►
From YouTube: Kubernetes Office Hours 20191016
Description
Office Hours is a live stream where we answer live questions about Kubernetes from users on the YouTube channel. Office hours are a regularly scheduled meeting where people can bring topics to discuss with the greater community. They are great for answering questions, getting feedback on how you’re using Kubernetes, or to just passively learn by following along.
Discussion thread and show notes for this episode: https://discuss.kubernetes.io/t/office-hours-for-october-16/8288
For more info: https://github.com/kubernetes/community/blob/master/events/office-hours.md
B
Is
going
to
win?
Welcome
everybody?
It
is
the
third
Wednesday
of
the
month.
That
means
it
is
time
for
the
kubernetes
office
hours,
the
monthly
livestream,
where
we
hop
online
and
try
to
answer
as
many
of
your
questions
as
possible
about
using
kubernetes
with
our
awesome
panel
of
experts
so
how's,
the
stream
sounding
you
should
be
in
Hache
office,
ours,
which,
if
you
read
the
text
below
here
on
the
live
stream,
feel
free
to.
Let
us
know
how
the
audio
sounds.
B
C
All
right
found
I
found
on
mute
hi.
My
name
is
Mario
Lauria
I
live
in
southern
Michigan
right
now,
working
for
stock
ex
as
a
senior
DevOps
engineer,
pretty
much
owning
our
entire
cluster
environment.
Eks
on
Amazon
I'm
also
really
involved
with
community,
especially
with
George,
and
the
Ann
Arbor
region
and
I
contribute
some
another
kubernetes
things
as
well
as
this
fantastic
podcast,
slash
broadcast.
D
Hi
everyone
I'm
Chris
Carney
I'm,
a
senior
developer
with
the
City
of
Ottawa,
who
I'm
more
or
less
in
charge
of
our
committees
infrastructure.
So
it's
kind
of
a
generalist
in
that
regards
so
try
to
be
involved
in
the
community
as
much
as
I
can
and
do
this
I'm
also
on
the
release
team,
where
I'm.
Currently,
the
release
lead
for
release,
notes
on
this
for
the
117
release.
E
Hi
I'm
Joel
speed,
I
work
for
a
company
called
pusher
out
of
money
I'm
on
the
construction
team
here,
so
I
do
a
lot
of
stuff
to
do
communities,
including
a
bunch
of
stuff
recently
around
extending
it.
So
if
you've
got
any
controller,
questions,
I've
been
building
lots.
Those
recently
I,
don't
know
lot
of
security,
stuff
and
all
just
getting
stuff
in
the
past
as
well.
So
those
kind
of
questions
vary.
My
way
awesome.
F
A
Okay,
so
I'm
Don,
Foster
I
work
for
pivotal
in
London
and
I'm
responsible
for
our
open
source,
kubernetes
contribution
strategy
and
the
program
that
goes
along
with
it.
So
I'm
focused
on
getting
the
right
people
within
pivotal,
contributing
to
open
source
communities.
That's
my
jam.
I've
been
doing
up
a
source
for
like
20-some
years,
so
I've
been
around
awesome.
B
And
this
is
Don's
first
time
so
welcome
and
bargah.
This
is
your
second
time
if
you're
new
feel
free
to
like
dip
in
as
much
or
as
little
as
you
want.
So
before
we
begin,
let's
set
some
ground
rules.
First
of
all,
everybody
I'd
like
to
thank
you
all
for
joining.
There
are
a
bunch
of
you
listening
today.
At
least
thirty,
that's
good.
So
this
is
judgment
free
zone.
Everyone
had
to
start
from
somewhere.
So
there
are
no
dumb
questions.
B
So
please
don't
flame
someone
who
for
asking
what
you
might
think
is
a
dumb
question
and
remember
that
the
kubernetes
code
of
conduct
is
in
effect,
so
be
excellent
to
each
other.
We've
never
really
had
a
problem
with
that
here.
Everyone's
excellent,
that's
awesome!
When
we
appreciate
that
and
while
we
will
do
our
best
to
answer
your
questions,
the
panel
doesn't
have
access
to
your
cluster,
so
live
debugging
is
off
topic.
B
You
know
we
can't
really
like
look
at
all
your
IP
tables
and
figure
out
what's
going
on,
but
what
we
will
try
to
do
is
try
to
like
at
least
show
you
where
to
look
or
some
tips
of
tricks
from
these
people
who
either
work
on
kubernetes
or
have
production
experience
to
help
at
least
get
you
going
somewhere.
So
you're
not
stuck
on
something
panelist.
B
You're
encouraged
to
expand
on
your
answers
with
your
experiences
and
pro
tips
feel
free
to
a
common
thing
is
you're
thinking
about
something
you
think
about
a
tool
or
something
take
the
URL
and
just
stick
it
in
chat.
We
always
appreciate
that
audience.
You
can
also
help
us
by
pasting
URLs
to
the
official
Docs.
Maybe
you
saw
a
blog
post
or
something
maybe
someone's
talking
about
a
tool
that
you
have
experience
with.
B
Stick
them
in
the
discuss
thread,
which
Bob
will
be
pasting
into
slack
today,
as
our
show
notes,
and
then
you
can
look
at
the
URLs
and
all
that
kind
of
stuff
there.
You
can
also
help
us
out
by
tweeting
spreading
the
word
paying
it
forward.
We
really
love
to
help
people,
so
you
know
anything
you
could
do
to
recommend
us
to
somebody
or
a
friend
or
even
if
you
can't
participate,
live,
we
always
love
when
people
go
back
and
listen
to
the
backlog
and
give
us
information.
B
We're
always
listening
on
how
to
make
this
show
better.
Before
we
start,
though,
I
would
like
to
thank
the
following
companies
for
supporting
the
community
by
lending
us
their
fantastic
engineers
that
you
see
here:
giant,
swarms,
stock,
X,
pivotal
pusher,
calm,
weave
works,
VMware,
the
University
of
Michigan
Red
Hat,
Utility
Warehouse
and
the
City
of
Ottawa
Ontario
Canada,
and
always
a
special
thanks
to
the
CAF
to
sponsor
away
the
sponsor
our
t-shirt
giveaway.
