►
From YouTube: Kubernetes Office Hours 20180321 (EU Edition)
Description
Come join us, third Wednesday of every month! https://github.com/kubernetes/community/blob/master/events/office-hours.md
A
I
am
kicking
off
the
stream
every
one
time
we're
ready
all
right.
Welcome
everybody
to
kubernetes
office
hours,
European
Edition
and
the
UK
I'm
Jorge
Castro
I'm
gonna,
be
your
host
I'm
gonna
introduce
everyone
here,
real,
quick
and
then
we'll
we'll
get
started.
Why
don't
we
go
Ilya,
then
booyah,
then
Jeffrey
and
then
Bob
introduce
yourselves
hey.
B
Everyone
I'm
here
from
Lee
works,
joining
from
our
London
office
and
I,
be
be
looking
forward
to
your
questions
today
and
the
the
areas
of
expertise
of
mine
are
around
continuous
delivery.
That's
working
cost
the
provisioning
and
various
other
things
too,
as
we
have
many
other
panelists
who's.
Next,
oh
yeah,.
D
E
A
Awesome
well
awesome
to
have
our
panelists
here
today,
like
we
said
we
do
this
of
the
third
Wednesday
of
every
month
for
those
of
you
who
are
new
viewers,
so,
first
of
all
feel
free
in
the
slack
channel
to
let
us
know
how
the
audio
quality
is.
We've
put
a
lot
of
effort
into
trying
to
making
this
a
high-quality
production.
So
let
us
know
how
the
volume
is.
A
Second,
let
me
just
crank
up
the
game.
There
how's
that
so
I'm
gonna
talk
about
how
it
works
like
why
we
did
this.
Basically
Ilya
and
I
were
there's.
There's
certain
ways
that
you
can
learn
how
to
use
kubernetes.
You
can
read
books
of
documentation
and
watch
videos,
but
they're.
Really
a
really
great
way
to
learn
is
when
you
go
to
a
conference
or
a
meet-up,
and
then
someone
actually
explains
something
to
you
and
like
high-bandwidth
manners,
you
know
method
that
you
can
understand
it.
A
So
one
thing
led
to
another:
we're
like
hey:
we
should
do
this
where
people
just
come,
they
ask
their
questions
and
then
we've
got
kubernetes
developers
on
the
line,
and
this
is
kind
of
your
hour
to
ask
expert
kubernetes
users
and
developers
questions
on
all
sorts
of
stuff.
So
first
things:
first,
the
slack
channel
is
gonna,
be
a
judgment-free
zone.
So
someone
shows
up
and
asks
a
quote-unquote,
dumb
question.
There
are
no.
A
Unanswered
questions
so,
let's
be
a
place
where
we
were
collaborative
and
supportive
of
all
of
our
efforts.
I
think
a
lot
of
us
have
struggled.
You
know
getting
kubernetes
to
do
exactly
what
you
wanted
to
do.
So,
let's
be
supportive
in
that
matter,
and
what
we
will
do
our
best
to
answer
your
questions.
A
Please
be
aware
that
we
can't
like
actually
ssh
into
your
cluster
and
look
at
your
stuff,
so
we
try
to
generally
answer
questions
that
are
more
general
or
that
can
teach
you
something
as
opposed
to
you
know.
Why
is
my
firewall
not
working
right?
Well,
tell
you
where
to
look,
and
perhaps
the
piece
of
documentation
that
will
help
and
worst
case
will
show
you
what's
sig
and
where
you
can
go
to
get
help
to
do
that
right
now,
we've
had
a
pretty
good
success
rate
with
that
I.
Don't
think.
A
We've
actually
had
a
question
yet
that
I'm
aware
of
that,
we
totally
were
stumped.
You
know
unless
it
was
someone's
restricted
network
at
work
and
there's
not
much,
we
can
do
for
you,
they're
panelists.
You
were
encouraged
to
expand
on
your
answers
with
your
experiences
and
pro
tips.
So
you
know
a
lot
of
this
might
be
just
like
hey
look
out,
you
know
this
is
the
spec.
