►
From YouTube: SIG-Testing Weekly Meeting for 20221018
Description
SIG-Testing Weekly Meeting for 20221018
A
B
A
Do
anything
particular
about
it?
Yeah.
B
I
I
I
I
hopped
on
earlier
on
and
I
think
it
was
recording
because
the
UTC
UTC
plus
one
business.
C
B
I
I
preferred
I
proffered
a
demo
of
some
work
that
I've
been
doing
to
explain
the
mysterious
appearance
of
a
weird
pull
request
that
I've
submitted
to
the
GitHub
client,
so
so
just
to
socialize
the
work
that
I've
done
and
just
explain
that,
where
I'm
at
and
just
give
a
little
bit
of
background,
it
may
not
be
of
huge
interest
to
sick
testing.
But
it
was
just
a
kind
of
a
show
and
tell
really.
D
A
I'm
also
happy
to
Point
those
folks
back
to
these
notes
or
stuff
like
that.
So
yeah.
A
C
Hi,
my
name
is
Brady
I
I'm
new
I
I
want
to
try
to
get
involved.
However,
I
can
I
work
at
Red,
Hat
I'm,
a
quality
engineer
at
red
hat,
but
this
isn't
really
based
any
work
basis
is
going
to
be
mostly
personal,
stuff,
I
work
on
testing,
openshift
and
rail,
with
different
products
of
of
Red
Hats
so,
and
we're
looking
at
using
prowl
a
lot
more,
so
I
definitely
want
to
learn
more.
A
Yeah
welcome
and
awesome
to
hear
yeah,
as
mentioned
I
think
we
don't
really
have
a
lot
of
craft
folks
in
this
call,
but
I'm
also
on
the
same
team
with
a
bunch
of
those
folks
so
and
they
do
definitely
check
the
slack
Channel.
C
A
All
right
great
to
hear
anybody
else
or
with
that
we
can
probably
turn
it
over
to
Robert.
B
Cool
so
I'll
just
can
I
share
my
screen.
B
B
My
name
is
Robert
kilty
I'm
I'm,
coming
from
Dublin
Ireland
I've
worked
on
Sig
release
for
a
few
releases
back
I
have
recently
been
working
with
II
and
hippie
asked
me
to
look
at
doing
some
work
to
basically
launch
prior
plugins
from
a
GitHub
action.
So
so
so
this
so
This
has.
B
This
has
been
a
a
piece
of
proof
of
concept,
work
that
I've
been
working
on
for
the
past
few
months
and-
and
there
are
some
people
who
are
interested
in
this-
and
this
has
culminated
in
me-
submitting
a
small,
a
small
PR
to
add
a
flag
to
indicate
to
the
GitHub
to
a
GitHub
client.
That's
used
by
proud
to
say,
hi
I'm,
starting
a
GitHub,
client
and
I
am
aware
that
I
am
a
being
launched
from
a
GitHub
action.
B
So
so
that
is
a
bit
weird
for
I
presume
that
the
people
are
testing
for
based
on
comments
that
I've
received.
So
I
just
want
to
describe
the
story
of
how
we
got
here
and
what
the?
What
the
rationale
for
this
work
is
and
just
explain
the
POC
that
I'm
working
on
so
for
those
who
are
new
here.
B
Prow
is
a
the
CI
solution
for
and
so
much
more
for
for
the
kubernetes
project,
and
there
is
a
component
within
proud
called
hook,
and
what
hook
does
is
it
listens
to
events
that
are
emitted
by
GitHub
and
reacts
to
those
events
and
the
the
reason
hook
exists
is
to
launch
a
set
of
plugins
and
at
a
very,
very
high
level,
and
the
job
of
those
plugins
that
are
run
by
hook
is
to
coordinate
all
of
the
incoming
changes
across
a
wide
range
of
repos.
B
In
order
to
do
things
like
maybe
automatically
label
the
size
of
an
incoming
PR
and
it's
a
lot
of
workflow
support
and
automation
to
Corral
changes
as
they
come
into
the
kubernetes
project.
Now
within
the
wider
cncf
landscape,
there
are
a
lot
of.