B
So
if
you
ask
a
question
and
we
read
it
live
on,
air
will
automatically
enter
you
in
a
raffle
and
at
the
end,
you
have
to
wait
till
the
end.
Sorry
I'm
gonna
give
away
for
t-shirts.
Normally
it's
two
but
we're
not.
We
don't
have
a
November
session
due
to
Cube
Khan,
so
I'm,
giving
away
four
shirts
so
ask
good
questions
and,
as
always,
when
you're
done
feel
free
to
hang
out
and
office
ours.
What
we
do
is
be
in
between
each
session.
B
We
do
collect
those
questions
and
use
it
to
see
our
next
episodes
and
it's
also
a
lot
easier
to
handle
them.
The
77,000
or
whatever
people
who
are
in
kubernetes
users
and
if
you're,
joining
us
from
that
channel.
Welcome
welcome.
Please
ask
a
bunch
of
questions.
We
have
someone
from
Serbia,
Australia
and
Kazakhstan
already
awesome.
Alright,
let's
get
started
panel.
Are
you
ready
give
me
a
thumbs
up
all
right,
I
see,
masks.
B
I
am
using
Mike
Gendler
Burnett
knees,
GSEs
SFTP
image
to
run
an
SFTP
server,
but
I
have
a
requirement
to
authenticate
users
via
SSH
keys
and
I
have
around
30
plus
users.
How
can
I
manage
users
and
their
SSH
keys?
I
know
those
questions
is
about
a
specific
docker
image,
but
maybe
you
can
panel,
you
can
give
us
some
general
tips
as
well.
Who
wants
to
take
this?
One
I
can.
C
C
So
people
would
do
a
PR
to
add
their
public
key
to
the
the
docker
image
and
that
you
know
I'll
throw
his
keys
file
in
the
docker
image
and
then
a
build
would
trigger
when
that
was
merged
and
then
it
would
be
deployed
and
they
would
have
to
be
able
to
have
access
to
that
that
image
it
would
get
redeployed
to
the
the
cluster
with
the
new
image.
So
that's
a
fairly
simple
way
of
doing
it.
C
E
E
What
it
can
do
is
you
can
get
it
to
sign
certificates
that
people
can
use
to
SSH
in
so
your
SFTP
server
would
authenticate
against
fault
trust,
vault
as
a
certificate
authority
and
then
every
time
a
user
wants
to
access
your
SFTP
server,
they
could
say:
can
I
have
a
certificate
for
15
minutes
assuming
their
vault
credentials
haven't
expired.
It
will
then
create
a
certificate
based
off
of
their
SSH
key
and
then
the
pair
of
those
can
be
used
for
them
to
SSH
or
SFTP.
E
H
B
Okay,
moving
on
sorry,
my
I
lost
my
place
there
and
slack
Alex
welcome
ass
hi
Yoann,
a
ready
cluster.
How
do
I
determine
the
CNI
in
use?
This
feels
like
the
help
my
clusters
broke
and
call
somebody,
so
you
know
actually
generalize
it
up
a
little
bit.
What
are
some
things
that
maybe
a
cluster
you
have
not
set
up.
C
Yeah
one
of
the
first
things
I
would
do
is
kind
of
a
accuses
he'll
get
all
start
looking
at.
Definitely
the
daemon
sets
are
running
that
those
main
top
level
objects
services,
daemon
sets
deployments
and
staple
sets,
started
looking
at
that
and
get
an
idea
of
what's
going
on
in
all
namespaces,
especially
cute
system,
so
there
isn't
really
economical
way
of
just
getting
the
C&I
there
isn't
a
cat
at
cos
release
if
you
will
so
yeah.
G
So
one
thing
you
can
do
is
get
quads,
as
will
tell
you
what
kind
of
CNI
and
that's
an
a
file
called
ET
c
/a
TCC
and
I'm
net
dot,
t--
that
who
can
who
and
look
at
if
you
do
a
character
for
the
consciousness.
It
will
probably
tell
you
the
cni
that
you're
trying
to
use.
So
I
think
that
is
the
best
way
that
I
know
of
today.
There's
no
way
like
you
can
go
very
certain
end
point
to
get
the
CNI
right
now,
but
this
is
what
I
know
you
can
get
to
see.
D
B
Okay,
moving
on
so
storm
or
asked
today's
curiosity
that
I
wonder
if
he
has
cross
anyone's
path.
I
just
had
an
entry
point
script
and
a
parent
container
behave
in
an
erratic
fashion.
After
the
child,
docker
file
was
executing
RM
dash,
FR,
slash
temp
semicolon
make
their
Dash
P
slash,
temp,
so
I'd
know
any
recommendations
for
emptying
slash
temp
as
part
of
the
docker
file.
The
script,
for
the
most
part,
is
a
simple
wrapper
and
if
I
run
into
the
underlying
command,
it
still
functions.
B
H
C
B
E
B
B
C
C
High
power
yeah-
that's
that's
an
excellent
guy.
It
can
take
a
while,
but
it's
gonna
give
you
the
fundamentals
and
really
break
you
in
there's.
Actually
a
book
that
hefty
Oh
released
awhile
ago
called
managing
kubernetes
and
I.
Believe
it's
it's
free.
You
should
be
able
to
find
it,
but
the
Google
search
was
free
release
that
I
actually
really
enjoy,
because
it
gives
you
both
kind
of
the
administrative
side
of
things
as
well
as
the
client
side
and
interacting,
but
also
you
know,
really
dig
into
the
control
paint
clean
elements
as
well.
C
So
that
is
super
helpful
and
then
there's
a
candidate
handbook
from
the
CN
CF.
There's
a
blog
form,
I
believe
one
of
the
have
deal
engineers
as
well.
How
have
the
engineers
aced
the
TKA
it's
a
little
bit
old,
but
I?