Oh
go,
read
it
right,
but
feel
free
to
put
in.
You
know
your
little
tidbits
of
information
that
you've
gotten,
especially
when
you're
running
things
in
production.
A
Inevitably
someone
will
ask,
what's
a
CI
CD
I
should
be
using,
and
then
everyone
goes
and
pastes
in
their
favorite
one,
and
we
come
up
with
a
really
great
list
of
tools
that
people
might
want
to
try
or
you
know,
cool
utilities.
Things
like
that
all
of
these
sessions
are
recorded
on
YouTube,
going
all
the
way
back.
So
what
we're
trying
to
do
is
build
this
as
like
a
corpus
of
community
expertise.
A
A
If
you
are
an
expert
and
doing
a
lot
of
great
things
with
kubernetes,
you
can't
be
on
this
panel
with
these
fine
set
of
people,
you
do
get
a
cool,
kubernetes
water
bottle
as
part
of
your
contribution
to
the
community,
by
sharing
your
expertise
soon
as
soon
as
I
get
approval,
we're
gonna
be
holding
raffles
and
things
like
that,
so
you'll
be
able
to
earn
a
cool
kubernetes
t-shirt
stickers.
Are
you
gonna?
Do
this
ask
questions
be
involved
and
we're
gonna,
make
it
all
fun
and
get
the
kubernetes
spinners?
A
Everyone
wants
and,
lastly
feel
free
to
hang
out
in
hash
office
hours
right,
like
the
the
user
channels,
how
many
people
kubernetes
def
right
now,
something
like
33,000
people.
Sometimes
it's
hard
to
get.
You
know
some
signal
out
of
that
noise
so
feel
free
to
make
this
like
one
of
your
home
channels
to
hang
out
with
if
you're,
finding
someone
who's,
you
know,
maybe
struggling
with
the
same
thing.
You're
doing
get
to
know
each
other
kind
of
hang
out.
This
is
like
our
little
I.
A
We've
carved
our
little
corner
in
in
the
slack
thing
to
be
like
hey.
This
is
where
we're
gonna
help
new
people,
here's
what
we're
gonna
do
office
hours,
type
stuff,
so
you're
more
than
welcome
to
hang
out
throughout
the
rest
of
the
month,
and
things
like
that
and
with
that
we'll
begin.
So
the
way
it
works,
you're
gonna
type,
a
question
in
hash
office
hours.
A
Please
preface
it
with
question:
I,
do
question
:
and
then
bla
so
I
see
it.
If
I
don't
see
it
and
accidentally
skip,
it
feel
free
to
just
repeat
it
again.
If
you
see
something,
that's
a
really
good
question:
do
a
reaction
on
it
with
like
a
thumbs
up
or
something.
So
if
something
is
an
overwhelmingly
great
question
that
everyone
wants
to
know,
it's
obvious
to
me,
and
we
could
do
that
and
with
that
I'm
just
gonna
panel,
how
I'm
feeling
we're
ready
we're
ready
to
do
this
I.
A
Good
to
go
yes,
okay,
all
right,
so
I'm,
gonna
scroll
back
as
we
had
some
questions
queuing
up
before
we
we
started
here
all
right.
The
first
question
is
gonna,
be
from
Vincent
SEOs,
hi,
I'm
new
to
Cooper,
nighties
and
DevOps
in
general.
My
question
is:
what
is
the
most
widely
adopted
community
supported,
full-feature
CI
CD
pipeline
to
use
with
kubernetes?
You
can
always
get
this
question
then
he
goes
on
to
say:
is
hound
Jenkins
the
standard?
So
I
know
this
group
has
some
opinions
on
CI
CD
stuff.
B
I
think
the
the
latest
and
greatest
thing
is
actually
what
Google
just
released
a
couple
of
weeks
ago
called
scaffold
and
I.
Think
that
is
something
that
that
should
be
of
interest
to
anybody.
You
can.
It
is
actually
a
pluggable
framework.
