►
From YouTube: Kubernetes SIG Testing - 2021-04-06
Description
A
Sure
the
full
script
but
hey
everyone,
welcome
to
sig
testing
bi-weekly
meeting
the
gist
is
respect
each
other
and
be
good
people
and
respect
the
kubernetes
code
of
conduct.
A
A
So
if
we
want
to
discuss
alvaro's
discussion
or
is
yours
alvaro.
B
Yes,
it's
also
just
a
small
point,
so
we
agreed
on
that.
We
are
going
to
enable
the
cherry
picker
plug-in
for
power,
at
least
for
sick
gorilla,
six
speakers
that
want
it.
However,
the
thing
needs
a
github
user,
and
this
github
user
can
not
be
the
standard
ci
robot,
because
the
ci
robot
doesn't
listen
to
itself
for
safety
reasons,
so
my
question
would
be:
do
we
either
have
another
github
user?
Already
we
can
use
or
if
not,
what
would
be
the
process
to
get
one.
A
That
sounds
like
a
good
errand
question.
Unfortunately,
I
feel
like
we
do,
but
I
think
aaron's
the
one
who
would
know
for
sure.
B
A
C
C
Yes,
github,
updated.
All
of
the
token
format
it's
nice
to
swap
is
not
necessary
at
this
point,
but
still
it's
a
pain.
If
github
ever
announced
a
breaking
change,
we're
going
to
know
where
all
of
our
bots
live
and
how
to
update
them
and
where
the
tokens
are
stored,
I
don't
keep
track
of
them.
I
don't
really
know
where
the
tokens
are.
Even
if
I'm
not
that
lazy,
I
want
to
swap
them.
I
don't
really
know
how
and
where.
D
Do
we
have
a
current
process
for
creating
robots
that
are
managed
by
the
community?
I
believe
so
am
I
hearing
about.
C
B
D
C
I
think
the
other
one
is
pretty
lightly
used.
If
you
ever
want
you
can
just.
I
think
you
can
reduce
it.
If
you
want
to.
E
Well
about
the,
but
I
may
have
an
answer
for
that.
So
basically
all
the
bots,
the
old,
but
I
still
imagine
it
by
say,
testing
and
the
password
for
per
second
are
still
somewhere
inside
google.
So
I
asked
this
question
last
week
during
the
games
information
at
darren
entirely
at
google
to
get
the
password
for
a,
but
they
don't
use
anymore
for
sick
testing,
but
it
takes
some
time.
So
basically,
what
I
did
is
just
create
a
new
bot
for
the
pro
pro
station
instance.
B
E
E
B
Yeah,
you
can
probably
short
circuit
this,
because
I
don't
just
join
aaron.
You
might
be
the
person
who
can
answer
the
question
of.
I
need
a
github
bot
for
the
cherry,
picker
plugin,
which
cannot
be
the
ci
robot
itself,
because
paul
refuses
to
listen
to
itself.
B
What's
what's
the
best
way
to
do
that
or
do
we
have
an
existing
one?
Maybe.
F
F
I
can't
seem
to
find
the
login
and
password
for
this
account
so
like
you
could
get
somebody
from
testing
for
on
call
like
chow
or
sean
to
like
get
the
token
out
of
the
test,
infra
cluster
or
like
you
could
use
that
token,
because
it's
in.
A
B
F
Yeah,
so
you
can
try
that,
and
I
can
work
to
see
if
I
can
get
the
credentials
for
that
offline.
If
that
doesn't
work,
the
the
thing
that
arno
and
I
are
trying
to
work
on
for
the
in
for
prow
instance
we're
gonna.
We
set
up
a
google
group
under
kubernetes.io.
F
We're
using
that
email
address.
Is
the
email
address
for
the
bond
account
so
that
multiple
people
will
have
access
to
it,
so
my
vote
would
be
try.
Try
using
the
existing
it's
either
called
kate's
merge,
robot
or
kate's
github
robot.
The
the
secret
in
the
proud
cluster
test
and
front
call
would
be
able
to
tell
you
and
if
that
doesn't
work,
go
back
to
me,
you
mean.
D
B
F
A
A
Docs
on
how
to
set
up
a
robot
or
how
to
get
access
to
the
robots,
I
was
looking
for
them,
but
I
couldn't
find
anything.
F
I'm
gonna
defer
to
testing
for
on
call
again.
There
are
instructions
in
the
proud
setup
instructions
for
like
how
you
could
go
about
creating
your
own
github
robot
account,
but
we
don't
have
a
doc
as
a
kubernetes
community
on
like
what
our
best
practices
and
policies
and
procedures
are
for
github
bot
accounts.
F
I
think
we
could
use
some
help
on
that.