Think
some
of
the
stuff
is
really
relevant
there.
So
long
as
you
apply
it
to
the
latest
cure
nice
version
which
will
be
tested
against.
So
definitely
do
some
googling
and
I'll.
Try
to
post
some
of
these
in
the
slack
as
well.
Mm-Hm.
E
But
I
didn't
mind:
I
found
this
practice
environment.
That
was
really
really
useful.
So
you
set
up
a
couple
of
key
baby
m
nodes
on
your
own
machine
and
it's
got
a
bunch
of
questions.
They're
kind
of
like
practice,
questions
that
you
can
execute
against
it
and
it
was
really
useful
to
just
actually
have
some
useful
physical
practice
who
opposed.
E
That,
in
slack
it's
yeah,
it's
eerily
similar
to
the
actual
exam.
Probably
shouldn't
say
that
that
is
really
really
good.
So
I'd
recommend
trying
that
out
and
at
Barcelona
coupe
con.
There
was
a
really
good
talk
which
again
are
all
leaked
by
olive
power
called
Ready,
Steady
CKA.
That's
got
some
really
good
tips
in
it,
especially
the
cube,
controlled
dry
run
stuff
that
she
mentions
really
really
useful.
B
Wow
that
practice
environmentalist
links
that
one
I
didn't
even
know
about
that
practice
of
iron
right,
that's
a
good
one.
There
meiosis
I
hope
my
kid
that
right
says.
Thanks
for
the
tips
on
the
cke
I've
bought
the
Linux
Foundation
kubernetes
fundamentals
course
plus
exam
yesterday,
that's
good
to
know:
does
anybody
else
in
the
audience
have
any
tips
for
the
cka
or
if
your
CK,
a
wave
feels
like
there
are
more
of
you
everyday
yeah.
D
There's
also
the
hashtag
CK
a
prep
thread
within
the
kubernetes
lock
they
annulled
bunch
of
good
resources
or,
if
you're
thinking
about
studying
it's
a
good
way,
a
good
place
to
kind
of
bounce
ideas
off
of
people,
and
they
have
a
spreadsheet
with
a
bunch
of
those
resources
for
helping
people
so
grab
the
link
and
drop
that
into
the
thread.
But
you've
really
helped
me
for
my
last
minute.
Cram.
B
H
B
Routing
problems
yep
and
Joe
tell
Rico
once
I
mention
that
there's
a
tool
called
metal,
3
IO
as
well
I'll,
go
ahead
and
paste
that
into
the
main
channel
and
Max
guy
says,
loves
metal
I'll
be
I've
been
using
it
in
BGP
and
l2
mode
didn't
run
into
any
problem
so
far
and
Aaron
Liefeld
wants
to
have
it.
He
also
uses
bare
metal
and
production
metal
help
bees.
So
it
sounds
like
people
are
using
it.
That's
good,
any
any
other
comments
here
on
metal
lb.
B
If
you
have
a
specific
problem
with
metal,
I'll
be
just
feel
free
to
ask
a
follow-up
question
and
then
we
will
get
to
it
as
soon
as
we
can.
Okay
moving
on
jorik
asks.
Is
there
a
way
to
access
control,
plane,
Prometheus
metrics
between
different
managed
environments,
able
to
retrieve
API
server
metrics
easily,
but
unable
to
reach
the
scheduler
at
CD
and
the
controller
manager.
H
G
Right
then,
when
you
want
to
come
back
to
you,
I
think
there
are
a
couple
of
questions
here.
One
is
to
access
the
Prometheus
metrics
from
different
management,
something
called
a
nose
which
pulls
from
multiple
Prometheus
servers
and
you
know,
show
all,
except
at
one
place
and
move
us
the
data
to
a
storage
system
and
that's
one
way
of
aggregating
all
the
metrics
from
different
environments.
As
far
as
collecting
from
the
control
plane
matrix
from
scheduler
xilian,
Control,
Manager,
I'm,
not
sure
if
they
do
have
a
matrix
endpoint.
Maybe
I
can
get
back
on
that.
G
B
B
All
right,
Alexander,
hello
from
Germany
cluster
AP,
has
did
a
good
job
of
managing
a
fleet
of
kubernetes
clusters
at
the
same
time,
cotta
containers
or
Marant
osvárt.
Let's
are
great
on
running
on
vm's.
Instead
of
container
/,
pods
storage
layer
maybe
saw
by
cluster
FS
or
rook.
Is
there
a
project
to
join
some
of
these
to
manage
multiple
kubernetes
clusters
hosted
in
VMs
on
top
of
another
bare
metal,
kubernetes
cluster.
A
D
B
D
B
B
B
E
Always
had
the
awesome
operators
list
handy
with
these
kind
of
things
and
they've
got
four
different
designer
operators
there.
So
one
might
be
a
good
one
to
look
at
okay.
It's
that
navigator,
I've
heard
good
things
about,
but
okay
ever
use
it
myself.
If
you
do
have
questions
mother's
on,
is
quite
active
on
slack
for
the
Jets
that
he
works
with.
Oh,
yes,.
B
Do
do
do
we
want
to
invite
him
in
maybe
I'll
poke
him
yeah.
We
should
have
him
on
this
show
that
would
be
fantastic
while
you're
at
it
ask
him
if
he
wants
to
join
the
show.
Okay,
hopefully
that
will
give
you
four
options
there
for
Cassandra
kubernetes
operator.
If
you
have
any
feedback
on
that,
please
get
back
to
us.
That
would
be
fantastic,
so
we
can
help
to
share
that
with
the
community.
Tiwari
asks
I
team
kind
of
an
open-ended
question
for
the
experts
with
ISTE
Oh,
slash,
linker
DV.
B
All
the
rage
nowadays,
when
do
you
feel
introducing
service
meshes,
makes
sense
for
your
kubernetes
setup.
A
lot
of
companies
are
moving
towards
these
technologies
aggressively.
But
what
would
your
ideal
parameter?
Slash
scale
when
you
might
consider
using
service
meshes
thanks
in
advance.
