►
From YouTube: Configure stage vision part 2
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
A
So
yeah
so
kubernetes
has
a
way
for
you
to
specify
things
like
replicas.
So,
for
example,
if
one
of
your
serving
parts
dies
like
another,
one
will
pick
up
the
slack,
but
that,
like
certain
applications,
may
behave
very
differently
depending
on
what
they
are.
So
it's
always
good
to
kind
of
test
your
system
for
resiliency,
and
this
provides
you
like
a
very
easy
way
to
kind
of
configure
and
forget
it.
So
did
you
hear
me
talk
about
like
each
one
of
the
things
that
you
can
configure
with
chipmunky
yeah.
A
A
The
vision
that
we
have
for
chaos
is
twofold,
so
first
one
is
adding
cube
monkey
as
a
part
of
a
good
lab
managed
app,
and
this
is
kind
of
changed
a
little
bit.
As
you
know,
right
now
we
install
lamps
with
a
single
click.
Cute
monkey
is
gonna,
be
one
of
these
new
apps
that
you
install
through
CI,
so
we're
gonna,
you
know,
provide
you
a
template
and
then
provide
you.
A
Hopefully
what
you
see
won't
change,
so
you
will
see
cube
monkey
in
your
good
lab,
managed
apps,
you
click
a
button
and
then
the
CI
job
will
run
and
we'll
install
it
in
your
cluster.
So
that's
great!
That
has
some
value,
but
the
reality
is
that
users
can
do
that
today,
so
users
can
install
cube
monkey
quite
easily
under
kubernetes
cluster.
The
real
value
or
the
more
added
value
will
come
when
we
automatically
allow
you
to
configure
chaos.
A
Engineering
through
Auto
DevOps,
so
I'm
really
excited
about
this
one,
and
this
is
about
adding
cube
monkey
support
to
Auto
DevOps.
So
what
I
picture
will
happen
here
is
that
you
don't
have
to
basically
do
anything
to
opt
in
to
chaos.
You
can
just
enable
cube
monkey
through
a
CI
variable
and
it
will
run
with
the
defaults
that
would
be
kind
of
the
first
stage.
Then,
when
it
kills,
one
of
your
parts
will
have
data
about
the
failures
I'm,
not
sure
where
this
data
will
be.
This
is
part
of
something
that
I
hope.
A
You
will
help
me,
which
is
that
we
inflicted
some
chaos
on
your
cluster
because
you
turned
it
on
and
the
auto
devops
pipeline.
We
we
use
the
key,
and
then
we
returned
a
report
saying
you
know.
This
is
how
much
time
your
app
was
down,
how
much
time
it
took
for
another
pot
to
spin
up
and
take
this
like
what
was
lost.
You
know
what
sessions
you
know,
I,
don't
know
what
kind
of
units
we
will
express
this
in,
but
you
monkey
has
a
lot
of
documentation
that
will
help
us
with
that.
A
So
this
is
kind
of
what
I
picture
is
gonna,
be
the
real
value
or
more
added
value.
Is
that
users
that
want
to
inflict
chaos
on
their
app?
They
don't
have
to
be
an
expert
on
the
tool
or
chaos
and
engineering.
They
can
just
basically
give
us
a
CI
variable,
saying
chaos
enabled
or
a
cue
Monkey
enabled
something
like
that
and
and
basically
set
to
yes,
and
that
will
tell
us
to
you
know,
kill
pots
the
default
values.
A
Then
they
do
some
code
changes
to
add
more
resiliency
to
their
app
and
then
they
do
it
again
and
then,
when
they
get
to
a
point
where
they
feel
that
they
have
added
in
a
resiliency
to
their
app,
then
they
test
it
in
production,
because
the
real
value
comes
in
like
real-world
testing
in
production.
When
something
happens,
how's
your
app
going
to
react
so,
like
Netflix,
is
famous
for
doing
this
in
production
like
they.
A
They
did
that
all
in
a
staging
environment,
they
run
a
bunch
of
tests
at
a
bunch
of
code
to
make
sure
that
things
die.
Grace,
gracefully
and
users
are
not
impacted,
and
then
they
just
setup
chaos,
monkey
on
production
to
randomly
kill
stuff,
and
that's
where
the
values,
like
that's
I
would
say.
That's
like
putting
let's,
let's
just
saying,
there's
the
saying
about
putting
your
dollars
or
your
where
your
mouth
is
or
something
like
that,
but
that
is
the
real
value.