So
if
you
want
to
use
hound
today
and
play
around
with
that,
you're
welcome
to
do
that.
B
If
you
would
like
to
stop
using
home
eventually
for
whatever
reason
and
switch
to
JSON
and
perhaps
or
a
case
on
it
or
anything
of
that
nature,
you
could
do
that
than
the
future
and
we
will
be
demoing
how
it
integrates
with
our
CI
TD
component
flux.
They
get
your
conservation
operator
for
the
genomes
flow.
So
then
yeah
watch
this
space.
It's
gonna
be
coming
shortly,
but
scaffold
is
definitely
something
you
should
be
looking
at.
D
I
can
say
with
experience
for
smaller
projects
and
if
you're,
a
tinkerer
I
use
drone
CI
for
my
home
set
up
and
it's
integrated
with
kubernetes
quite
well,
however,
for
more
enterprise
applications.
What
we're
using
right
now
is
get
lab,
CI
and
get
lab.
It
has
some
really
nice
integrations
and,
if
you're
doing
more
than
just
kubernetes
I
think
that
tends
to
it's
a
bit
more
flexible.
So
you.
C
Mainly
on
on
circle
CI
these
days
but
kind
of
independent
from
the
CI
itself,
most
of
our
customers
that
using
Jenkins
began
to
prices
so
they're
used
to
it.
Jenkins
also
has
just
released
this
Jenkins
X
I
think
it's
called
a
Jenkins
specially,
especially
for
communities
haven't
looked
into
it
yet,
but
it
sounds
promising
at
least
should
be
a
bit
more
lightweight.
Maybe
do.
B
You
mind
if
I
insert
opinion
ads
when
I
looked
at
it.
If
it
seems
a
little
strange
that
they
kind
of
try
to
extend
cute
little
experience.
But
that's
that's
what
it
seems
like
you
know.
They
kind
of
are
the
new
command
for
you
to
learn
and
you've
sort
of
relates
to
capital.
But
it's
something
else
which
I
thought
is
a
little
bit
strange.
E
E
E
D
C
C
E
A
Mm-Hmm
and
one
thing
I
wanted
to
piggyback
off
Bob's
answer
is
there
is
no
golden
standard,
it
seems
a
lot
of
it
are
sometimes
it's
like
business
dependent.
You
know
like
it.
Sometimes
it
feels
like
whatever
cic
DS
solution.
Pipeline
thing
you
have
is
like
almost
directly
tied
to
your
company
culture
and
in
some
cases,
so
there's
no.
Like
default.
One
I've
seen
I've
seen
everyone
make
the
argument
for
every
single
one.
A
So
all
right,
I
hope
that
answers
your
question
and
any
last
bits
on
CI,
CD
stuff.
All
right.
Moving
on
MRD
em
I
asked
question
deploying
on
Prem
darrell
VM
to
using
ki
Badman
in
an
enterprise
lab
environment,
can
I
use
pre-generated,
search
for
kubernetes
components
or
do
I
work
on
getting
the
Kate
see
a
quote:
trusted
not
likely.
E
C
A
E
E
It
is
like
you
can
use
it
in
production,
but
you
like
you,
cannot
treat
it
like
a
normal
kubernetes
end
point
it's
more
of
like
a
place
where
you
can
it's
like
a
deployment,
end
point
and
you
let
the
Federation
then
decide
where
it's
gonna
place
things
in
they're,
not
all
crazy
objects
are
supported
and
there's
a
lot
of
other
things
that
you
have
to
do
to
get
it
like
to
fully
support
it.
It's
it's
pretty
decently
explained
in
there.
But
if
you
want
to
know
more,
I
highly
recommend
hopping
into
like
cig
multi
cluster.
A
And
then
one
thing:
if,
if
someone
could
Google
this
for
this
and
whack
it
in
the
channel,
is
list
of
kubernetes
SIG's?
Well,
we
have
one
page
that
has
all
the
SIG's
and
for
those
of
you
that
aren't
aware
all
the
work
in
kubernetes
is
done
by
SIG's
special
interest
groups
around
a
specific
area
or
a
working
group.
