►
From YouTube: Kubernetes Office Hours 20200318 (EU Edition)
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.
For more info: https://github.com/kubernetes/community/blob/master/events/office-hours.md
A
Now
we
are
alive.
How
is
everybody
doing?
Let
us
know
in
chat
how
we're
we're
doing.
We
had
a
quick.
We
had
a
little
technical
issue
today
where
George
was
streaming
to
the
wrong
event,
but
we
will
fix
that
all
right,
we're
all
set.
Welcome
everybody
on
YouTube.
This
is
the
kubernetes
office
hours
hold
on.
Let
me
see
here,
how's
everybody.
Can
you
just
check
the
stream
real,
quick
Jeff
on
an
actual
YouTube.
B
A
Good
good,
all
right,
everybody
welcome
to
the
kubernetes
office
hours
is
the
third
Wednesday
of
every
month.
We
hop
on
with
a
bunch
of
panelists,
and
we
answer
a
bunch
of
kubernetes
questions,
so
we
kind
of
burned
some
through
sometimes
with
some
technical
issues.
Luckily,
that
only
costs
us
five
minutes
so
we're
gonna
do
here
is
we're
gonna,
go
right
into
your
questions.
I'm
gonna
tell
you
how
it
works
and
then
panelists
when
you
answer
your
first
question,
just
give
yourself
a
quick
intro
and
then
we'll
solve
it
that
way,
yeah
all
right.
A
So
before
we
start,
let's
give
you
some
ground
rules.
This
is
a
kubernetes
event,
so
the
code
of
conduct
is
in
effect,
so
please
be
excellent
to
each
other.
That's
in
chat
and
everything
how
we
sounding
Jacob
welcome.
Thank
you.
This
is
also
a
judgment-free
zone.
Everyone
had
to
start
from
somewhere,
so
we
don't
really
have
any
dumb
questions
or
stupid.
A
Questions
so
feel
free
to
ask
whatever
question
you
need
in
the
slack
channel,
that
is
hash
office
hours
on
slack
buckets
that
I
owe,
and
we
will
do
our
best
to
answer
it
any
time
that
we
have
feel
free
to
hang
out
and
chat.
We
like
it
when
people
hang
out
and
have
their
own
discussions
if
you're
in
chat
now
listening.
Let
us
know
where
you
are
in
the
world
which
work
on
that
kind
of
thing
say
hello,
please
be
kind
of
said
that
the
chat
is
being
live-streamed
on
the
Internet.
A
So
while
we
will
do
our
best
to
answer
your
questions,
the
panel
doesn't
have
access
to
your
cluster,
so
anything
that's
live.
Debugging
is
like
off
topic:
we're
not
gonna
SSH
into
your
like
cluster,
or
anything
like
that
and
dig
into
your
like
stuff.
But
what
we
will
do
is
at
least
get
you
going
in
the
right
direction
on
where
you
need
to
go
to
fix
your
thing.
A
That's
why
we
have
you
here
and
you
can
help
us
out
by
adding
your
expertise
into
the
chat,
because
that
gets
put
into
the
video
you
can
help
by
pacing
in
URLs,
so
the
official
Doc's
blogs
or
anything
that
might
be
relevant
to
the
topic
at
hand.
We've
been
doing
these
for
what
almost
two
years
now
and
almost
every
single
session
we've
learned
about
a
new
tool.
A
So
that's
part
of
the
reason
that
we
have
this
a
part
of
the
community
have
so
many
people
out
working
on
this
stuff
that
it's
really
a
great
place
for
you
to
share
your
expertise,
a
warning,
though,
if
you
do
to
get
a
job
I
might
invite
you
to
be
on
the
panel
like
some
of
the
people
that
are
on
here.
So
it's
a
good
way
to
pay
it
forward.
If
you
want,
please
feel
free
to
start
posting
your
questions
in
slack
now
just
do
question
:
blah
something
that
makes
it
obvious.