There
are
a
lot
of
other
incoming
projects
that
are
separate
from
kubernetes,
but
in
the
cloud
native
space,
the
the
the
thinking
behind
allowing
those
smaller
projects
that
will
be
resource
constrained
in
terms
of
human
resources
and
and
maybe
also
infra
resources.
B
The
the
idea
of
launching
allowing
prior
plugins
to
be
launched
from
a
GitHub
action.
It
is
to
allow
those
smaller
projects
to
leverage
the
workflow,
goodness
provided
by
prayer
plugins
and
in
the
context
of
not
involving
themselves
with
an
actual
proud
deployment
of
their
own
or
interacting
with
the
centralized
prior
deployment.
So
so
that's
the
that's
the
rationale
that
was
given
to
me
and
they
said
Rob.
Can
you
just
go
and
see
if
you
can
kick
off,
have
a
go
at
kicking
off,
proud,
plugins,
launch
from
a
GitHub
action?
B
So
so
I
wanted
to
go
and
do
that
and
in
a
nutshell,
it
is
possible
to
do
and
it's
possible
to
create
a
custom,
GitHub
action
and
the
way
this
works
at
runtime
is
that
you
create
a
container
and
you
make
it
comply
with
what
the
expectations
of
GitHub
are
in
relation
to
how
that
container
will
behave.
And
then
you,
when
you've
done,
that
you
effectively
have
a
custom,
GitHub
action.
So
what
I
have
done
is
I.
B
So
so
what
I've
done
here
is
essentially
I've
created
a
goal
line,
application
that
kind
of
mirrors
how
hook
behaves,
but
it
is
a
little
bit
different
because
the
runtime
is
different
and
like
hook
this
prior
GitHub
action,
container
spins
up
a
default
set
of
pro
plugins
and
it's
it's
life
cycle
is
a
little
bit
different
in
so
far
as
it
is,
kicked
off
by
a
GitHub
action,
and
then
it
dies
at
the
end
when,
as
when
as
processes,
the
the
events
just
kicked
off
that
GitHub
action
the.
B
So
it
has
a
much
shorter
life
cycle
because
Hulk
runs
as
a
long-running
service.
But
essentially,
if
you,
if
you
know
hook,
you
know
you
know
PGA
this
application
that
I've
developed
there
is
one
runtime
difference
that
has
resulted
in
me.
Submitting
this
PR
upstream,
and
the
runtime
difference
is
that
when
you
kick
off
a
when
a
GitHub
action
kicks
off,
it's
assigned
a
an
Ubuntu
runner.
B
So
you
get
an
Ubuntu
environment
for
free
a
little
bit
of
compute
for
free,
and
then
you
can
go
about
your
business
and
the
the
big
difference
from
an
authorization
point
of
view
is
that
the
GitHub
action
grants
you
an
ephemeral,
GitHub
oauth
token,
and
what
what
this
means
is
for
the
end
user
because
of
action,
end
user
or
the
project.
B
Maintainer
then
doesn't
need
to
worry
about
maintaining
that
oauth
token,
because
it's
ephemeral
because
it
only
lasts
for
an
hour
but
the
one,
the
one
limitation
that
that
oauth
token
has
is
it
cannot.
It
doesn't
allow
a
GitHub
client
to
access
the
slash
user
endpoint
in
the
GitHub
API.
So
so
the
pr
that
I've
submitted
adds
as
an
option.
B
The
fields
is
GitHub
action
and
that
then,
when
you
create
your
GitHub
client,
when
you
create
that
GitHub
client
within
within
PGA
or
or
anywhere
else,
you
can
tell
the
client
to
look
I'm
being
launched
from
your
GitHub
action
and
then
that
flag
is
then
used
to
guard
against
accessing
the
slash
the
slash
user
endpoint.
So
that's
why
that
PR
has
has
been
submitted.