So
usually
we
could
start
off
with
a
general
area,
and
then
you
can
drill
down
and
find
this
thing.
A
That's
actually
working
on
that
feature
and
all
the
SIG's
have
public
meetings
that
are
similar
to
this
on
zoom'.
You
can
watch
the
archives
on
YouTube
and
things
like
that,
and
that's
usually
where
we
defer
to
when
we're.
Looking
for
a
very
specific
question
about
a
feature,
I
think
you
attend
those
meetings
regularly
about
yep.
A
C
If
you
have
a
CI
pipeline
anyway,
like
CI
CD
having
one
endpoint
or
five
shouldn't,
be
a
huge
difference,
so
at
least
from
the
deployment
side,
you
can
get
it
and
from
the
Geo
DNS
side
on
on
the
outside,
you
can
get
it
through.
You
won't
have
failover
between
single
services
between
clusters
might
be
able
to
use
core
DNS
there.
He
does
that's
right,
that's
good,
too
yeah.
B
I
was
concerned
that
you
could
use
the
core
DNS
with
multiple
clusters
if
you
need
to
cross
cross
the
service
discovery
right,
so
you
could
say
my
servers,
don't
cost
the
one
or
whatever
I
mean
the
naming
scheme
would
would
be
potentially
a
bit
more
complicated,
but
essentially
I
could
do
that.
So
you
could,
you
could
do
the
cross.
Cluster
service
discovery
score
DNS
and
if
you
need
cross
cross,
the
connectivity
I
I'm
more
than
happy
to
tell
you
more
about
how
that
could
work.
E
B
Like
there
are
few
use
cases
that
the
Federation
control
plate
is
good
for,
but
there
are
many
things
you
could
do
without
that,
and
you
can
implement.
Federation
is
sort
of
loosely
coupled
fashion
with
cross
cluster
network
service.
Discovery
in
networking
was,
for
example,
core
DNS
and
we've
met,
and
this
point
I
said
for
deployment.
Lee
I
won't
see
once
you
have
a
very
robust
CIC
new
pipeline
and
it's
not
not
that
hard
to
synchronize
configuration
to
different
postures
and
if
you,
the
detox
approach
that
will
help
you
a
lot.
A
E
A
Right
phaser
ask
I'm
running
a
cluster
of
1.9
point
Exxon
EWS
to
play
with
cops
still
struggling
with
the
persistent
problem
of
EBS.
Bang
equals
read/write
many
I
wonder
if
there
will
be
a
solution
to
the
SUNY
FS
is
too
expensive.
Sassa
FS
are
clusters
seem
to
need
some
adjustments
of
the
worker
nodes.
Is
there
some
proposed
solution
to
have
a
read,
write
many
persistence
based
on
EBS.
D
E
D
Shouldn't
say
right
in
this
context:
yeah
the
way
we
solved
it
wound
up.
Actually,
we
weren't
using
EF
or
we
didn't
use
EFS.
We
actually
spawned
an
ec2
instance
to
act
as
an
NFS
server
and
then
mounted
EBS
to
that,
because
that
gave
us
better
performance
and
we
knew
it
was
a
hack,
but
we
tried
to
get
something
going
faster
and
it
worked
for
us
for
a
few
months,
but
I
think
that's.
That
is
a
limitation
with
EBS,
not
kubernetes.
Just.
A
B
Just
wanted
to
add
that
we
listened
to
the
like
many
experts
here
on
storage,
most
of
the
time
lost
lost
time.
We've
also
had
a
storage
question
and
that
there
wasn't
anybody
to
to
help
us
out
without
perhaps
perhaps
some
some
of
the
the
viewers
who
may
know
more
about
storage
would
like
to
join
us
fun.
The
next
time
perhaps
or
I
mean
you
can
always
hop
on
the
final
now,
if
you'd
like
to
so,
please
do
let
us
know
and
yeah
did
anybody
want
to
add
anything
to
this
topic?