Let's
time
limit
box,
this
question
to
five
minutes
overdue:
it
who's
got
opinion
on
service
meshes
and
when
you're
ready
to
I
guess,
move
up
the
stack
I.
C
I
just
want
to
say
something
like
to
preface
everything
else
when
you
need
it
is
the
answer:
don't
jump
into
something,
because
it's
all
the
hype
and
sounds
really
exciting
like
make
sure
that,
in
your
operational
troubleshooting,
monitoring,
alerting
processes
and
for
what
you
were
companies
and
your
goals
are
trying
to
achieve
like
make
sure
it
makes
sense
first
before
jumping
in
because
it
can
be
some
dense
waters,
and
you
know
definitely
lots
of
other
important
things
as
well
when
running
a
kubernetes
cluster,
especially
in
reliability
and
security
and
scalability.
So
that's
my
two
cents.
E
There's
a
lot
of
overhead
that
comes
with
them,
as
Mary
was
saying,
like
we've
got
sto
running
on
our
clusters
in
production,
that
sort
of
only
just
going
to
production.
Having
been
worked
on
for
about
six
seven,
maybe
eight
months
now,
but
it
really
I'm
not
sure
we
actually
needed
it,
but
a
lot
of
the
stuff
it
provides.
We
maybe
could
have
got
away
with
something
much
smaller
and
yeah.
E
Last
night
we
had
a
three
or
four-hour
instant,
because
our
customer
just
exploded,
because
we
got
loads
of
this
new,
auto
scaling
and
new
constant
stuff
everywhere,
and
it
just
just
it
can
kill
the
cluster
so
bear
in
mind.
There
is
like
extra
containers
in
every
pot
and
then
there's
a
whole
bunch
of
other
pods
on
top
of
that
logging
metrics
it.
Yet
there's
a
lot
of
overhead
that
you
might
want
to
be
wary
of.
If
you
are
thinking
about
running
it
so
yeah
you.
H
B
Okay,
anything
else
before
we
move
on
to
the
next
one,
all
right,
everybody
we
are
slowly
but
surely
catching
up
to
all
the
questions
feel
free
to
answer
them.
If
I
haven't
asked
your
question
and
you
feel
like
I've
accidentally
skipped
it,
please
just
let
me
know
it
helps
if
you
put
question
in
like
in
all
bold
to
make
it
easy.
Everyone
is
tossing
tons
of
information
today,
you're
gonna
make
gathering
the
notes
really
painful,
I
appreciate
all
of
that.
B
So
please
keep
all
the
information
going
on
macaroni
and
welcome-back
says
sto
as
Abby's
I
reduce
all
the
resource
requests
for
testing
and
max
guy
wants
to
mention
that
rook
has
Cassander
support
as
well
for
a
previous
question,
but
the
Cassandra
operator
so
he's
dropping
a
link
there.
Okay,
anything
else
on
service
meshes
before
we
move
on
awesome,
great
nobody's
feelings
were
hurt.
Outstanding
linen
asks
any
way
to
access
a
deployment
/
job
such
stateful
sets
annotations
from
a
container
I'm
only
able
to
get
annotations
from
the
pod
spec
through
environment
variables,.
H
B
Anybody
else
seems
pretty
straightforward
for
this.
One
I
see
lots
of
people
typing,
we'll
give
you
a
second.
If
you
want
to,
we
are
about
halfway
through
those
of
us.
Those
of
you
joining
us
from
YouTube
welcome.
This
is
the
kubernetes
office
hours,
the
monthly
live
scene,
where
we
try
to
answer
on
your
questions,
hop
into
office
hours
on
the
kubernetes
slack,
and
ask
your
question:
we
are
giving
away
four
t-shirts
today,
so
ask
your
question
and
we
will
give
wait
for
t-shirts
and
I'll.
Tell
you
what
I'll
reserve
two
of
those
we'll
do.
B
Two
two
t-shirts
for
the
people
that
ask
questions.
Those
of
you
giving
a
really
great
expert
advice,
I'd
like
to
give
you
a
t-shirt
at
well.
You
know
what
it's
cute
cars
coming,
I'm
just
gonna,
give
away
a
bunch
of
t-shirts
today
uh-huh.
How
does
everyone
feel
about
that?
If
you're
just
joining
us
at
chat,
please
let
us
know
what
you're
working
on
and
where
you're
in
the
world,
so
we
can
see
where
we
are
reaching
everybody.
Okay.
B
B
H
C
B
Will
I
know
one
of
them:
I'm
gonna
ask
for
a
demo
for
a
community
meeting
or
something
I.
Remember
the
booth
a
cube
con,
but
we
should
check
it
out
so
yeah,
sorry
about
that.
If
anybody
else
has
experience
on
garner
feel
free
to
type
it
in.
Please
keep
answering
your
questions.
These
are
fantastic.
All
right
ma
asks
anyone
got
tips
for
people
wanting
to
start
contributing
to
the
kubernetes
projects
and
I've
got
a
question
for
this.
I've
got
an
answer
for
this,
but
I'll
go
last.
I
can.
A
Jump
in
on
this
one,
there
are
some
great
contributor
Docs,
which
will
post
the
links
to
in
the
in
the
slack
channel
and
in
the
notes.
So
that's
a
great
place
to
start.
Another
good
place
to
start
is
to
pick
pick
a
cig
pick,
a
cig
that
you're
interested
in
and
you
know,
poke
around
in
their
issues
and
pull
requests
and
take
a
look
at
what
they're
doing
and
find
something
interesting
to
work
on.
A
If
you're,
not
sure
what
cig
you
want
to
work
on,
get
your
bitter
experience
and
Doc's
are
two
really
super
friendly,
great
starter,
SIG's
and
from
there
you
can
actually
learn
a
lot
about
contributing
to
the
project,
and
so
those
are
good
places
to
start.
If
you
don't,
if
you
don't
know
elsewhere,
those
are
my
tips.
Awesome.