Like
I
said,
I
think
arno
arno
and
I
had
a
chat
with
like
the
github
management
folks
and
their
their
request
was
like
use
a
kubernetes
dot,
io
group
to
as
the
login
for
the
account,
and
then
we
could
like
store
the
password
in
google
secret
manager,
where
we
already
have
like
get
ops,
driven,
config
changes
for
who
is
allowed
to
access
that.
F
There
and
ideally
maybe
that's
a
good
question
for
the
cherry
creek
robot.
Do
you
need,
do
you
need
like
more
owner
privileges,
or
could
you
get
away
with
less.
B
It
literally
doesn't
need
anything
at
all
it
just
like
creates
a
fork
on
under
its
own
user,
pushes
to
a
branch.
There
creates
a
pre-request,
so
it
doesn't
need
anything
at
all.
Yeah.
It
could
be
fade
about.
That's
true,
sorry,
I'm
perfectly
cool
with
feta,
but
too
it
just
might
confuse.
Maybe,
but
I
really
don't
care.
C
I
think
it's
casey
a
robot.
That's
weird!.
F
No,
I
don't
have
coop
ctl
on
this
laptop.
Apparently
that's
interesting.
Does
anybody
have
access
to
the
cluster?
I
mean
we
could
hash
this
out
real,
quick,
see
what
secrets
are
available
on
the
custom.
Requester,
maybe
figure
out
what
tokens
are
available
to.
F
C
F
C
C
E
F
Looking
sorry,
I
don't
feel
like
this
is
maybe
the
most
productive
use
ever.
Okay,
there's
something
called
triage
bot
token
in
the
test.
Pods
test,
cod's
namespace.
If
you
could.
F
Anyway,
I
feel
I
feel
like
I
can
work
offline
with
the
testing
for
call
on
call
folks
to
see
if
we
can
find
the
login
and
password
or
go
through
the
recovery
flow
for
the
kate's
github
robot
in
terms
of
names.
That
would
be
my
preferred
account
if
not
I'll,
take
a
look
at
maybe
reusing
the
the
I'll.
Take
a
look
at
just
setting
up
a
new
account
for
you.
F
Why
not
something
to
be
aware
of
I'll
check
in
with
the
other
github
admins?
I
believe
that,
like
github
is
very
aware
of
prow
and
the
bot
accounts
that
it
uses
and
has
lifted
like
internal
bad
behavior
detection.
I
guess
so
like
the
ktci
robot
account
gets
away
with
a
lot
more
than
like
any
old
bot
account.
I
went
and
set
up
as
a
user.
F
So
as
long
as
you
don't
think,
the
cherry
pick
plug-in
is
gonna.
Do
anything
too
crazy.
We
can
probably
set
up
any
old
bot
account.
F
Yeah,
let
me
let
me
ping
you
in
slack,
once
I've
got
either
the
login
credentials
for
the
github,
github
account
or
I'll
go
ahead
and
create
a
new
one
awesome.
Thank
you.
A
Yeah,
I
was
going
to
say
if
I'm
maintaining
the
reigns,
but
if
nobody
else
has
any
other
points,
I
guess
we
can
call
it
really.
C
A
C
C
So
I
think
it's
as
far
as
I
can
tell
it's
pretty
promising,
and
it
certainly
can
can
save
the
testing
for
on
call
several
hours
of
rotating
secret
for
clients
and
clients
are
also
pretty
happy
to
self-service.
The
secret
rotation
so
seems
like
it's
a
win-win
situation,
and
I
have
a
pr
enabling
that
for
build
cluster,
and
if
anyone
has
a
chance,
please
take
a
look.
F
C
Awesome,
I
also
figured
it's
pretty
awesome
that
you
can
also
help
us
on
some
other
cases
like
this
morning.
I
realized
it
can
also
help
us
automatically
deploy
the
monitoring
stack
because
it
used
to
be.
We
still
have
a
last
step
of
manual,
replacing
playing
secret
secret
value
in
yaml
file
before
deploy,
and
it
can
actually
be
solved
by
the
entire
secret
management
stuff.
C
F
F
A
F
We
want
our
plans
to
be
for
the
v122
kubernetes
release
cycle.
I
don't
know
about
y'all,
but
I
have
like
five
or
six
different
ways
of
attempting
to
plan
a
quarter
and
I'm
trying
to
reconcile
them
all
so,
maybe
I'll
I'll
do
that
next.
F
C
F
F
I
was
thinking
about
maybe
leveraging
the
same
sort
of
stuff
to
set
up
some
auto
deploy
jobs
for
kate's
in
for
stuff,
mostly
just
the
kate's
in
for
a
pro
probability
clusters,
but
it
could
be.
F
We
get
to
a
point
where
some
of
the
other
applications,
like
slack
and
triage
party,
could
also
be
just
auto
merged
and
auto
declared.
C
F
Okay,
I
will
get
to
it
as
I
am
able.