I'm.
A
C
C
A
Okay,
we
got
some
awesome
information
here
for
everyone
from
the
viewers
right
now
so
Ben
Hall
likes
to
add
+1
for
drone.
It's
a
very
nice
and
straightforward
for
small
projects
feel
similar
to
get
lab,
which
is
nice
plug
cat.
A
coda
has
some
content
to
help.
You
understand
how
to
configure
it
and
then
he's
popped,
a
link
on
how
to
configure
that
into
the
channel.
Thanks
for
that,
Ben
and
Vlad
would
like
to
mention
that
Jenkins
x
user
exists
on
slack
as
well
as
hash
jenkins,
CI
on
the
kubernetes
slack.
C
A
E
C
It's
soon
to
be
production
already,
I
guess,
but
there
is
already
a
lot
of
users.
You've
tried
it
out
we're
also
kind
of
looking
forward
to
it
being
closer
to
production
already,
so
people
can
use
it
to
see
more
different
scenarios
and
yeah.
Basically,
it
does
what
served
us
just
with
less
headaches
for
yourself,
because
you
don't
need
to
manually
set
any
steps
up.
Yeah.
A
A
F
C
E
B
C
You're
on
a
juror
I
think
there's
a
lot
of
New
Ager
functionality
like
VM
scale
sets
coming.
Hopefully
there
is
a
last
buck
we
found,
hopefully
that'll
be
fixed,
a
great
team
at
Microsoft
and
other
than
that.
Some
type
of
occasions
might
be
there
at
workloads.
Api
is
one,
but
that
was
already
in
1/9.
A
A
A
E
A
A
D
C
A
D
G
A
Been
shared,
a
really
interesting
link
here
for
kubernetes
feature
OSS
tracking
board,
which
is
a
spreadsheet
that
has
all
the
features
I
totally
forgot
that
existed.
Thank
you
for
tossing
that
in
there
suresh
we'd
like
to
ask
question:
can
you
guys
share
your
opinion
on
the
kubernetes
service?
Catalog
open
service
broker
API?
How
will
it
be
used
in
a
cluster.
C
So
my
eye
opening
use
case
was
because
I
was
really
thinking
about
this
hard
like
when
all
in
the
guys
kind
of
started,
introducing
it
and
my
eye
opening
news
case
was
the
integration
of
image
out
so
that
you
can
provision
basically
cloud
managed
databases
manage
database
denied
ugliest
like
an
RDS
or
what
you
have.
What
have
you
on
measure
and
and
Google
cloud
through
companies
and
get
get
back
like
your
username
and
password
injected
into
your
pod?
It
was
from
me
kind
of
the
eye
opening
news
case.
A
So
I
have
sent
in
a
few
of
their
their
meetings
in
the
past
and
they're,
a
great
bunch
of
guys.
So,
if
you're
looking
to
give
them
feedback,
absolutely
just
join
one
of
their
meetings,
they're
always
looking
for
user
feedback
and
contributors
as
well.
So
all
right
moving
on
angel
says:
are
there
any
actual
real
use
cases
for
rooting
kubernetes
at
your
house
other
than
just
playing
around
see?
It
started
off
as
a
joke,
and
here
we
are
Tanna.
D
A
For
me,
for
me:
it's
all:
it's
all
learning
stuff
I
mean
I,
don't
like
how
do
you
I,
don't
know
I
can't
I
can't
see
myself.
It's
not
like
I
have
a
Power
Distribution
unit
at
home
and,
if,
like
one
of
my
nose
loses
power
like
the
whole
house
loses
power
right,
so
I
don't
know
what
I
would
actually
use
it
for
so
I
kind
of
I
I.
Do
it
for
learning
Jeff
I
know
you
so.
D
D
A
Meant
for
me,
that's
all
I
learned
I
set
up
a
mail
server
at
home
back
when
you
could
set
up
mail
servers
and
not
have
everything
blacklist
you
so
I.