B
And
I
would
like
to
mention
that
we
do
have
a
sister
show
to
this
called
meet
our
contributors
that
goes
on
the
first
Wednesday
of
every
month
with
pears
Pitman.
It's
basically
just
like
this
format,
except
you're,
asking
questions
about
how
to
contribute.
There's
Cooper
daddy's
developers
there
you
can
ask
them
how
they
got
started.
They'll
show
you
how
they
do
code
review.
Sometimes
we
have
code
walkthroughs
through
architecture.
It
is
a
fantastic
book
end
to
the
show
this
shows
about
using
kubernetes
and
that
shows
about
contributing
to
kubernetes.
H
Couple
other
like
little
resources
within
the
careers
YouTube
channel,
you
can
find
a
set
of
playlists
for
like
a
new
contributor
workshop,
that's
sort
of
like
it's
broken
up
and
it
will
go
over
like
our
github
workflows,
there's
an
intermediate
track
that
does
a
little
bit
of
a
few
like
code
based
tours.
We
actually
have
the
code
based
tours
broken
out
into
a
separate
list
now
and
if
you
are
still,
you
know
just
looking
to
jump
right
in
take
a
look
for
issues
that
are
labeled
with
like
no
help
want
it
or
good.
B
B
C
C
You
have
a
requirement
for
it
or
you
think
you
could
give
a
lot
of
value
from
whatever
that
next
core
service
could
be
or
prometheus
metrics
or
something
like
that
and
then
I
would
definitely
start
with,
and
maybe
even
if
you're,
not
using
home,
but
charts
home
charts
and
operators.
We
we
linked
that
awesome
operators,
that's
a
great
way
to
get
kind
of
an
understanding.
Powell
applications
are
put
together
and
deployed
for
things
like
aj
and
whatnot,
so
yeah
it
definitely
take
it
slow,
but
yeah.
B
B
B
Okay,
so
Dimitri
asks
hey
guys,
wanted
to
get
your
thoughts
on
service
on
kubernetes.
We
currently
use
AWS,
Landa's
and
I
had
a
thought
to
get
more
utilization
out
of
our
cluster
by
running
lambdas
inside
our
cluster.
There's
a
project
called
queue.
Bliss
just
wanted
to
get
your
thoughts
on
this
approach.
I
know:
we've
mentioned
open
faz
before
here
on
the
show
who's
who's
messing
with
this
I
suspect,
Mario.
C
C
We're
not
we're
not
actually
doing
land
on
kubernetes
and
I.
Don't
think
we're
planning
on
to
our
edge
team
has
been
doing
graphics
well
on
kubernetes,
which
I
don't
have
many
stories
from
yet.
But
let's
wait
for
an
outage
in
the
middle
Black
Friday
and
then
I'll
have
a
good,
a
good
one,
but
yeah
lambda
has
been
fun.
I
think
lambda
leaves
a
lot
to
be
desired
and
I
think
you
lose
a
lot
of
control,
and
so
there
has
been
talk
about
like
do
we
go
to
2q
Bernays
with
it
I
think.
C
If
we
did
again,
we
definitely
look
at
the
kubernetes
options.
I
think
what
is
it
now?
Kay
native
and
I?
Don't
even
remember
cube
lists
I
forget,
so
someone
help
me
out
open
fast,
the
other
one
kata.
B
B
B
If
Jonas
was
listen,
he
might
be
able
to
help
me
out,
but
that's
okay,
keeping
popping
in
those
service
URLs
in
slack
FM
project.
That's
that's!
A
new
one.
I
always
learn
of
a
new
project.
Every
time
I
do
this,
which
is
part
of
the
reason
why
we
do
do
this.
Venkat
asks
see,
unite
questions
so
I
have
a
cube
admin,
initialized,
1.16
cluster
running
with
flannel
as
the
overlay
Network.
Is
it
possible
to
switch
to
calico,
slash
weave
network
on
a
live
cluster
without
having
to
rebuild
the
cluster
I?
B
Don't
care
about
the
pods
or
deployments
I
deployed
with
all
the
other
cluster
components
running?
Can
this
be
achieved
thanks?
Actually,
this
seems
easier
than
the
last
person
who
wanted
to
switch
seeing
eyes,
I,
remember
and
keep
it
all
up
and
running,
and
then
we
ask
them
to
report
back
and
we
never
heard
from
them
again.
B
H
Have
successfully
done
this
before
there
are
lots
of
caveats,
and
it
really
started
depends
on
what
you're
actually
using
and
you
might
need
to
do
like
an
IP
Tate
like
clear
out
your
IP
or
something
like
that
in
between
I
would
not
recommend
it,
but
it
is
possible
to
do,
and
every
situation
is
probably
gonna
be
a
little
bit
different.
Is.
B
C
Think
they
should
start
over
per
se.
I
think
maybe
like
a
rolling
reboot
of
nodes.
So
instead
of
rolling
update,
dude
the
manual
option
and
then
and
then
rolling
reboots
of
you
know,
one
node
comes
up
new
scene
is,
you
know,
continue
I,
don't
think
they
need
to
like
blow
away
their
entire
cluster.
I
know
that's
a
little
easier
like
for
sure
cloud.
Has
it's
an
deserve,
but
like
for
Amazon,
you
can
blow
away
a
node
group.
No
new
node
group
move
your
workloads
to
that.
C
H
Actually
left
the
pods
up
and
going
we
were
able
to
you,
you
know
we.
We
went
through
your
Dana
that
know
that
the
rebooted
notes
and
brought
them
back
later
and
just
sort
of
like
progressively
shifted
things
around
it.
It
did
work
and
it
was
in
a
production,
live
cluster
that
it
was
kinda
had
to
run
into
a
problem
with.
H
B
Yeah,
so
you
know
they're
saying
they're
using
cube
admits
so
be
right.
You
know
I,
don't
obviously
we
don't
have
access
to
their
environment,
but
on
paper
that
should
make
reads,
standing
up
the
cluster
very
much
easier,
I,
don't
know
that
depends
on
their
needs.