A
A
Yeah,
so
it's
it's
funny,
but
they
have
like
times
of
the
day
like
around
from
6
p.m.
to
9
p.m.
it's
their
busy
time,
and
because
most
people
like
watch
net
Netflix
to
unwind
after
work
and
stuff
like
that
and
their
traffic
is
just
insane
so
yeah.
At
those
times,
I've
also
seen
like
an
error
message.
They
have
to
try
again,
but
then,
when
I
try
a
second
time,
it
works.
It's
not
like
it's
standing
for
an
extended
period
of
time,
right.
B
A
No,
so
you
can
default
the
way
that
you
deploy
to
be
like
straight
deploy
to
production,
which
is
less
usual,
and
then
you
can
say
continuous
deployment
to
or
no
automatic
deployment
to,
staging
and
manual
deployment
interruption.
So
what
I
picture
is
that
this
is
how
most
people
have
it
setup
is
that
it
automatically
deploys
into
staging
and
then
I
picture
them
running
their
tests
there
and
see
how
things
react,
making
changes
and
stuff
like
that
and
then
manually
the
point,
interruptions.
B
A
A
A
That
that's
a
good
point,
that's
something
that
probably
the
testing
group
will
have
to
work
on
and
I
wonder.
Testing
group
unit
testing
integration
test
load
testing
see
so
we
already
have
a
plan
from
load
testing
in
2020.
So
how
great
would
it
be
to
kind
of
unify
these
two
things
of
saying
you
know:
I,
want
load
testing
with
my
chaos,
engineering
and
like
have
a
dashboard
with
all
the
results
that
would
be
pretty
cool.
That.
B
A
B
A
I
this
is
anecdotally.
I
can
tell
you
the
users
that
I've
spoken
with
they
have
the
setting
in
Auto
DevOps.
You
know
as
far
as
I
know,
this
is
not
actually
something
that
is
instrumented.
So
that's
a
good
point.
That's
a
good
takeaway!
Let's,
let's
take
a
look.
If
it's
in
on
the
usage
pane,
we
doubt
it
yeah.
B
A
A
A
You
cannot
do
that
through
the
API,
so
you
can
create
a
cluster
through
the
API
or
not
create
you
can
add
an
existing
cluster
for
the
API
and
you
can
set
the
base
domain
to
the
API.
But
those
are
all
the
those
aren't.
The
only
two
things
that
you
can
do,
so
we
have
issues
pending
to
provide
the
ability
to
install
charge
via
the
API.
This
is
all
very
fluid
right
now,
because
we're
kind
of
fundamentally
changing
the
way
the
charts
are
installed
into
your
and
pure
cluster,
so
I
had.
B
A
Though
you
you
can
check
out
the
configure
dashboard,
and
here
it
gives
you
kind
of
a
total
of
Auto
DevOps
pipelines.
So
here
I
have
tiles
that
give
you
the
total
number
of
CI
pipe.
Why
pipelines
that
run
in
your
lab
versus
the
number
of
Auto
DevOps
pipelines
that
run
so
we
can
see
that
you
know
out
of
28
million
on
average
nine
hundred
and
forty
four
thousand
are
Auto
DevOps
pipelines
yeah.
So
we
that's,
like
you
know,
small.
A
Well,
there's
a
couple
of
issues
there,
but
I
think
that
the
main
one
is
that
all
of
them
UPS
is
running
everywhere.
If
you
have
like
a
text-based
project,
that's
a
wiki
or
you
know
anything
it'll
run
on
any
sort
of
code,
so
we
have
to
be
smarter
and
we
have
to
make
sure
that
out
of
DevOps
only
runs
when
it
can
add
value
and
then
we'll
start.
A
Those
were
built
out
from
like
legacy
apps,
so
the
images
are
very
heavy
and
they
take
a
long
time.
So
we
want
to
make
that
more
efficient
as
well
with
the
use
of
cloud
native,
build
packs.
So
just
kind
of
you
know
a
couple
of
things
that
we
have
in
the
pipeline.
We
hope
will
kind
of
add
more
value
to
the
autumn.
Devops
workflow,
but
we
know
we
have
some
work
to
do
for
sure.