Do
it
mostly
to
learn
I,
just
a
cube
admin
blow
away
my
thing
and
just
reinstall
it
every
other
week
and
I'm,
just
like
I
kind
of
I
kind
of
do
it
for
that
kind
of
familiarity
as
well.
A
E
The
big
thing
is:
look
at
SEF,
look
at
the
performance
requirements
around
SEF
and
look
at
how
you're
going
to
set
that
up,
and
it
really,
if
you're,
going
to
be
setting
up
with
like
one
or
two
nodes,
you're
not
going
to
get
good
performance
out
of
it.
You
need
a.
You,
definitely
need
to
invest
in
your
storage
infrastructure
to
get
decent
performance.
D
A
A
E
A
C
A
lot
of
certs
and
I
mean
for
one
customer
be
looking
into
SSO,
but
I
would
try
to
go
with
something
Oh
IDC
and
then
maybe
have
something
in
between
like
Tech's,
keep
low
or
zero
or
something
to
abstract
away.
Your
AIT
management
provider
will
be
able
to
federal
rate
a
few
providers
into
because
you
can
only
have
one
or
IDC
integration
and
in
the
API
server.
A
A
A
E
A
E
A
E
A
A
B
So
well,
essentially,
one
one
simple
way
to
do.
This
would
be
to
let
scaffold,
run
your
build
and
then
and
then
push
the
images
to
its
registry,
and
once
that's
done
you
could
you
just
you
know,
wait
for
those
images
to
sink,
at
least
with
our
implementation
of
get
your
installation
operator
which
is
reflux
on.
We
can
have
placed
a
link
to
that
momentarily,
but
essentially
yeah.
We
just
pick
up
the
images
from
the
registry.
B
B
That
new
image
is
available
before
you
before
the
following
internal
16
right
so
yeah.
It's
essentially
just
let
it
deploy
it
straight
away,
so
that
that's
this,
that's
one
one
of
the
integrations
that
I
have
in
mind
right
now
that
there's
a
little
bit
more
to
it
that
I'm
thinking
of
doing,
but
it's
all
it's
all
in
flux
at
the
moment,
but
I'm
yeah.
B
A
Okay,
wanted
to
push
a
push
an
example
to
and
then
there's
a
link
of
a
kubernetes
off
example
there
and
we'll
make
sure
we
put
all
of
these
in
the
show
notes.
So,
mr.
ball
CBS,
how
do
you
manage
local
versions
of
all
your
tools
tops
cube
control,
cube,
CL,
etc?
In
particular,
when
you
have
multiple
clusters
of
varying
versions,
1.7
X,
1.8
X,
all
the
different
point
releases-
is
it
all
manual?
Are
there
tools
for
managing
the
tools?
A
E
B
For
example,
g-cloud
as
an
official
container,
the
image
that
we
can
use
I'm
not
entirely
sure
about
all
the
other
ones,
but
yeah
so
definitely
cube,
could
always
pretty
good
at
backup
visibility.
You
know
very
few
things:
I
saw
that
broke
actually
struggling
to
recollect,
and
it
is.
It
is
also
always
possible
to
use
all
the
version
of
Takata
was
a
new
version
of
the
cluster
API.
C
C
A
A
A
Alright,
so
moving
on
Joel
speed
wants
to
mention
the
first
part
of
my
SSO
series
is
available
on
the
new
stack
now
the
next
few
parts
coming
Friday
next
Friday
happy
to
discuss.
If
anyone
has
particular
questions
and
then
he
plopped
a
link
to
that,
so
make
sure
you
check
that
out
on
the
new
stack
dot,
IO
Ivan
goes
on
to
ask
okay.
If
we
talk
about
scaffold
here
goes
another
question:
how
do
you
do
your
local
development
with
kubernetes
telepresence
mini
cube,
mounting
volumes,
etc,
etc?
A
G
C
Think
the
user,
a
lot
of
mini
cube,
some
of
us
use
development
containers
inside
a
communities
cluster,
so
developing
right
on
the
classic
kind
of
and
against
it.