B
I
can
I
can
move
quickly
to
a
to
a
demo
of
this,
and
then
I
could
take
questions
and
comments
and
I
can
leave
that
simple
people,
and
that's
just
a
little
bit
of
some
of
the
stuff
that
I
spoke
to
so
I
have
a
a
PGA,
sandbox
repo,
and
this
this
mock
project
repo
is,
is
meant
to
simulate
a
client
project
that
a
cncf
project
that
would
make
use
of
this
functionality
and
there's
a
little
press
say
as
to
what
this
repo
is
for.
B
If
we
go
to
issues
and
if
I
was
to
so
this
is
this
is
an
issue
that
I've
that
I
just
loved
it
moments
before
the
meeting.
What
happened?
I
tried
the
thing:
what
I
expected
to
happen?
I
expected
thing
to
work,
steps
to
reproduce,
etc,
etc.
Anything
else
we
need
to
know
so
so
that's
just
a
little
bit
of
book
templating
if
I
run
slash
help
here.
B
What
happens
here
is
that,
because
I
have
a
registered
GitHub
action
to
react
to
issue
comments,
I
launch
the
GitHub
action,
so
this
is
the
Run
of
that
there
and
we
can
see.
We
have
an
issue
comment
so
so
this
is
the
GitHub
action
that
makes
use
of
that
makes
use
of
Prior
GitHub
actions
latest
and
that's
the
container
being
pulled
down
there,
and
this
is
the
setup
for
the
job
very
useful.
B
If
you
are
developing
this
thing
to
developing
with
GitHub
actions
to
go
through
that-
and
then
here
is
here,
is
that
custom,
GitHub
action
running
and
one
of
the
one
of
the
things
of
note
in
terms
of
implementation
is
that
PGA
unlikehock
is
just
consuming
a
lot
of
environment
variables
and
because
this
is
being
launched
from
a
specific
GitHub
event
and
that
event
payload
has
consumed
as
an
environment
variable.
So
that
will
be
a
small
difference
between
PJ
and
hook.
B
And
then
this
is
just
the
log
of
everything
running
there.
So.
B
And
this
will
pretty
much
I'm
just
going
to
pull
that
stuff
out
of
the
way.
So
if
you
just
come
through
there,
you
would
see
that
that
that
helped
plug-in
running.
B
So
that
is
just
a
you
know,
a
whirlwind
tour
of
that
and
just
a
demo
of
of
that
working
the
right
now
right
now,
the
container
I'm
using
Co
to
build
the
container
and
publish
the
container
the
the
custom
get
of
action
to,
and
the
GitHub
container
registry
and
I
am
I,
have
embedded
a
plugins
that
yaml
to
configure
the
behavior
and
that
mirrors
plugins.yaml
in
in
Hook,
and
the
thing
that
I'm
working
on
at
the
moment
is
to
expose
that
to
a
project
maintainer
as
an
end
user
have
somewhere
possibly
in
a
repo,
so
they
can
customize
the
behavior
of
the
plugins
to
suit
their
needs.
B
But
yeah.
That's
that's
pretty
much!
That's
pretty
much
the
end
of
that
as
an
update.
So
if,
if
anyone
is
interested
so
hopefully
that
provides
context
and
I'll
link
into
this
presentation
to
that
PR
and
I
think
so.
Testing
and
the
sake
would
have
to
have
a
think
about
whether
or
not
they
want
to
accept
this
PR,
but
but
hopefully
that
will
provide
context
as
to
whether
or
not
that
was
something
that's
suggesting
we're
interested
in
taking
on.
B
And
if
we're
not
that's
fine,
we
can
work
through
that,
but
yeah,
that's
pretty
much
it.
So
if
anyone
has
any
questions,
I'd
be
free
to
take
them
now
or
later
or
in
the
slack
Channel.
B
A
B
I
can
I'll
happily
stop
if
there's
no
questions,
if
there's
a
question,
I
was
just
leaving
it
open
for
questions,
but
if
there's
no
questions
now,
that's
fine,
yeah
I'll
be
up
I'll,
be
up
to
kubecon
n
a
if
anyone
sudden
becomes
more
interested
in
this
or
if
anyone
looking
at
this
later
and
so
just
just
hit
me
up
and
and
I'll,
be
happy
to
feel
questions
or
give
Demos
in
person.