B
Good
I
think
it's
course
having
a
conversation
with
you
again
since
you've
spoken
to
many
customers
and
users,
maybe
a
separate
meeting
again
for
taking
me
through
like
companies,
you
know
that
you
get
club
that
leads
out
of
DevOps
how
they
use
it.
What
their
needs
are
etcetera,
I
think
you
have
the
best
number,
even
more
than
so.
Expressor.
A
A
I'll,
send
you
this
link
so
because
our
DevOps
doesn't
fit
like
a
hundred
percent
of
the
use
cases,
and
we
know
that
don't
want
to
build
that
in
a
lot
of
people,
take
inspiration
from
our
DevOps
and
the
plug
ability
right.
So
you
know
our
DevOps
is
composable,
so
you
don't
have
to
use
all
of
it,
so
you
can
use,
let's
say
only
code,
quality,
sass
and
dependencies
scanning.
A
If
you're,
if
you're
composing
your
own
pipeline
based
on
audit
Auto
DevOps,
we
are
not
counting
those
as
far
about
pipelines.
So
that's
kind
of
another
problem
that
we
have
is
that
in
my
mind,
those
should
count
towards
our
DevOps
use,
because
you're,
taking
advantage
of
this
great
templates
that
we've
built-
and
we
are
not
seeing
that.
So
we
need
to
instrument
the
usage
of
templates.
A
B
I'm,
so
confused,
what
part
is
worth
looking
at
doing
on
the
UI
and
what
part
would
be
more
adoptable
if
that's
word
on
from
the
command
line.
So
if
you
want
to
use
even
a
chunk
of
the
other
DevOps
will
be
worth
splitting
them
on
the
you
like
use
as
a
complete.
What
DevOps
solution
or
a
pick
testing
or
pick
I.
A
B
A
B
B
A
A
B
A
A
A
A
A
B
A
A
Yes,
that
would
be
helpful,
so
you
need
to
kill
actually
a
couple
of
things
because
it's
not
as
simple
as
if
you've
killed
a
cluster.
There
are
all
these
other
things
that
you
had
to
create,
like
you
know
your
fancies
and
your
notes
and
all
that
crap
basically
you'll
have
to
kill
because
it
cost
money.
Well,
you
don't
have
to.
A
A
A
When
come
across
a
certain
scenario
and
I'll
show
you
some
real
world
one
book,
so
you
kind
of
know
what
what
we're
talking
about
so,
for
example,
this
are
they
get
lab,
run
books
that
the
operation
teams
use
to
basically
know
what
to
do
when
they
come
across
a
certain
scenario.
So
let's
say
there's
some
downtime
or
there's
some
thing
that
they
need
to
do
so
like
here
it
tells
you
what
happens
when
you
come
across
an
incident.
First,
don't
panic
and
things
like
that
so
like
it
walks,
you
step-by-step
on
what
to
do
right.
A
A
A
Notebooks
and
Jupiter
notebooks
are
have
you
come
across
this
notebooks
before
yeah,
so
they
are
very
popular
in
the
data
science,
world
and
people
use
them
to
share
data,
and
the
great
thing
about
notebooks
is
that
you
can
build
in
code
to
your
notebook
and
run
it
right
inside
the
notebook
and
get
a
result
right
from
the
notebook
and
there's
a
really
cool
video
here
that
basically
those
over
how
this
works
in
ingot
lab.
Let
me
pause
this
and
how
this
looks.
Well,
you
know
what
I'm
gonna,
pull
it
up
and
I'll.
A
So
you
can
look
at
that
video.
So
that's
what
a
jupiter
notebook
is
basically
and
jupiter.
Notebooks
are
served
by
a
jupiter
server
right
and
we
allow
you
to
install
this
thing
called
jupiter
hub
into
your
kubernetes
cluster,
and
it
basically
provides
you
a
way
to
have
multiple
instances
of
jupiter
notebook
servers.
So
when
I
install
it-
and
I
go
to
a
bit,
let's
take
a
look.
A
And
sign
into
my
notebook
with
get
lab,
I'm
going
to
authorize
good
lab
to
use
my
authorization
information,
and
then
I
thought
that
this
server
was
already
up
running.
I
guess
not
so,
while
that
comes
up,
this
is
where
we
want
you
to
store
your
runbook
information,
basically
because
of
two
things.
The
first
thing
is
the
rubix
library,
so
this
library
is
an
open
source
library.
That
is,
that
allows
you
to
take
pre-built.