I
feel
like
was
something
like
it's
important
to
see
and
now
you're
better
off,
protecting
your
future
self
but
yeah.
So
there's
there's
some
opinions
on
that.
Okay,
moving
on
actually
Jane's
ass,
how
stable
is
kubernetes
for
stateful
applications?
How
much
support
is
there
for
database
applications?
H
The
majority
of
workloads
that
we
run
on
our
stuff
is
databases.
The
big
thing
is
just
you
got
to
know,
sort
of
the
caveats
going
in
you
know
it's,
it
is
for
things
to
not
shut
down
cleanly.
You
definitely
don't
want
to
use
deployments.
You
want
to
use
stateful
sets,
yeah,
we've
be
hosts,
Postgres,
MySQL
and
I
hate
it,
but
we
do
sort
of
we
have.
We
have
a
little
woman
there.
H
B
Okay
and
as
usual,
I
will,
post
to
Josh
purchases,
now-infamous
blog
post,
it's
like
a
three
part,
blog
post
on
running
databases
in
kubernetes.
That
always
has
a
lot
of
good
information
there
and
there
are
some
discussed
threads
as
well,
where
people
are
constantly
talking
about
how
to
run
databases
on
kubernetes.
So
we
will
toss
those
into
the
show
notes
if
any
of
you
out
there
in
the
community
have
a
handy
I
want
to
whack
that
into
the
slack
Channel
I
would
appreciate
it.
B
D
Yeah
just
a
little,
we
don't
use
it,
but
I
saw
a
really
nice
talk
at
a
loss.
Cube
con
from
the
I
can't
remember
what
company
they
work
for,
but
they
went.
They
were
in
dealing
with
my
sequel
by
their
than
by
themselves.
They
switched
over
to
a
V
test,
which
is
from
the
YouTube
team,
I
think
so
that's
how
Google
uses
what
Google
uses
internally
to
scale
up
their
YouTube
platform.
So
in
terms
of
scale
you
know
it
can
handle
it
and
it
just
kind
of
helps.
I
think
it's.
D
There
just
replicates
the
sharding
down
to
my
sequel
clusters,
so
it
makes
that
a
lot
easier
to
manage
if
you're,
using
by
sequel
clusters.
Otherwise
I
think
that
if
for
postcards,
there's
you
lando
operator
is
quite
good
from
what
I've
heard
so
yeah.
So
it
depends
on
what
database
you're
running
and
then
just
make
sure
you're
using
the
right
tool
and,
like
Bob,
said
to
make
sure
you
know
what
you're
getting
into
yourself
into.
B
All
right
actually
has
a
second
question,
but
we
will
come
back
to
it
in
a
second.
Our
John
asks
we
have
a
requirement
to
run
elastic
search
on,
kubernetes
was
planning
to
deploy
using
a
stateful
set
and
local
persistent
volume,
but
no
persistent
storage,
so
he
doesn't
have
a
persistent
storage
solution.
Yet
so
it's
a
recommended
setup.
We
use
cube
admin
based
cluster
on
VMware.
A
C
Just
gonna
say:
I'm
I,
just
there's
a
blog
post
from
elastic
that
they've
got
elastic
cloud
burning
on
kubernetes,
it's
been
beta,
I,
don't
know
a
ton
about
it,
but
yeah.
We
definitely
follow
their
kind
of
pattern
for
deployment
and
managing
that
and
definitely
be
mindful
of
the
storage
you
needin
and
the
type
of
storage
and
now
speed
out.
That's
so
looks.
B
Looks
like
lots
of
people
have
opinions
on
this
one.
Let's
see
so
our
John
mentions
that
operator
looks
like
the
one
on
operator.
Hub
is
not
yay,
yet
nick
says
they're
a
couple
vs
operators
now
and
I
think
the
official
went
to
a
lot
of
operators
always
seem
to
be
in
the
alpha
status.
I
hear
ya:
let's
see,
there's
a
helm,
a
less
exact
and
max
guy
recommends
Zalando,
who
has
their
elasticsearch
operator
and
then
the
blog
posts
from
my
goostin
says.
B
H
B
B
For
one
more
plus
one
from
someone's
get
some
kind
of
some
kind
of
quorum
on
on
out
of
run
elasticsearch,
but
yeah
that
should
give
you
some
options.
I
think
this
is
something
that
could
be
an
open
question
and
we
should
definitely
keep
an
eye
on
it
report
back
if
you
do
decide
on
one
or
the
other.
Let
us
know
because
I
think
that'll
help
us
at
least
the
next
person
figure
out,
which
one
to
use
moving
on
James
asks.
Does
someone
you
cilium
without
q
proxy
any
feedback
on
this?
B
B
Anybody
else
have
an
opinion
on
this.
One
bob
just
wants
pretty
much
psyllium
everywhere,
all
over
the
world.
Dimitri,
yes,
and
one
more
question
at
I
hope
I'm,
getting
your
name
right,
one
more
question:
if
it's
alright
recently
learned
about
mix-ins,
especially
the
kubernetes
mixing
em.
For
me
since
operator,
I'm
wondering
how
everyone
is
deploying
these
with
helm,
should
I
have
a
skirt
to
build
my
mix-ins,
create
my
templates
and
deploy
with
home,
or
is
there
a
different
approach.
D
B
C
I
found
it
I
would
definitely
start
playing
around
just
with
home
jerk
tuning
options
lightly
and
kind
of
a
playground
cluster
and
get
an
idea
for
what
kind
of
defaults
are
and
how
it
runs
just
from
have
a
normal
deployment
and
then
start
looking
into
do
you
know
you
need
to
write
scripts
or
sans
running
scripts,
tune
things
to
your
liking.
Most
of
the
time,
you
can
tune
a
chart
with
variables
and
options
to
get
it
to
your
liking.
C
If
that
isn't
exactly
the
case,
you
could
always
fork
the
the
chart
over
and
run
it
yourself
for
your
own
needs.
Tuning
templates
and
things
like
that,
so
definitely
play
around.
