►
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
A
Yep
looks
good,
okay,
hello,
everyone,
I'm
ashitos,
and
this
is
kubernetes
C
cluster
lifecycle,
cluster
API
provider,
Azure
officers
and
today
is
third
August
2023.
A
We
abide
by
the
CNC
of
card
of
conduct,
which
means
that
be
kind
to
each
other
and
raise.
Hence,
if
you
want
to
talk,
I,
don't
see
any
new
faces,
but
I'll
just
pause
for
like
a
couple
of
seconds.
If
anybody
wants
to
introduce
themselves.
B
Yeah
I'm
a
new
year
yeah
I'm,
Thomas,
I,
I,
folks,
yeah
I'm.
Basically,
software
engineer
in
Microsoft
and
I'm
trying
to
use
capsi
so
I
had
some
question,
but
I
can
wait.
I
mean
I,
didn't
have
a
chance
to
add
myself
into
the
agenda
for
today,
because
I
I
just
asked
her.
Yesterday
the
I
mean
access
to
edit
the
document,
but
I
can
wait
until
then
I
mean
no
rush.
B
A
Yeah
feel
free
to
add
to
the
agenda
Thomas
sure,
okay,
so
I
am
going
to
the
agenda
right
now
and
John.
C
Yeah,
this
is
just
a
quick
announcement
that
we
did
cut
new
patch
releases
this
week
for
v19
and
v110.
There's
just
one
tiny
bug
fix
affecting
manage
clusters,
but
no
other
user
facing
changes,
or
at
least
significant
user
facing
changes,
but
yeah
feel
free
to
check
out
the
releases
for
more
details
on
those.
A
Thanks
Tom,
anybody
has
any
questions
comments
on
this
foreign.
A
D
Yeah
I
put
this
down,
but
I
think
anyone
could
speak
about
it.
I
think
this
is
mostly
relevant
for
you
and
Billy,
but
so
just
wanted
to
give
some
just
context
for
folks
who
haven't
been
following
on
slack.
We
are
as
part
of
moving
to
the
SDK
V2
and
workload
identity
running
into
some
issues
where
the
default
flow
of
essentially
creating
credentials
in
the
SDK
V2
is
a
bit
different
and
because
of
how
our
cab
Z
credentials
work
right.
D
Now,
where
you
can
have
a
fallback
to
The
Container
environment
variables,
which
has
been
deprecated
for
a
long
time
since
release
1.5,
but
it's
still
available,
we
have
to
kind
of
do
some
very
like
work
around
PR.
So
one
thing
that
we
discussed
this
week
is
why
don't
we
just
completely
remove
the
manager
credentials
since
they
have
been
deprecated
for
over
five
releases,
and
we
did
say
when
we
introduced
Azure
cluster
identity,
that
they
would
be
deprecated
eventually,
so
just
to
clarify
what
this
means.
D
This
means
that
you
wouldn't
be
able
to
set
environment
variables
for
your
Azure
service
principle
as
part
of
the
overall
manager
anymore,
to
be
shared
with
all
of
your
clusters
that
are
reconciled
by
kebzy.
It
would
be
a
per
cluster
reference
in
the
Azure
cluster
spec.
D
That
doesn't
mean
that
you
need
to
use
Azure.
That
doesn't
mean
sorry
that
you
need
to
use
adpod,
Identity
or
workload
identity,
there's
still
an
option
for
manual
service
principle,
which
is
just
a
client
secret,
client,
ID,
10,
ID
pair,
so
you'd
still
be
able
to
use
eyes
just
like
you'd
have
to
add
the
reference
for
each
on
the
actual
clusters
like,
as
opposed
to
using
the
environment
variables
in
the
manager.
D
So
if
anyone
out
there
is
still
using
this
way
of
authenticating-
and
you
know,
has
a
good
reason
for
it-
I'd
love
to
hear
about
it.
Please
speak
up,
but
other
than
that
I
think
we're
planning
on
deprecating
the
environment
variables
as
a
release,
111,
which
is
coming
in
September.
E
Mike
just
a
comment:
I
well,
unless
I'm
reading
it
wrong,
I
believe
the
Azure,
the
cluster
API
provider,
Azure
book,
so
references
those
environment
variables
for
the
quick
start
guide.
D
So
the
environment,
so
you
would
still
set
environment
variables
like
in
your
local
environments
and
that's
what
gets
put
in
the
template
so
that
flow
doesn't
change.
It's
just
that
the
environment
variables
are
set
on
the
yaml
spec
that
you
apply
to
the
management
cluster
as
opposed
to
on
the
deployment
itself.
When
you
do
cluster
still
in
it,.
D
A
Yeah
I'm
I'm
personally
plus
one
to
this,
because
the
spindle
source
of
confusing
for
a
while
so
yeah
anybody
have
any
thoughts
like.
F
And
I
just
had
a
question
since
I
haven't
checked
in
on
the
pr
in
a
while
this.
This
is
going
to
simplify
our
implementation.
Quite
a
bit
right,
Willie,
there's
just
going
to
essentially
be
one
auth
method
where
we
inspect
the
Azure
cluster
identity
and
then
do
the
right
thing.
So
that's
sort
of
how
it's
going
to
pan
out.
G
Yeah
I
found
think
I
needed
to
change
any
like
implementation
into
code.
