of
the
problems
that
we're
facing
are
so
you're,
not
alone,
you're,
not
doing
anything
wrong.
C
That
I
can
that
I
can
see,
but
the
best
thing
to
do
is
simply
post
on
the
demo
systems,
Channel
or
specifically,
the
CI
KS
support
channel.
If
you
have
those
challenges
and
then
we'll
work
through
them
and
triage
and
figure
out,
if
it's
us,
if
it's
a
variable,
if
it's
a
hate
rear
and
the
job
it
works,
is
fine
and.
H
Good
morning,
Jeff
yeah,
I'm
gonna
echo
mark
sentiments
great
job
thus
far
and
big.
Thank
you
and
thank
you
so
much
for
adding
some
videos
and
documentation
in
the
system
as
well.
I
think
that's,
you
know
going
to
help
from
an
enable
enable
net
perspective,
so
yeah
great
stuff.
So
the
first
part
it's
a
two-part
question.
The
first
part
is
you
had
mentioned
that
we
have
provisioned,
essentially
some
of
the
folks
that
joined
our
workshop.
C
It
more
to
the
username
for
privacy
reasons.
We
obviously
do
not
map
over
the
students
first
and
last
name
into
our
into
the
get
lab
instance
because
of
the
sandbox
nature.
However,
we
do
want
to
see
what
the
name
of
the
person
who
redeemed
that
invitation
code
is
on
our
portal,
which
is
ref
data
protected.
So
we
can
actually
map
hey
I,
see
20
students
that
are
in
my
zone
call.
Only
15
names
are
here
which
ones
are
not
on
this
list
right,
so
we're
going
to
bomb
it
out
as
well,
but
ultimately
yeah.
H
Awesome,
thank
you
so
much
for
that.
The
second
part
is
really
kind
of
more
channel
focus
and
it
could
be
for
the
channel
team,
but
I
thought
I'd
ask
it
here.
We
know
when
the
channel
team
is
enabling
partners.
Are
they
going
to
be
using
this
as
their
demo
system,
for
example,
if
we're
enabling
some
of
the
essays
that
you
know
X
Y
Z
partner,
is
this
a
system
that
they'd
be
using
so.
C
We
have
a
strategy
in
scope
and
call
last
week
with
the
channel
team
about
the
specific
question,
the
consensus
that
we
came
to,
and
this
is
a
work
in
progress.
So
this
is
subject
to
change
is
that
we
will
use
our
training
cloud
for
their
initial
onboarding,
their
initial
certification
and
learning
the
get
lab
product
from
their
ongoing
sandbox
needs
and
demo
system
needs.
They
will
be
either
using
github,
comm
or
spinning
up
their
own
omnibus
infrastructure
for
that
partner,
with
the
help
to
channel
essays
who
will
be
guiding
them.
C
C
J
Someone
a
few
really
discovered
this,
but
we
are
experiencing
trouble
creating
new
classroom.
The
key
needs
platform
because
we
are
running
out
of
IP
addresses,
and
this
is
unfortunately
not
to
our
abused
of
them,
but
to
do
the
full
configuration
of
Google
that
it's
high
consuming.
So
if
you
need
a
new
class
sir,
please
asked
me
and
Jeff
because
usually
it's
required
some
attempts
to
get
the
Newcastle
time
and
this.
J
So
we
are
putting
the
correct
configuration
in
the
class
of
creation
to
avoid
a
larger
consumption
that
what
is
needed
and
I
saw
also
that
we
have
a
lot
of
classes
that
are
outdated.
So
the
version
that
you're
running
is
not
supported
anymore.
This
green
part
adds
to
the
integration
we'd
asked
because
we
are
expected
the
specific
version.
So
if
you
don't
have
anything
specific
or
a
heart,
should
be
migrated
on
the
cluster.
J
Keep
in
mind
that
the
upgrade
is
similar
on
a
time
perspective
to
the
colossal
rebuild
you
can
ask
to
us
to
rebuild
the
cluster,
so
we
are
going
to
free
IP
addresses
said
because
the
old
cluster
are
consuming
up
to
32
new
cluster
space.
So
this
can
help
some
of
your
workmates
to
have
a
closet
that
actually
may
not
be
possible.
So
thanks
to
you
for
the
collaboration.
C
I
Good
stuff,
again
echo
what
everyone
is
saying:
fantastic,
your
quick
question:
when
we're
creating
new
issues
and
everything
do
we
have
like
a
group
of
like
sample
users?
I
know
we
had
talked
about
that
at
one
point,
there's
something
we
can
assign
other
users
to
so
it
looks
more
real.
If
you
will
like
when
we
have
a
you
know:
data
through
there
yeah.
C
So
we're
working
on
that
right
now
and
I'll
give
you
the
work-in-progress
thought
on
this.
So
with
the
EU
cluster,
we
want
to
have
a
sample
company
with
it
right
and
then
use
the
marketing
persona,
users
and
all
those
different
characters,
as
well
as
additional
randomly
generated
users
that
we
have
in
there.