A
So
we
can
see
it
and
then
we
put
those
into
a
little
hack
and
be
working
document.
And
then
we
ask
all
the
questions
and
then,
at
the
end,
I
take
all
of
the
URLs
all
the
notes,
and
then
we
whack
it.
We
toss
it
on
the
kubernetes
forums.
So
you
have
a
link
in
the
reference
to
all
the
things
that
we
have
been
talking
about
today.
You
can
also
help
us
up
by
tweeting
spreading
the
word
and
paying
it
forward.
A
Jeff
is
gonna,
give
you
a
tweet
that
he's
gonna
post
in
chat
now
for
you
to
retweet,
because
I
was
streaming
to
her
own
URL,
so
you
can
really
help
us
out
by
retweeting
we're
just
about
to
do
it
or
letting
a
friend,
know
or
whatnot.
This
panel
is
also
made
entirely
of
volunteers.
So
if
you
want
to
rotate
in,
let
us
know
we
kind
of
have
like
a
cadre
of
people
that
are
available
in
a
week
before
I,
say:
hey
who's
around
and
we
figured
out.
A
We
rotate
in
the
commitment
is
about
one
hour
a
month,
but
sometimes
we
have
too
many
people
or
whatever.
So
it's
just
a
cool
little
rotation
and
lastly,
I'd
like
to
thank
all
the
organizations
for
allowing
their
people
to
hang
out
with
the
community
and
help
out.
So
thanks
to
VMware,
Red,
Hat,
Spectrum
Giants,
warm
Google,
Microsoft,
stock,
X
UW
and
whatever
your
affiliation
is
those
of
you
out
there
participating
today.
So
with
that
panel,
you
ready
I,
see
Marcos
here
in
chat,
welcome,
Marco
anybody,
anybody
ready
we're
ready
to
go.
I
am
ready
for.
A
The
first
question
is
from
Simone
who
asks
recently
I've
been
playing
with
a
3-node
RabbitMQ
cluster.
The
clusters
deployed
as
a
stateful
set
with
replicas
equals
three
four
summary:
the
pods
refused
to
join
the
cluster.
This
led
to
some
readiness,
liveliness
and
entry
points
to
fail.
So
what
are
the
pod
refused
to
start?
According
to
the
rabbiting
P
Doc's
are
way
to
solve.
This
is
to
issue
a
reset
command,
but
how
could
I
do
that?
A
If
the
pod
refuses
to
run
I,
also
thought
of
changing
in
the
entry
point
or
the
readiness
pros,
but
this
would
have
impacted
the
other
pods
since
they
are
replicas.
Removing
the
readiness
probe
could
prevent
them
to
start
in
the
right
order.
For
example,
I
ended
up
resetting
the
RabbitMQ
cluster
in
a
somewhat
less
Orthodox
way,
but
this
is
not
something
I
would
have
done
in
production.
A
Environment
I
wondered
if
I
could
have
recovered
it
in
some
other
way,
possibly
less
application-specific
like,
for
example,
if
I
could
get
a
command
to
run
in
a
pod
that
has
the
problems
on
starting
or
if
I
could
override
the
readiness,
/,
liveliness,
/
entry
point
for
specific
replicas
or
something
else
entirely
panel.
What
do
we
think
I.
A
C
F
Guess
I
think
it's
like
super
specific
to
RabbitMQ,
because
somehow
one
node
couldn't
actually
join
the
cluster
and
if
you
would
like
try
to
modify
like
readiness,
probes
or
liveness
probes,
they
actually
call
like
actually
execute
that
recipe.
Oh
man
I
think
it's
going
to
like
reset
your
quest,
so
you
might
be
like
losing
data
or
something
so
I
would
definitely
like
check
out
rabbitmq
cluster
try
to
figure
out
why
that
single
pod
didn't
actually
didn't
manage
to
join
your
cluster,
because
it's
like
it's
super
specific
to
this
application.
Yeah.