We
don't
do
a
lot
of
kind
of
mounting
volumes
and
like
having
an
editor
on
that
mount
reloading
a
container.
That's
usually
still
build
the
thing
and
run
it
in
a
new
mini
cute.
A
C
E
D
E
C
A
D
A
A
E
B
B
C
E
C
A
A
A
A
A
A
C
A
B
B
Isn't
that
easy
to
start
with
so
yeah
so
that
that's
that's
one
thing:
I,
don't
exactly
know
whether
the
husband
used
all
the
Hellinger
could
get
fairly
reasonable
set
over
things,
but
you
still
need
to
do
quite
a
bit
of
work
to
get
to
go,
find
that
dashboard.
Things
like
that
puto
and
I
can
I
can
really
suggesting
things
other
than
our
product.
For
that
purpose,
it's
a
beef
quality
supposed
to
Prometheus
I
wish
there
was
an
alternative
method
to
suggest
that
is
easy
to
use.
But
for
now
that's
that's!
F
C
Dashboards,
it's
always
hard
because
there's
no,
no
nice
open
dashboards
out
there
and
you
need
to
build
your
own
and
then
you
need
to
know
Griffin
and
firm
uses
queries.
Then
you
need
alert.
So
maybe
a
good
entry
like
seeing
people
use
we've
cloud
and
you
have
unlimited
storage
to
rent.
That's
that's
a
big
benefit.
A
Let's
see
some
more
opinions
here,
Vlad
says
many
cube
works
for
small.
You
know
talking
about
local
development,
mini
cube,
works
for
small,
isolated
experiments.
If
you
use
it
for
serious
to
have
work,
you
have
to
mock
a
lot
and
you
have
to
maintain
a
completely
separate
environment
with
separate
bugs,
which
is
how
so
it's
a
little
just
a
warning
there
and
dims
how
you
doing
today.
He
goes
I
use,
hack,
slash,
local
up,
clustered
Sh
script
for
my
dev
workflow.
B
A
Please
do
not
use
that
for
work,
don't
use
things
in
and
that
for,
like
your
work,
production,
cluster
Sean,
says
topology,
aware
routing
sounds
great
I'll.
Look
into
that.
Thank
you
thanks
for
coming
web.
Worse
paste
it
in
the
Prometheus
operator,
which
you
might
want
to
investigate
here
for
setting
that
up.
A
Why
p21
sad
that
the
Prometheus
operator
includes
some
pre-configured
dashboards
for
Griffin
as
well.
Thanks
for
that
follow-up,
Sean
says
ideally
the
same
box.
We
have
a
proprietary
logging
daemon
that
are
shared
between
pods
that
are
on
each
host.
Also,
the
idea
of
these
global
packages
that
all
app
services
need
access
to,
ideally
moving
to
fluent
D
are
similar,
but
these
use
their
own
protocol.
Any
of
the
comments.
Now
we
have
some
some
more
information
there
from
Sean.
C
A
Okay,
10
minute
warning,
so
if
you
have
questions
get
them
in
now,
we're
gonna
finish
in
about
five
minutes
here
and
then
we'll
go
again
in
about
six
hours
for
for
the
u.s.
folks.
Well,
lead
was
to
Asshai
would
like
to
play
with
the
HP
HPC
scheduler
slurm
under
cube.
It
has
one
role
as
a
master
and
others
as
agents
should
I
use,
stateful
stuff
for
the
first
role
for
the
master
and
every
time
I
would
like
to
scale
the
nodes.
I
need
to
reconfigure.
D
Unreasonable
in
my
wheelhouse,
because
we
work
in
an
HPC
shop,
we're
like
the
outliers
yeah,
so
we
actually
wrote
a
an
integration
with
kubernetes
and
slurm
that
we
dub
slurm
Nettie's
I
have
linked
it
in
the
chat.
I
could
go
in-depth,
except
I
have
a
feeling.
This
is
not
gonna
apply
to
like
all,
but
maybe
two
people
that
are
interested
in
this,
the
silic.