A
Patterns
to
operate
on
infrastructure,
for
example,
if
you
were
to
write,
let's
say
a
query
to
an
AWS
database.
You
know
you
were
to
write
all
that
code.
Let's
say
would
be
a
hundred
lines
right
where
you
have
to
like
specified
database
and
what
kind
of
interaction
you
want,
and
you
want
what
you
want
to
return.
A
Let's
not
say
a
database.
Let's
say
some
cloud
watch
metrics
like
metrics
around
your
own
service.
So
what
the
rubix
library
does
is
that
it
has
a
bunch
of
interactions,
kind
of
built-in
that
make
it
as
simple
as
a
single
line
to
basically
interact
with
a
kubernetes
cluster
or
interact
with
Klat
watch
or
some
other
AWS
constructs.
A
So
this
rubric
library
is
built
by
a
company
called
Newark
and
there
are
friends
of
gitlab
a
meet
who's.
The
founder
and
CEO
has
helped
us
with
a
bunch
of
their
own
box
functionality
inside
get
labs.
It
largely
has
been
kind
of
an
effort
that
parts
of
it
have
been
built
at
good
laughs
and
part
of
its
have
have
been
built
by
an
urge
alright.
So
let's
take
a
look
at
the
at
the
run
book.
A
So
here
when
you
install
Jupiter
onto
your
cluster,
we
give
you
a
sample
one
block
here,
and
this
is
what
the
run
book
looks
like.
It
basically
has
a
setup
that
tells
you
what
kind
of
credentials
you
have
to
run
and
this
sold
in
the
video
by
by
the
way,
and
it
would
be
great
to
do
as
SN
as
an
experience
baseline
to
this
is
how
to
configure
an
executable
one
book
with
good
lab.
B
A
But
basically,
this
is
all
this
is
our
vision
for
run
books
that
it's
basically
it's
not
only
text
that
you
go
from
this
system
and
then
come
out
of
it
and
then
go
back
to
it
and
then
come
out
of
it,
but
all
the
actions
that
you
need
to
run.
You
run
them
right
within
within
the
notebook
within
the
run
book.
Sorry.
A
So
well,
you
can
think
of
a
how
to
run
book.
It
is
kind
of
a
script.
A
script,
however
I
think
is
different
in
that
you
run
it
in
in
in
an
isolated
context.
Right
so
you
say,
run
this
script
and
then
you
run
it
and
you
have
to
go
to
the
CI
job
to
view
the
results
and
then
then,
what
do
you
do
next
right?
So
the
great
thing
about
having
this
build
and
you
can
think
of
our
run
book
or
sorry.
A
You
can
think
of
a
Jupiter
notebook
as
a
set
of
scripts
that
are
built
in.
But
the
great
thing
is
that
when
you
run
them
like
well,
this
is
not
not
configured,
so
it's
not
gonna
work,
but
all
you
need
to
do
is
press
this
play
button
at
the
top,
and
then
it
will
run.
You
know
whatever
queries
you
you
have
and
it
will
return
the
values
for
those
queries.
So
let's
say
if
one
is
for
plant
watch,
metrics
it'll
plot
the
grunts
right
into
your
notebook
and
then
based
on
that
data.
A
A
Run
books
are
not
used
to
test
the
code;
no,
they
are
used
to
like
deal
with
scenarios
right
like
hey.
Your
production
database
is
down
what
do
I
do
so
as
an
SRE,
you
say:
okay,
let
me
pull
up
the
run
book
for
database
now
and
I'll
pull
it
up
and
it
has
a
series
of
steps
and
I
followed
those
steps.
So
if
you
look
at
what
run
books
are
today,.
A
If
you
look
at
what
run
books
are
today,
there
are
a
series
of
steps
that
are
written
down
somewhere
now,
I
have
to
go
to
other
system
and
execute.
So
the
great
thing
about
executable
run
books
is
that
imagine
that
each
one
of
the
steps
had
a
play
button
where
the
screen
size
and
it
does
it
for
you
automatically
yeah.
B
A
It
could
be
a
script
or
it
could
be
some
code
to,
let's
say,
run
a
query
in
a
double.
Yes,
it
could
be
plotting
metrics,
it
could
be
so
see.
This
is
what
is
returned
here
is
an
error
because
my
things
are
not
or
not
figured,
but
if
you
watch
the
video
I
think
you'll
get
a
better
view
of
how
how
it
would
work.
Oh
well,.