D
So
most
probably
the
focus
should
be
not
on
the
pot
that
is
not
working
but
on
the
ones
that
are
and
are
basically
not
letting
the
other
one
in
so
most
probably
the
commands
or
the
bug
fixing
needs
to
be
done
on
the
on
the
existing
cluster
there,
but
I'm
also
I'm
very
almost
not
worked
with
very
private
MQ.
So
no.
A
A
Marco's
says:
could
you
exact
into
one
of
the
other
two
pods
and
inspect
/
modify
the
state
interesting
Ming
Ming.
Welcome,
yes,
bringing
your
own
stackoverflow
question
is
definitely
on
topic,
we'll
put
that
in
the
queue
and
get
to
it.
Thanks
for
dropping
by
it
was
something
like
the
note
things
to
be
joined
of
the
cluster,
but
the
other
nodes
disagree.
The
official
way
to
recover
from
this
is
launch
rabbit
and
Q
control
reset.
But
how
is
this
like
a
rabbit,
quorum
issue?
Yes,.
D
C
C
E
Was
just
going
to
ask
what
the
the
rabbitmq
community
had
to
say
about
this
issue?
I'm
Sur
sure
it's
probably
not
the
first
time
someone's
come
across
that.
So
if
their
solution
is
run,
that
RabbitMQ
reset
or
is
there
another
way
like
spin
up
a
pogrom
that
reset
or
there's
something
I've
been
fortunate
not
played
around
with
me
up
at
MQ?
But
yes,
trying
to
take
many
other
debugging,
workarounds
or
and.
B
B
Yeah
I
was
gonna
say
he
really
needs
to
engage
with
the
community
because
I
guarantee
you
other
people
have
written
to
this
gotten
frustrated
and
just
like
use
the
happy
handed
solution,
but
maybe
there's
a
better
solution
out
there.
So
we
can
put
in
the
docs
kind
of
a
warning
for
people
that
are
deploying
so.
D
Maybe
one
thing
to
to
kind
of
circumvent
the
issue
in
the
future
would
be
to
give
yourself
a
bit
more
time
and
maybe
have
less
aggressive
probes.
So
you
have
a
bit
of
time
to
run
a
debug
container
against
that
and
run
the
right
reset
commands
before
it's
just
get
getting
restarted
all
the
time
some
readiness
rate
of
60
seconds
or
so
to
give
yourself
the
time
to
to
run
that
reset
command
against
it
would
just
mean
that
the
containers
are
starting
a
bit
slower.
A
E
A
A
I'm
kind
of
interested
to
seeing
I
mean
obviously
nuking
it
from
orbit
is
an
option,
but
you
know-
and
that
did
fix
the
problem,
but
hopefully
there's
like
a
cleaner,
more
recommended
way,
but
I
would
be
very
interested
to
know
and
if
you
do
find
out,
if
you
could,
let
us
know
that
would
be
fantastic
and
how
about
so
I
think
we're
all
tapped
out
any
any
any
any
other
creative
solutions
that
we
can
think
of
before
we
move
on
all
right
thank
Simone.
What
we
do
here
is
when
we
read
your
questions.
A
We
you
will
be
entered
in
a
teacher
giveaway
contest,
so
I
will
be
giving
away
two
shirts
at
the
end
of
the
session.
So
stick
around
everybody
you
have
to
be.
You
have
to
wait
till
the
end,
though,
for
the
raffle
so
moving
on.
We
have
this
question
from
Jay
Mohr
car,
which
is
actually
kind
of
long
because
they
posted
he
being
the
API
and
got
a
bunch
of
information.
So
hey
anyone
had
used
telegraph
with
an
F
kubernetes
input
plug-in
inside
of
a
pod.
This
call
to
the
kubernetes
API
URL.
A
You
know
I'm
wondering
if
I'm
be
just
easier
to
repost
this
one
here.
Has
anyone
looked
at
this
one?
Yet
I.