That
was
a
work
in
progress
that
I
did
over
the
holidays.
So
it's
not
exactly
production.
In
fact,
it's
not
even
really
alpha
a
few.
D
Yeah,
the
the
other
thing
that
I
would
like
to
link
to
any
other
HPC
type.
People
is
a
an
SSH
session
gateway
that
Bob
wrote
and
I
just
kind
of
sat
over
his
shoulder
and
told
him
what
to
do
for
a
little
bit
of
it,
but
with
those
two
combined
we're
actually
looking
to
take
HPC
workloads
and
move
them
towards
kubernetes.
There's
also
another
project
that
I'm
trying
to
build
out
that
is
HPC,
but
completely
in
kubernetes.
It's
not
trying
to
shoehorn
some
other
project
to
use
kubernetes.
A
Day,
that's
that's
perfect!
That's
why
we
have
these
meetings.
Well,
you
know,
hopefully,
that'll
give
you
some
information
there
to
get
started
thanks
so
much
any
final
questions
before
we
wrap
it
up
and
color
to
call
it
a
month
thanks
so
much
everyone
for
attending.
As
we
say
we
we
can
do
these.
Basically,
as
often
as
we
want
right
now,
we're
doing
it
once
a
month.
The
only
limiting
factor
is
getting
people
on
the
panel
and
questions.
So
if
you
have
I
think
we're
probably
about
do
to
do
like
a
storage,
one
I
think
so.
B
Wanted
to
I
wanted
to
to
note
that
we
also
do
one
session
at
Cube.
Calm.
Yes,
well,
we've
been
a
live
session
at
coupon,
so
those
of
you
who
are
who
are
looking
to
come
to
cube
con,
no
I,
don't
know
George.
What
do
we
do
there?
Exactly
can
I
like
lock
ourselves
in
their
room
or
we
got
to
be
in
front
of
an
audience,
then
pass
the
mic
around.
It's.
A
Gonna
be
just
like
this,
except
either
I'll
be
running
around
with
a
microphone
to
the
audience,
questions
or
we'll
have
like
a
place
for
you
to
stand
in
line
and
then
ask
your
question
and
yeah.
Well,
it's
literally
gonna
be
one
long
table.
We'll
probably
have
more
people
on
the
panel.
You
know
to
make
sure
we're
more
well
rounded,
like
gonna.
B
F
A
B
A
A
Kubernetes
contributor
summit,
we
actually
have
there's
going
to
be
like
sessions
for
people
who
are
existing
contributors
where
they'll
you
know
talk
about.
The
features
are
working
on
whatever,
but
we
also
have
a
new
contributor
track.
Then
I
want
to
talk
about
which
will
have
people
saying:
hey
here's
how
you
work
through
your
first
pull
request:
here's,
how
you
here's,
how
I
get
github
workflow
works,
because
ours
is
different
because
they're
so
big
and
it's
not
a
traditional.
A
Well,
you
go
to
github
and
click
buttons
start
workflow,
so
we're
gonna
go
that
we're
kind
of
tailoring
content
for
new
contributors
for
people
that,
let's
say,
you're
an
advanced
user
or
maybe
you're
you're,
ready
to
kind
of
maybe
fix
your
first
bug
or
you
know,
get
more
involved
instead
of
just
consuming
kubernetes.
So
if
you're
gonna
be
a
Q,
Khan
and
you're
interested
on
kind
of
leveling
up,
you
know
crossing
over
from
user
to
developer.
This
would
be.
Ideally,
this
would
be
a
great
place
to
go.
A
F
B
A
B
B
A
Yes,
that's
right.
We
forget
that
one
all
right
and
with
that
we're
gonna
go
into
the
outro
real,
quick
thanks
thanks
to
everyone
who
who
tuned
in
to
listen
and
ask
your
questions.
This
really
helps
us
out
thanks
to
our
panel,
for
showing
up
your
expertise
is
incredible,
and
I
would
like
to
thank
the
following
companies
for
allowing
these
fine
engineers
to
volunteer
their
time.