C
So
the
demo
designer
is
going
to
have
a
new
users
tab
that
will
show
the
users
that
crew
get
created
as
part
of
your
PlayBook
and
then
once
those
users
are
in
the
playbook,
you
can
then
assign
them
through
issues
milestones
so
on
and
so
forth.
From
that
perspective,
as
when
the
playbook
runs,
those
user
accounts,
let
you
get
created
and
on
the
demo
portal,
you
will
then
see
the
username
and
password
to
sign
in
as
those
users
or
absolute
administrator.
C
You
can
always
hit
the
impersonate
button
and
do
with
that
approach
as
well
from
an
individual
project
perspective.
You
can
imagine
if
you
would
create
15
users
or
20
users.
Out
of
a
part
of
that
we're
gonna
read
a
license.
It's
pretty
quick!
Nothing!
We
can't
mitigate
that
problem,
so
ultimately
I'm
trying
to
find
the
right
balance
of
does
every
you
know:
si
have
five
to
ten.
C
If
the
users,
if
all
the
projects
H
number
gets
created
or
do
we
use
with
the
user
part,
only
works
on
the
shared
instance,
so
I'm
working
through
some
of
those
technical
details,
but
the
answer
is
yes
that
that
is
coming
in
the
very
short
term,
with
a
you
know,
generate
a
list
of
ones.
We
plus
you
can
add
your
own
names
if
you
want
so,
if
you
want
have
some
fun
with
it.
So
yes.
G
C
So
we've
been
supporting
workshops
so
when
I
refer
to
the
training
cloud,
this
is
what
I'm
referring
to.
So
we
have
a
process
that
goes
through
and
creates
all
of
the
it
basically
does
the
user
account
provisioning
for
users,
so
these
are
the
workshops
or
training
classes
that
we've
run
in
the
last
three
months.
So
it's
a
known
thing
to
us.
C
Our
theoretical
capacity
right
now
is
around
200
people,
and
so
I
know
that
you
know
the
field
marketing
team
is,
you
know,
spinning
some
stuff
up,
we've
been
doing
monthly,
virtual
DevOps
ones,
and
so
a
lot
of
where
we're
at
right
now
is
the
answer
is
yes,
it's
not
a
problem.
There
is
a
process
for
doing
it
and
if
you
look
at
the
question
slide,
deck
itself
have
a
look
at
slides,
34
through
42,
and
that
will
give
you
the
full
context
of
how
we
do
it
great
Jeff.
K
Can
I
follow
up
on
that?
There's
an
active
thread
going
on
right
now
about
wanting
to
increase
these
DevOps
automation,
workshops
to
500
people
instead
of
200
I
just
want
to
make
sure
we
have
confidence,
and
we
have
experience
using
live
environments
in
a
virtual
format
for
like
a
workshop
with
200
people,
but
we
think
500
was
not
something
we
can
support
today.
There
yes.
C
I'll
answer
this
question:
just
I
got
asked
that
question
yesterday
by
Joel
and
he
was
something
filled
that
conversation
with
them
so
where
this
really
comes
down
to
is
right
now,
our
omnibus
infrastructure.
It
has
a
certain
size
of
an
instance
on
GCP,
so
there's
only
so
much
CPU
and
memory
assigned
to
that
instance.
The
training
cloud
itself,
as
we
spin
that
up
will
have
double
or
triple
that
capacity.
C
However,
there
is
a
theoretical
user
limit
to
that
capacity
and
how
much
we
spend
right
now
are
our
environment
is
perpetual,
so
we
basically
whatever
we
we
pay
for
for
the
duration,
so
I'm
not
going
to
spin
up
something
that
it's
going
to
cost
us
thousands
of
dollars
a
month
for
something
that
will
in
normal
use
20
to
30
people.
So
what
we've
seen
so
far
is
its
twofold
one,
the
infrastructure
capacity
which
we
can
solve
that
problem,
but
solving
it
next
month,
it's
not
going
to
be
on
the
roadmap
to
solve
that
right.
C
The
second
piece
of
it
is
when
we're
running
the
larger
workshops.
We
have
a
lot
of
essays
who
volunteered
their
time
to
be
teacher's
assistants
and
help
out
the
users
and
there's
a
limit
of
scale
to
that.
That's
the
biggest
challenge
I
listen
to,
and
so
the
the
virtual
DevOps
I'll,
let
those
who
are
actually
in
running
that
workshop
speak
to
any
details
around
that.
But
the
highlight
is:
if
you
get
to
500,
you
know
users
in
a
workshop.
How
many
essays
you
need
to
support
that
yeah.
K
So
I'm
just
asking
I
can
do
it.
Yeah
I
just
want
to
know
from
a
technical
demo
environments
because
I've
been
in
training
classes
before
we
overload
the
system
that
everything
goes
crashes
or
constitutes
screaming,
halt,
so
I
think
we
can.
We
have
to
have
that
conversation
in
terms
of
how
many
essays
do
we
need
to
support
this
if
we
can,
but
I
just
want
to
know
what
that
so
that
maximum
is
that
we're
confident
from
a
technical
demo,
and
so.