A
So
here's
a
question:
sorry
that
that's
a
long
one
and
I
didn't
you
know:
I
didn't
want
to
I,
didn't
want
to
read
the
outputs
of
a
curl
command
out
on
the
podcast,
so
he's
basically
getting
a
404
error
and
then
he's
filed
an
issue
there,
which
is
here
in
github.
I,
will
give
the
panel
a
few
minutes
to
digest
that.
A
Does
anyone
want
to
take
this
one
and
read
one
so
I
could
do
the
next
one,
because
we
have
enough
panelists
I
think
we
could
split
up
yeah.
Why
elite
it's
it's
pretty
loud
who's
gonna
read
this
one
and
then
we
can
move
on
to
this
nutty
survey.
One
anyone
I
got.
B
B
A
My
screen,
okay,
perfect
all
right,
so
just
gonna
investigate
that
one
and
get
back
to
us.
We're
moving
on
a
question
from
Prashant
Rathore
says:
I'm,
an
external
nettie
server,
not
in
a
cluster.
How
can
I
share
the
same
WebSocket
connection
from
my
pods
clients
to
this
external
server?
I
mean
multiple
pods,
so
that
message
is
sent
by
the
service
can
be
distributed,
but
I
don't
want
the
service
to
have
separate
connection
for
each
pod
thoughts.
D
G
E
C
Our
Prashant
here
I
mean
he
can
for
me.
He
can
extract
it
away
and
put
like
another
service
that
connects
to
the
external
WebSocket
connection,
but
that
will
be
like
a
lot
more
work
instead
of
just
doing
multiple
WebSocket
connections,
so
it
would
be
like
the
second
option.
I
would
go
to
if
multiple
WebSocket
connections
are
not
possible.
I
would
have
one
services
that
kind
of
manages
the
WebSocket
connection
and
T.
They
like
redirects
the
traffic
to
the
other
services.
D
Yeah
and
maybe
if
we
knew
more
about
the
use
case,
it
would
be,
we
would
be
able
to
give
some
some
hints
at
other
solutions,
because
I'm
not
sure
like
from
this
question
what
the
use
case
would
be,
and
if
that
is
like
a
good
way
of
doing
this,
the
more
ugly
workaround.
But
there
might
be
a
good
reason
to
do
an
ugly
workaround
because
it
might
be
legacy
software
and
you
might
not
be
able
to
change
it.
D
B
So
they
actually
answered
it
them
so
well
kind
of
answered
it
themselves
with
some
help
from
Rocco.
Actually
there
were
a
couple
problems,
one
they
didn't
have
our
back
set
up
with
this
helmet.
It
was
a
helm
chart
that
was
a
in
song
Telegraph
and
the
other
thing
was
it
was
pointed
out.
You
need
to
use
the
downward
API
proxy
rather
than
just
trying
to
hit
kubernetes.
So,
okay.
E
B
A
A
Great
thanks
for
digging
in
that
shout
out
to
rock
owed
great
guy.
Next
question
is
from
Felix,
and
this
is
another
long
one,
so
hi
all
waiting
the
next
office
hours.
Meanwhile,
I
have
a
problem
with
a
volume
in
a
security
context
that
the
pods
are
running
with
I
have
a
security
context
for
our
stateful
set.
That
runs
the
containers
as
user
user
one,
but
a
volume
config
map
are
mounted
as
root.
I,
don't
understand!
Why
can
someone
explain
to
me
what
happened
here
and
then
they
pasted
their
bla
calm
calm.
D
A
A
This
is
a
very
Oso
Alexander.
Welcome,
says
this
is
very
long-running
issue
on
github,
only
groups
are
defined,
not
user
and
then
Joe
speed,
hey
Joe
says
it's
a
limitation.
How
the
mount
system
works.
I've
looked
into
this
before
about
six
months
ago.
Unless
anything
has
changed,
you
can
only
mount
config
Maps
as
root,
and
then
alexander
confirms
the
volume
mount
user
is
always
route.
A
Guess
we've
seen
uglier
alright.