Definitely
use
the
dash
dash
debug
dash
dash
dry,
run
options
to
help
make
sure
using
the
latest
Elm
as
well.
I,
don't
know
much
about
mixing
Prometheus
operator
is
pretty
straightforward.
It
takes
a
little
bit
of
tuning,
though,
but
other
than
that
you
know.
C
Definitely
after
you
do
something
with
helm
and
very
much
use
your
client
tools
to
check
what's
actually
running
in
a
sense
of
the
objects
that
are
running
pods,
and
you
know,
connections
and
things
like
that.
So
you
know
exactly
if
something
does
happen
or
it's
taking
over
the
cluster
or
you
know
something
isn't
happening
in
a
way
it's
supposed
to
in
here.
You
know
you're
you're,
trying
to
troubleshoot
and
make
it
easier
on
yourself.
So.
B
Mm-Hmm
mix
adding
a
link
to
the
previous
operator,
since
this
hunter
I
believe
incorporates
the
mixings
but
not
sure
how
well
they're
maintained
Bobby
Tables
is
doing
+1
to
customize
actually
reading
back.
That's
what
Nick
says.
We
use
those
mix
things
but
don't
use
helm.
We
actually
use
them
via
customize
and
also
get
sync
which
will
pre
render
the
mixings
and
use
get
sync
to
apply
them
hook
me
up
with
a
link
to
that
Nick
that'd
be
great
yeah.
C
B
Okay,
Bogdan
says
hello
from
Berlin
hello
question:
what
are
the
services
that
you
would
never
keep
in
kubernetes
for
safety
and
disaster
protection,
point
of
view,
I
mean
for
the
moment:
I
use
e
KS
and
AWS
F
in
the
databases
I
use
RDS,
but
what
other
services
do
you
recommend
to
be
outside
of
kubernetes
for
easy
recovery?
I
want
to
add
a
corollary
to
this
question
as
well
is:
do
you
all
figure
out
when
it
makes
sense,
especially
if
you're
in
a
managed
cloud
provider
should
I
just
use
the
manage
database?
B
D
For
us,
it
depends
on
our
DBAs
level
of
how
much
do
they
need
to
support
I.
Think
we're
looking
at
the
hosted
solutions
just
because
it's
offloading
law
out
of
extra
support,
kubernetes
experience
is
still
kind
of
growing,
so
it's
just
familiar
territory,
so
it
depends
on
what
direction
your
organization
wants
to
go
to
and
D
level
of
experience
here.
Devs
have
mm-hmm.
B
C
I
want
to
completely
Mir
that
we
were,
you
know,
startup
life
moving
very
fast
and
a
lot
of
its
just
deferring
extra
extra
load
on
us
and
and
I'm
really
the
only
quarter,
kubernetes
person-
and
so
you
know,
running
something
like
kubernetes-
is
really
exciting
and
you
get
a
lot
of
benefits,
but
there's
also
a
little
bit
more
management
overhead
than
just
deferring
it
to
a
cloud
provider,
especially
these
days.
So
you
know
they
give
you
everything.
C
You
need
the
endpoint
there's
even
TLS
kind
of
built
in
as
at
least
with
RDS
as
well
things
that
you
don't
have
to
worry
about
as
much
so
you
know
we
would
love
to
run
a
lot
more
like
a
good
examples
of
ElastiCache
at
Redis.
We
actually
use
Redis
through
AWS,
but
I'm
very
much
considered.
Just
you
know,
taking
a
weekend
and
trying
to
get
a
really
scalable
a
solution
on
on
grenades.
It
doesn't
seem
too
hard,
but
it
just
doesn't
seem
to
make
sense.
C
B
Aaron
would
like
to
add
that
if
you're
bare-metal,
you
don't
really
have
much
of
a
choice,
nice
Miguel
says
SEF,
don't
ask
me
why,
but
it
does
seem
like
it's
a
good
idea
right.
If
you
have
limited
kubernetes
knowledge,
why
ad
databases
to
your
life,
you
probably
have
your
hands
full
seems
like
a
not
bad
way
to
do
it,
especially
with
the
managed
service
as
being
as
good
as
they
are
lately.
B
B
D
E
From
experience,
anything
that
holds
long-running
connections
can
be
a
real
pain
in
the
arse
on
people
that
is
like
trying
to
just
gracefully
drain.
Those
has
been
a
big
problem
for
some
Royal
Engineers
for
best
part
of
two
years,
so
maybe
reconsider
that
that's
a
good
when
I
hadn't
hadn't
tried
it
and
try
that
one
okay,
most
of
our
products,
use
WebSockets
and
stuff.
So
our
customers
look
like
consistently
connected
to
us.
So
it's
problems
when
we
want
to
roll
clusters
and
upgrade
pods
and
stuff
awesome.
B
And
then
we
have
time
for
three
questions:
Bob
I've
got
the
raffle
at
the
top
of
them.
How
Candide
documents,
if
you
wanna,
snag
us
four
winners.
That
would
be
fantastic.
Well,
I
finished
these
under
all
right,
Dan
ask
questions.
What
are
some
people
are
doing
to
end-to-end
encryption
within
the
cluster
itself?
Is
it
just
using
a
service
smash
like
sto,
or
do
you
have
something
within
the
cluster
like
cert
manager,
issuing
certificates
to
applications.
D
We
just
rolled
up
the
linker
D
to
provide
that
and
it's
working
really
well.
It
doesn't
give
you
the
overhead
that
sto
does
and
it's
quite
easy
to
maintain
so
far
and
that's
the
kind
like
Mario
and
this
whole
kind
of
kubernetes
e
expert
net.
Just
it's
I,
don't
have
to
deal
with
that
extra
headache
and
I
get
some
a
lot
of
those
little
benefits,
which
is
nice,
mm-hmm
heard
good
things
about
certain
manager,
too
so
depends
on
your
use
case.
B
Okay,
so
linker
DRD
have
any
other
recommendations,
feel
free
to
type
them
moving
on
Venkat
second
I'll
ask
questions
for
those
who
are
scared
of
writing.