A
D
Thanks
for
sharing
I
I
think
the
main
thing
is
one
that
we
can't
answer
here.
I'm
I'm
interested
in
if
proud
maintainers
like,
are
comfortable,
maintaining
this
additional
mode
of
using
these
tools
like
it
like.
Basically
just
you
know,
if
there's
interest
in
supporting
this
from
owners,
I
think
if
there's
not
then
yeah
that
that's
not
a
lot,
the
rest
of
us
can
do.
B
B
Absolutely
yeah
and
that's
why
I
try
to
socialize
the
idea
like
I,
think
long
term.
The
the
people
who
are
who
are
expressing
interest
in
reviewing
this
as
a
POC
are
are
the
it's
tight
contributor
strategy,
group,
I,
think
and
and
I
think
the
the
long
yeah
the
long-term
vision
for
this
is.
B
If
adopted,
we
try
and
make
it
a
separate
Community
thing:
that's
community
supported
Community
can
try
and
create
Community
around
us,
but
from
a
strategic
point
of
view,
the
the
there's
plenty
to
think
on
and
and
chew
on
and
debate
really.
You
know
because.
D
Like
if
it's
a
separate
Community
point
of
view,
I
don't
know
if
it
needs
to
be
like
this
GitHub
client
like
like
I'm,
actually
not
recalling
what
it
was
exactly
that
caused
us
to
wind
up
writing
our
own
client,
like
instead
of
one
of
the
existing
client.
D
Libraries,
for
example,
I
think
that
prow
has
gotten
a
bit
trickier
as
we
wound
up
supporting
use
cases
other
than
kubernetes,
but
it
is
a
theory
that
we're
doing
like
there
is
code
in
there
for
interacting
with
like
Jarrett,
and
you
know,
the
kubernetes
project
has
no
Jared
instances.
That's
simply
other
adjacent
users,
so
you
know
I
think
there's
plenty
imprecedent
for
this,
but
yeah
it's
still
okay.
You
know
it's
up
to
other
folks
that
aren't
in
this
call
yeah.
B
If
I
was
to
animalistically
anthropomorphized
proud,
it's
almost
like
a
product
topping
away
at
a
Shell
trying
to
break
out.
You
know
and
both
yeah
absolutely
but
but
the
main
reason
I
wanted
to
socialize.
It
here
was
to
just
explain
that
explain
that
PR
and
give
background
on
that.
A
I
think
either
polar
chow,
or
somebody
has
already
seen
a
bit
of
this
on
the
slack
but
I'd
be
happy
to
appoint
them.
This
way
for
some
more
contacts,
and
maybe
part
of
strategizing,
because
our
team
is
definitely
doing
some
longer
term
planning
right
now
so
I'm
sure
that
will
be
a
fun
thing
to
throw
in
as
well.
Okay,.
C
A
Cool
all
right,
but
yeah
I,
agree
otherwise,
probably
need
someone
from
the
prep
team
to
talk
more
about
that
yeah
awesome
demo
thanks.
A
A
Apologies
I
misread,
which
name
that's.
A
E
Robert
cool
PR:
that's
some
really
impressive
work!
Great
demo
thanks!
Thank
you!
Pr
I
got
up.
It's
not
impressive,
like
yours,
there's
a
flaky
test.
The
reason
I'm
trying
to
get
the
floor
here
is
because
I
gotta
persuade
people
that
my
proposed
solution
is
the
smart
way
to
proceed.
The
test
is
failing
because
of
multi-art
support
on
a
very
old
version
of
tensorflow
test
that
is
out
and
I
tried
to
replace
it
with
a
newer
version,
and
even
that
would
require
a
substantial
amount
of
work.
E
I
tried
to
detail
that
work
in
the
pr,
so
you
can
see.
I
didn't
just
do
a
shortcut
to
get
something
to
turn
green.
The
PR's
been
open
for
some
time
now.
The
punch
line
is
stop
trying
to
run
the
test
on
Arch,
64
and
I.
Expect
this
single
turn
green
again
and
the
blast
radius
has
shrunk
down
to
just
the
very
small
test
itself,
so
it
should
be
safe.