Well,
unfortunately,
that's
probably
not
the
answer
you
were
looking
for,
but
it
seems
to
be
the
right
one.
It
looks
like
people
are
still
typing
in
here,
though.
So,
let's
see
what
happens
here,
Alexander
has
length
to
the
issue
that
we
are
talking
about.
It
is
26,
30
volumes
are
created
and
container
with
ownership
to
strict
permissions.
I'll
make
sure
that
makes
it
to
the
notes
and
there
is
a
related
discussion
and
everything
Joe
mentions
this
really
annoyed
me
when
trying
to
get
proud
to
run
as
non-root.
B
A
Joel
agrees
and
Pierre's
dropped
in
a
link
there
to
the
relevant
documentation,
so
that
should
help
you
out
all
right.
Moving
on
those
of
you
listening
feel
free
to
keep
asking
questions
just
start
to
type
them
in
the
channel,
we're
almost
caught
up
in
the
queue
here.
If
at
some
point
we
miss
your
question
or
it's
not
obvious,
just
feel
free
to
ping
me
and
just
repost
it
in
but
put
it
in
the
channel.
Don't
PM
me
because
I
have
so
many
windows.
Open
I
have
no
idea.
A
What's
going
on
so
moving
on
emmanuel
asks
since
kubernetes
1.18
was
released,
I
encountered
a
lot
of
problems
with
the
mental
load.
Balancer
external
IP,
all
the
services
of
tut
load,
balancer,
don't
get
the
address.
It
remains
in
a
pending
state.
I've
checked
the
metal
lb
logs
and
I
didn't
find
anything
apparently
meaningful
for
the
problem.
I'm
working
on
a
virtual
environment
deployed
automatically
so
I'm
100%
sure
that
nothing
has
changed
in
the
configs.
That's
nice.
The
only
things
changed
is
the
kubernetes
version.
A
I've
tried
to
deploy
the
cluster
from
scratch
many
times,
but
the
result
was
the
same.
I've
tried
a
kubernetes
manual
installation
too,
but
the
result
is
the
same.
I've
used
this
configuration
for
three
months
without
any
problems,
so
I
know
that
before
cube
1.18
all
worked
as
expected.
Has
anyone
experienced
the
similar
issue.
A
B
A
F
One
idea
that
the
heaviest
may
be
like
a
long
shot,
but
maybe
you
yet
like
in
metal,
will
be
defined
on
address,
pool
which,
basically,
what
address
is
to
use
for
external
apiece?
Maybe
you
are
like
out
of
a
piece
or
something
because
it
seems
like
a
metal
will
be
issue
more
than
anything
else.
So
I
would
definitely
check.
What's
your
pool
range
and
what
actually
peas
are
used.
A
B
Say
or.
B
F
A
Mean
I
mean
that
could
be
added,
I
mean
I,
don't
know!
Oh
it
manuals
here.
Awesome
says
the
change.
I
see
is
coupon,
daddy
T's,
the
Presidents
of
manage
fields,
I've,
never
seen
or
used
them
before
I'm
testing
in
a
/
24
network
that
is
free
for
90%
of
the
IPS,
so
he's
pretty
sure
he's
not
running
out
of
IPs
but
they're
still
typing.
Let's
see
I'm
using
a
slash
20
as
IP
pool
reserved
to
me.
A
D
B
D
F
C
A
I
was
I
was
gonna,
say
if
men
I'll
be
was
just
outright
broken
one
day
in
18,
I
think
you'd
see
a
bug
like
an
issue
almost
immediately
right,
so
there's
got
to
be
something
specific.
He
knows.
I
didn't
check
the
cube
API.
If
you
think
that
could
help
I
will
do
it
ASAP
so
question
here
for
the
panelists
actually.
Is
it
always
just
a
good
idea
to
check
to
keep
I
know?
This
seems
like
advanced
common
sense,
but
is
does
everything
well
everything
has
to
go
through
there
right.
E
E
B
Be
some
things
in
like
an
area
you
think
is
unrelated.