Llamo
not
me.
Is
there
any
front
end
that
generates
the
mo
for
them?
There's
dry
run
like
you,
cuddle
create
to
play
engine
excess
fish,
I
mean
that's
nice
right
dry
run
then
Yutaka,
oh
yeah,
mo
into
a
file,
but
they
want
to
use
a
form
or
a
GUI
or
something
like
kickstart
GUI
to
generate
a
kickstart
file
from
like
your
Linux
days.
C
Yeah,
this
is
a
really
tricky
one
I'm,
a
ya
mole
engineer.
3
at
this
point,
so
home
has
a
command
called
helm
template,
and
this
is
really
nice.
Maybe
starting
out
when
you
say
you
find
a
chart,
something
that
looks
interesting.
You
can
actually
helm
template
that
the
chart
net
just
kind
of
prints
out
all
of
the
yeah
mo
for
all
the
objects.
So
you
know
the
values,
and
you
know
variables
and
other
things
like
that
they
go
would
go
into
the
chart.
C
It
you
know,
matches
those
into
the
the
template
creates
the
resulting
manifest
and
prints
that
out
for
you
and
that's
at
least
a
great
place
to
start
I.
Think
you're
gonna
have
to
understand
some
Hamill
to
some
point
for
editing
certain
certain
things
are
pulling
pulling
certain
things
out,
but
you
can
definitely
kind
of
get
around.
At
least
at
this
point,
with
all
the
projects
that
are
out
there,
you
can
get
around
having
to
start
from
scratch
for,
for
most
things,
mm-hmm.
B
And
Aaron
once
add
use
an
ID
with
a
kubernetes
plugin
that
you
can
generate
templates
from.
That
seems
like
a
simple
thing
you
could
do.
Yes.
Code
has
one
and
I'm
gonna
whack
in
a
link
there
for
plumie,
which
is
kind
of
like
a
higher
level
SDK
for
your
favorite
language.
That
kind
of
has
like
native
fingers
from
your
thing.
That
might
be
like
way
to
the
other
side
of
what
you're
looking
for,
but
some
really
interesting
work
happening
there
and
I
wanted
to
make
sure
I
mentioned
it.
Any
other
tips.
D
D
Animal
out
of
that
ballerina
also
does
something
similar,
but
I'm
not
sure
if
it
that's
a
very
specific
use
case.
What's
the
ones
I
can
think
of
off
the
top
of
my
head,
it's
just
a
tool
to
kind
of
help.
You
validate
your
animal
if
you're
kind
of
yeah
pointing
things
things
break
instrument,
it
has
cube
Val
and
so
that'll
validate
your
kubernetes
against
the
current
API
standard.
So
it
really
helps
identify
what
you're
doing
wrong
and
tells
you
where
to
fix
it.
Yeah.
B
And
then
let
me
get
the
URL
for
ballerina
and
then
Brendan
burns
I've
seen
him
do
this
talk
like
four
times
and
I
can't
remember
the
name
of
the
subject:
it's
just
hard
to
google
for
because
he's
a
kubernetes
co-founder.
So
all
you
get
his
interviews
when
you
search
for
him,
but
he
has
a
tour
he's
trying
to
solve.
A
lot
of
people
have
tried
to
solve
this.
B
B
B
B
C
Yeah
and
it
also
works
on
Windows,
so
when
you're
on,
you
can
still
spawn
a
big
cluster,
but
it
also
takes
a
config
file
which
you
defined.
How
many
nodes
you
want
other
parameters
which
is
really
cool
Wow,
so
I
definitely
recommend
it
for
testing
kind
of
Multi,
multi,
node,
setups,
so
really
cool
utility,
yeah
I
know
our
developers
actually
really
only
know
docker
super
well,
and
so
we
kind
of
let
them
test
locally,
with
docker,
so
they're.
C
Actually,
our
docker
compose
files
for
their
application,
as
well
as
kubernetes,
manifests
and
whatnot,
and
that
I
think
kind
of
the
demo,
ops
versus
developers
and
we'd.
There's
an
inflection
point
where
the
developers
don't
really
care
as
much
about
the
queries
configurations.
We
try
to
abstract
that
away
in
templates
and
then
they
just
fill
in
the
values
for
what
environment
variables
or
maybe
scaling
parameters,
and
things
like
that
should
be
so
yep.
B
Yep
and
Konstantinos
would
like
to
remind
everyone
that
micro
Kate's
is
also
available.
I
know
people
use
that,
and
both
micro
Kate's
and
kind
have
channels
on
the
kubernetes
slack,
which
is
really
handy.
Whatever
tool
you
use,
those
developers
are
available,
especially
Constantin
bother
him
with
all
your
questions.
He
really
loves
that
okay
we're
running
a
tad
over
and
we
want
to
wrap
this
up.
I'd
like
to
thank
our
panelists.
Thank
everybody
for
joining.
Our
winners
are
Alexander
PMC,
Graf,
Venkat
and
Agustin.
You've
won
the
quest.
B
B
So
who
knows
you
join
I
just
might
be
in
the
mood
to
give
away
more
t-shirts
and
I
am
allocated
by
the
ciouds
yeah.
So
with
that
everyone
please
feel
free
to
hang
out
office
hours
in
the
discussed
forums.
We
will
go
ahead
and
post
in
all
the
show,
notes
and
all
the
links
we
will
be
at
Q,
Khan
and
cognitive
consul.
There
is
no
session
next
month.
B
I
will
make
sure
that
in
our
notes,
I
put
our
little
schedules
and
word
our
talks
are
so
if
you
want
to
come,
say
hi
to
us,
come
everybody
Bhargav,
welcome
back
dawn,
I
hope
you
had
a
good
first
time
and
that
you
want
to
show
up
again
and
with
that
any
last
words,
everybody
audience
any.
Last
any
last
comments
thanks
everybody.
We
really
appreciate
you
coming
and
with
that
you
will
see
you
all
in
descent.
Are
we
going
in
December?