At
this
point,.
D
Yeah
I
think
it's
a
little
bit
more
signal
call.
But
you
know
we
interview
this
up,
like
within
sick
testing.
We've
been
trying
to
push
e
to
e
tests
towards
like
wherever
possible,
we're
just
removing
all
these
extraneous
images,
and
we
have
this
one
image,
agnost
or
something
someone
came
up
with
and
it
it
it
targets
being
multi-arch
in
part
by
just
we
have
a
go
binary.
D
That's
a
Swiss
army
knife
of
like
you
can
tell
the
go
binary
like
be
an
HTTP
Echo
server
or
something,
and
we
have
just
a
bunch
of
sub
commands
for
useful
test
things,
and
then
we
cross
compile
that
and
have
like
one
image
that
works
everywhere.
But
for,
like
you,
know
running
this
tensorflow
workload,
I
can
I.
D
Ci
with
gpus
on
like
s390x
anyhow,
we
don't
have
any
CI
Upstream
on
s390x,
so
like
same
thing,
for
most
of
them
arm
is
one
that
we
do
have
some
ewci
I,
don't
think
any
of
it
has
gpus,
though
I
think
we're
just
talking
about
like
AWS
graviton
instances
in
terms
of
like
the
project
has
the
chaops
sub-project
has
run
some
CI
with
kubernetes
and
caps
on
on
arm
AWS.
D
D
We
don't
have
a
way
to
actually
exercise
this
on
those
platforms
anyhow,
so
building
for
it
is
wasted
energy
versus
just
like
deflating
the
test,
so
I
I'm,
a
big
plus
one
on
this,
but
I've
also
been
more
generally
plus
one
I'm
pushing
on
like
you
know.
D
Why
is
the
kubernetes
project
releasing
for
these
architectures
that
we
can't
test,
whereas,
like
you
know,
other
people
may
disagree,
I
think
as
long
as
there's
not
any
indication
that,
like
Sig
node,
disagrees
here,
I'm
plus
one
on
this
approach
to
get
it
deflaked?
And
you
know
if
someone
else
wants
to
come
through
and
upgrade
it
and
get
it
working
on
a
new
multi-arch
they're
welcome
to
but
I.
E
Right,
it's
not
that
it's
difficult
to
make
it
work
on
our
arms
sit
on
the
arms
that
tensorflow
thing
doesn't
even
work
on
that
anymore.
Whatever
image
was,
there
doesn't
seem
available
anymore
anyway,
let's
not
discuss
it.
Thanks
for
your
your
discussion,
that's
enough
said.
D
Well,
let
me
just
ask
you
real
quick,
like
have
you
got
any
indication
from
any
of
like
signode
leads
or
anything
on
like
how
they
feel
about
this.
E
D
I'll
lend
my
comments
to
it,
but
I'll
hold
on
like
mer,
like
you,
know,
pushing
it
to
merge
until
we
confirm
that
right.
C
D
I
think
that
was
the
last
topic,
so
I'm
gonna,
just
like
something
real
quick.
We
have
kubecon
next
week.
There
is
a
Sig
meet
and
greet
session.
I
will
drop
a
link
again
in
the
slack.
If
folks
are,
you
know,
feel
comfortable
helping
like
meet
and
greet
people
coming
to
learn
about
the
Sig.
D
We
need
people
to
sign
up,
I,
think
I'm
a
little
bit
behind
on
the
compass
but
I
think
I'm,
the
only
one
that
signed
up
right
now
and
I'm
also
one
of
the
only
ones
signed
up
for
infra,
so
actually
really
either
of
those
things
if
you
will
be
at
kubecon
and
would
like
to
help
meet
and
greet
people
coming
to
look
just
like
ask
questions
about
the
second
whatnot.
D
There
will
there's
a
sign
up
issue
and
I'm
gonna
go
grab
that
link
and
drop
it
in
the
suggesting
slack
shortly.
Foreign.
D
Thanks
for
coming,
y'all
I
think
we
can
close
that
early
and
enjoy
the
rest
of
your
Tuesday.