That
will
have
breadcrumbs
I
think
it's
it's
best
just
to
capture
all
of
the
logs
and
all
of
our
CI
jobs.
We
literally
capture
everything
about
the
state,
all
of
the
logs
everything
that
was
running
literally
everything
just
so
we
can
debug
because
yeah.
E
A
Like
rock
the
next
Ravi's
next
question,
which
we'll
get
to
here
in
a
minute
right,
the
log
is
pretty
explicit
there
about
a
TLS
bad
certificate
right.
So
all
right
so
in
manual,
it
sounds
like
check
out
the
cube
API
logs.
If
I
don't
know,
if
you
have
access
to
the
cluster
now
you
know
check
it
out
and
then
maybe
report
back.
But
that
sounds
like
a
good
place
to
start
any
other
tips
here
that
we
can
figure
out
as
far
as
mental
I'll
be
I.
B
F
So
it
feels
like
somewhere
in
like
another,
will
be
control
loop.
It's
not
doing
what
it's
supposed
to
be
doing
so
yeah
like
if
you
can,
like
turn
on
the
blocks
on
metal,
will
be
or
astrays
it.
Yeah
I
definitely
look
at
cube,
API
logs
for
like
any
like
or
forbidden
or
like
some,
some
kind
of
like
I.
Don't
know
a
back
issues
for
a
metal
will
be
right.
A
Okay,
well,
the
manual.
Let
us
know
how
that
goes.
Hopefully,
that'll
give
you
plenty
of
places
to
start
looking.
Okay
thanks.
Everyone
for
joining,
we
have
about
20
minutes,
left
welcome
to
the
kubernetes
office
hours,
just
hang
out
and
hash
office.
Our
zine
slack
that
case
at
I/o.
Ask
your
question:
will
queue
it
up
and
answer
aim
the
best
way
possible
and
then,
at
the
end,
we'll
give
away
a
bunch
of
t-shirts
and
everyone
will
have
a
good
time.
A
The
next
question
is
from
mingming,
which
is
a
link
to
a
stack
overflow
question
which
I'm
gonna
repost
here
always
appreciate
it.
When
people
bring
their
own
questions
with
the
detail
already
formatted,
so
I
will
go
ahead
and
link
that
in
there
and
says
I
created
a
CRD
like
this
and
then
feel
free
read
the
link
there
and
I
noticed
that
I
reuse,
pod
spec
from
the
core
API
group
in
mycr
D,
to
avoid
user
applied
invalid
yamo
files,
I
decided
to
add
validation,
logic
in
mycr
D
controller
for
simple
fields
like
name.
B
If
you're
trying
to
unmarshal
EML
into
that,
it
should
automatically
validate
because
it's
trying
to
validate
against
the
pods
back
type
and
like
in
in
my
own
development
of
controllers,
one
of
the
things
that
I've
done
if
I'm
dealing
with
like
so
one
of
the
things
that
I'm
working
on
it
receives
a
string
that
is
gamal
and
I,
just
try
and
unmarshal
that
into
the
correct
type
and
that
validates
it
or
not.
Mm-Hm
now
laid
around
in
the
Stack
Overflow.
A
D
Forward
because
it's
validation
mentioned
there,
maybe
what
they
want
is
also
to
get
like
actual
CID
validation
on
the
API,
so
like
open,
API,
schema
validation,
which
is
basically
I.
Think
what
you
need
to
do
there
is
you
need
to
copy
over
what
what
is
in
the
schema
for
a
part
into
yours.
Cid
schema
to
get
it
kind
of
the
whole,
because
there
is
a
mention
of
reg
X,
which
is
usually
used
in
a
schema
validation.
B
You
can
define
it
in
the
the
crap
the
labels
and
comments
to
the
right
of
it
and
above
the
types,
the
other
thing
is
if
they
are
looking
for
more
than
just
this
is
valid.