C
K
F
Yeah
so
I'm
new
here,
so
if
I
was
any
understanding,
these
excuse
these
nice
to
meet
everyone,
but
yeah,
so
I
know
that
we're
talking
about
in
ephemeral,
environments
I
think,
that's
great.
Obviously,
that's
a
huge
cost
saver
right,
I'm
just
wondering
if
there's
some
sort
of
like
automated,
like
reaper
script,
that
just
cleans
up
my
environment
or
if,
when
you're
done
with
it
like
how
do
you
get
rid
of
your
environment
so.
C
Good
question
there's
multiple
ways
to
solve
this.
So
within
get
loud,
everyone
kind
of
has
shared
access
to
a
several
different
cloud
accounts
right
and
so
in
GCP
we
have
gitlab
internal
group.
We
have
a
group
CS
group
and
we
have
to
get
love
demos
group
and
there
might
be
others
that
I'm
not
aware
of,
and
you
simply
create
your
stuff
up
in
there,
there's.
C
C
The
majority
of
our
users,
who
have
been
doing
a
lot
of
this,
are
using
infrastructure
as
code
tools,
terraform,
Confirmation
and
so
on
and
so
forth
to
simply
spin
up
what
they
need
to
and
then
destroy
it
when
they're
done,
and
so,
if
you're
going
to
be
doing
a
little
bit
of
it,
don't
worry
too
much
about
it
clean
up
what
you
can
if
you're
going
to
be
doing
very
comprehensive
environments,
I
really
suggest
using
terraform,
because
it's
easy
to
do
a
terraform
apply.
Then
a
terraform
destroy
that'll
clean
up
after
you.
L
C
Basically,
the
sooner
that
you
were
done,
it
can
mark
ready
the
better.
What
I'm
ultimately
trying
to
do
is
ban
or
delete
user
accounts
on
i2p,
and
so
I
can
basically
get
that
list
down
to
zero.
So
the
sooner
that
we
get
everyone
saying
I,
don't
need
ITP
anymore,
the
better.
So
if
you
move
your
projects
out,
even
it's!
C
Basically,
if
you
have
projects
in
ITP
that
you're
not
using,
but
you
might
need
them
and
you
don't
want
to
transition
right
yet
simply
do
a
project
export
to
get
them
out
of
there
if
you're
still
actively
using
ITP
for
performing
demos,
then
specify
a
date
in
which
you're
saying
I'll
be
moved
everything
over
by
then,
but
right
now,
I'm
simply
trying
to
reduce
the
active
user
list
in
ITP
to
work
toward
a
deprecation.
So.
L
L
C
Now
we're
gonna
be
doing
is
using
the
EU
instance
of
the
demo
cloud,
so
we've
had
the
u.s.
one
that
provides
the
sandbox
the
EU
one
will
be
available
very
in
the
next
week
or
two
here,
and
the
goal
of
that
is
everyone
has
admin
rights
on
it,
so
you
can
go
in
and
see
it
and
because
we
have
sample
data
and
projects
in
it.
You
can
see
the
admin
settings
for
those
sample
projects,
because
that
instance
will
be
rebuilt
on
a
recurring
basis.
It
will
keep
things
clean.
C
You
know
as
we
go,
but
it's
effectively
that'll
be
the
right
option
for
it.
If
you
need
an
alternative
to
it,
you
can
always
spit
up
a
docker
container
with
get
up
in
it
and
grab
it
off
of
that.
If
you
really
need
it,
but
that's
what
we
are
trying
to
solve
that
problem
with
the
EU
instance
perfect.
Thank
you.
That's
awesome,
yeah.
C
We
are
close
on
time,
so
this
is
a
two-minute
time
warning
so
mark
your
question
will
be
the
last
one
and
if
you
do
have
an
additional
follow-up
questions,
there
is
a
AMA
I'm
having
tomorrow
open
forum
up
for
anything
else.
If
there's
no
questions
on
the
AMA
will
go
ahead
and
cancel
it,
but
just
mark
feel
free
to
they.
G
C
For
a
while,
you
don't
have
anything
to
worry
about,
so
we're
not
going
to
be
launching
and
doing
the
Rd
on
the
ephemeral
clusters
until
q3
and
I
suspect
we'll
do
another
three
to
six-month
deprecation.
So
if
I
had
to
make
a
guess,
November
December,
maybe
sooner
if
we
end
up
with
cost
control
issues
but
it'll
it'll,
be
it's
a
ways
out
great.
C
Very
good,
thank
you,
everyone
for
for
joining
us
today
and
yeah.
If
you
have
any
follow-up,
questions,
feel
free
to
join
the
AMA
tomorrow
and
that'll,
be
basically
a
demo
demo
designer
deep
dive
and
focusing
on
how
we're
going
to
build
up
our
sample
projects,
it'll,
probably
dive
into
the
EU
instance
and
play
with
a
couple
things
if
there
are
not
additional
Martians
on
there,
but
otherwise.
Thank
you
so
much
for
joining
taking
the
time.