This
is
not
valid
if
they're
looking
for
more
than
just
that
boolean
they
want
specifics
on
what's
invalid,
then
things
like
the
validation
package
in
kubernetes,
like
the
validate
pod
spec,
will
actually
do
like
down
to
the
field
level,
validation.
A
All
right
any
other
comments
here,
all
right.
This
Raavi
welcome.
This
is
I'm
getting
the
following
errors
from
API
server
logs.
Does
anyone
have
any
idea
what
these
errors
are
and
how
to
fix
it?
Prometheus
alerts,
API
server
errors,
hi.
If
you
actually
scroll
up
just
a
tad
a
little
bit
there
panel
you'll
see
his
post.
This
is
the
one
I
had
mentioned
earlier
before,
where
he's
clearly
getting
a
TLS
handshake
error
for
a
bad
certificate.
Thoughts
on
this
one.
A
E
D
G
D
D
A
Right,
like
an
internal
IP,
any
other
tips
on
this
one,
here's
a
question
from
Zane:
this
is
an
interesting
one.
Why
can't
cube
admins
certificates
be
valid
for
10
or
100
years?
It's
too
painful.
When
we
deliver
kubernetes
to
the
customers,
production
environment,
we
need
to
manually,
compile
the
cube
admin
source
or
manually,
create
a
large
number
of
certificates.
Can't
you
just
give
us
an
optional
parameter
or
variable
to
flexibly
configure
this
option.
Anyone
have
opinions
here,
I
think.
A
D
D
B
B
A
D
Mrs.
security
standpoint
you
should
definitely
rotate
certificates
and
I
would
even
recommend
more
often
than
a
year,
because
kubernetes
does
not
support
revocation
of
certificates.
So
if
one
of
your
certificates
gets
out,
people
have
access
to
your
cluster
for
a
full
year,
all
right
or
hundred
years.
If
you
want
to
think
about
200
years
yeah
and
maybe
not
gonna
fly
with
yes,
it's
it's.
A
A
The
reason
I
say
that,
especially
with
things
like
cube
admin
is
they
try
to
cover
so
many
use
cases
that
sometimes,
without
the
background
on
why
a
default
was
chosen,
it
could
be
difficult
to
figure
out
why
they
would
choose
that.
So
I
will
find
out
the
link
to
cube
and
office
hours.
Here
we
go.
It's
ones.
A
C
A
So
saying
check
this
out
here:
cube
admins
part
of
SiC
cluster
lifecycle.
If
you
click
here
and
you
scroll
down
about
halfway,
there's
a
cube
admin
office
hours.
Of
course,
there's
a
cube
admin
slack
channel
as
well,
if
you
just
want
to
ping
them
on
there,
so
that's
Wednesdays
at
9:00
a.m.
Pt
weekly,
so
that
is
that
is
today
in
two
hours
right,
because
that
conflicts
with
our
office
hours
that
we're
doing
again
in
two
hours.
A
A
Alright,
we've
got
about
ten
minutes
left
so
time
for
two
or
three
more
questions,
so
keep
lining
them
up
next
question
from
Sammy.
He
says:
I
have
a
scenario
where
I
see.
Sometimes
the
end
points
do
not
get
created.
Even
when
the
pot
is
running
service
is
there
and
labels
are
matching.
I
did
see
this
blog
from
Dollar
Shave
Club,
but
it
didn't
help
finally
ended
up.
Writing
my
own
Python
script.
To
handle
this
scenario
where,
when
you
delete
and
recreate
the
service,
it
immediately
gets
created.
A
B
E
B
F
I'm,
looking
at
that
blog
which
person
referenced
and
it
looks
like
it
was
like
qbp
I
thought
that
link
issue,
so
basically,
after
controller
manager,
sort
of
sends
a
lot
of
requests.
It
just
sort
of
throttles
itself
and
the
way
we
fixed
it
is
they
basically
tuned
up,
cube
API,
QPS,
flag
and
cube
API
boost
flag,
which
seems
like
a
reasonable
fix.
D
Yeah
we'll
make
sure
it's
it's
actually
the
same
problem
like
if
you're,
if
you're,
seeing
the
same
issues
in
the
controller
manager
that
it's
getting
getting
throttled
I'm,
not
sure.
If
you
see
that
on
the
controller
manager
logs
or
the
API
logs,
then
then
I
would
definitely
adjust
that.
Otherwise
you
would
definitely
need
to
go
deeper
and
see
where
it's
actually
coming
from,
because
yeah
the
Python
script
might
help.
A
A
We
are
running
out
of
time
here,
Jeff's
about
ready
to
fire
up
the
old
Dungeons
&
Dragons
dice
and
give
away
two
t-shirts
here
and
the
last
question.
Well,
here
we
go
thanks.
Pierre
I
was
waiting
for
that
one
Rodrigo,
Villa
Blancas,
welcome,
says
I'm,
just
wondering
why
kubernetes
is
using
making
basil
I
made
a
simple
PR
for
an
error.
Handling
and
I
had
several
CI
stages
fail
because
of
basil.
We're.
A
A
A
Anytime,
you
link
on
a
github
issue
and
it
has
like
the
fold
in
it
that
says:
load
more
items.
You
know
it's
gonna
be
a
good
one.
So,
hopefully
that
will
answer
your
question
there.
It
looks
like
Alexander's
typing
anybody
else.
We
have
time
for
one
more
question
before
Jeff
gives
away
the
t-shirts
will
will
I'll
go
ahead
and
start
out
trolling
here.
A
If
someone
has
a
really
good
question,
we
could
do
it
so
with
that
Jeff
is
gonna,
pick
two
people,
here's
the
way
it
works
when
you
come
to
office
hours
and
we
address
one
of
your
questions
on
air.
We
give
away
two
t-shirts.
So
these
are
the
dope
kubernetes
t-shirts
from
the
CN
CF
store.
The
way
it
works
is
we'll
announce
the
two
winners
and
then
I
will
PM
you
a
code
and
you
go
to
store
dot.
C
and
C
F
dot
IO
apply
the
code
for
your
kubernetes
shirt
and
you
get
it
for
free.
A
Also.
If
anybody
wonders
where
all
these
awesome
people
are
getting
all
these
cool,
kubernetes
shirts,
you
can
find
those
always
at
store.
That's
CFI,
oh
and
as
always
a
special
thanks
to
the
CNC
F
for
sponsoring
the
t-shirt,
giveaway
so
Jeff,
who
are
two
winners.
Today
we
got
a
manual
and
Ravi,
alright,
so
Manuel
and
Ravi.
If
you're
around,
you
have
won
a
kubernetes
t-shirt,
I
will
PM
you
right
after
this
show
and
as
I
organize
the
notes
and
everything
last
call
for
URLs
or
anything.
A
A
So
one
session
is
useful
for
Europeans
and
once
for
the
west
coast,
and
those
of
us
stuck
in
the
middle,
both
of
them
are
useful
for
us
and,
as
always
everything
we
do
here
is
open-source
and
public
and
written
down.
So
if
you
have
interest
and
running
your
own
office
hours
in
your
time
zone
or
want
to
help
volunteer
any
of
these
things,
we
would
always
appreciate
that
help.
Something
we
are
working
on
this
cycle
is
to
bring
more
topic.
A
E
E
C
A
C
A
B
D
A
E
A
Dope
awesome,
alright
and
with
that
we're
gonna
sign
off
we'll
be
back
in
two
hours
with
our
West
Coast
Edition,
once
I
fix
the
YouTube
URLs,
it's
gonna
be
a
hot
mess
for
me,
but
it
will
be
fine
thanks.
Everybody
for
listening
in
panels
stick
around
for
a
second
and
for
everybody
else,
we'll
see
everybody
next
month,
always
a
third
Wednesday
of
every
month.
Thanks
for
participating
and
enjoy
and
stay
safe
out
there.
Everyone.