It's
just.
We
have
some
yaml
files
that
import
the
cus,
the
environment
variables
into
the
tabsy,
editor,
spec,
I,
think
and
yeah.
Just
removing
that
would
would
get
the
job
done.
I'm
testing
the
pr
now
with
it
works
with
like
non-workload
identity,
but
right
now,
I'm
just
trying
to
get
it
to
work
with
workload,
identity
and
running
into
a
little
bit
of
an
issue
with
just
getting
workloaded
I
need
to
work
but
I'm.
A
G
Of
course,
thank
you,
I
forgot
to
send
a
message
about
there.
I
got
before
I
signed
up
last
night,
but
I
just
sent
it
in
the
thread
so
yeah
awesome.
Thank
you.
A
Thanks
for
taking
notes,
map
looks
like
I,
don't
see
anything
else
before
I
go
to
the
next
item.
Anything
else
folks
once
wants
to
comment
or
say:
I'll
just
pause
for
a
couple
of
seconds
season.
D
Yeah
I
just
want
to
answer
my
Mike's
question
in
the
chat,
and
so
I
did
I
thought
you
were
talking
about
the
Cappy
quick
start,
which
I
just
took
a
look
at
so
the
way
that
the
manager
environment
variables
get
that
is
with
the
base64
encoded
variables
so
that
we
can
get
rid
of
but
having
them
said
it
won't,
hurt
anything
it's
just
that
they
won't
be
needed
anymore
and
then,
when
you
apply
the
template
and
create
a
secret
with
like
the
Azure
cluster
Identity
or
for
the
Azure
culture
I
didn't
need
to
use,
then
you
would
typically
use
the
non-base
64
encoded
environment
variables,
which
the
getting
started
guide
kind
of
points
to
so
I.
D
Think
the
main
difference
is
the
order
of
things
where
like
before.
If
you
were
to
use
the
manager
environment
variables,
you
would
set
those
environment
variables
in
your
environment
and
then
do
closer
steel
init
dash
dash
in
for
Azure
and
those
environment.
Variables
would
get
used
to
create
a
secret
to
be
used
by
the
container
to
set
environment
variables.
But
now
you
don't
need
credentials
to
initialize
Azure
anymore.
D
You
can
actually
initialize
the
management
cluster
without
credentials
and
then
once
you
want
to
create
a
workload,
cluster
you'll
need
to
have
some
identity
reference
that
has
those
credentials
in
it.
So
that's
where
those
environment
get
variables
get
used.
Does
that
make
sense?
Does
that
answer
your
question.
E
D
A
Awesome
so
the
tmdr,
it's
a
sale
if
I
get
it
rightly
with
this
deprecation,
the
quick
start
guide
will
still
function
appropriately
without
any
problems.
Correct
awesome,
I
know
like
Thomas,
was
talking
about
adding
to
agenda
like
Thomas.
Do
you
want
to
like
talk
about
your
item
right
now,.
B
F
You
you
might
have
to
go
to
the
sorry
to
interrupt.
You.
You'll
probably
have
to
share
hosts
with
him
ashtush.
If.
A
A
B
B
Yes,
okay,
good,
so
let
me
give
you
a
little
bit
of
context
about
what
we
are
trying
to
do
with
the
cap.
Z.
Fundamentally,
we
have
a
end-to-end
pipeline
that
we
want
to
enable
to
test
our
product
and
in
the
past
we
used
to
use
AKs
engine
to
provision
this
self-managed
kubernetes
cluster,
not
AKs,
so
in
azure
is
deprecated.
We
are
moving
to
capacity.
B
Fundamentally,
the
goal
is
to
have
a
skilled
test
environment
for
us,
so
we
would
like
to
run
an
Edo
pipeline
that
provision
on
demand
one
or
two
are
and
capacity
clusters,
with
even
big
number
of
nodes,
in
my
case,
I'm
trying
to
create
a
capacity
with
400
nodes,
for
instance,
but
the
scale
that
we
would
like
to
test
ultimately
is
even
thousands
of
nodes
like
four
thousand
five.
Thousands,
because
our
customers
are
basically
that
big.
B
The
thing
that
I
noticed
that,
when
I
tried
to
switch
to
capacity
is
that
I
try
to
basically
provision
a
cluster
like
this
through
our
pipeline
and
the
management
cluster
was
actually
creating
all
the
resources
in
Azure.
In
fact,
this
is
basically
the
resource
Group
that
I'm
using
to
provision
all
the
resources,
and
you
can
see
that
there
are
401
VMS,
because
I
have
one
control,
plane,
node
and
for
Android
the
worker
nodes.
These
are
basically
400
working
nodes
that
I
want
to
have.
B
The
problem
is
that
I'm?
Seeing
is
that,
basically,
even
if
these
resources
are
present
in
Azure,
they
don't
show
up
through
the
kubernetes
apis.
They
basically
are
less
than
that,
and
in
particular
I'm,
always
seeing
that,
even
after
an
hour
of
provisioning,
all
the
VMS
are
basically
listed
here.
B
They
are
397
actually,
because
if
I
do
get
notes
on
the
cluster
that
I
provision
and
do
work
out,
basically,
this
thing
has
two
rows:
one
is
the
adder
and
one
is
the
control
plane,
so
actually
397
web
provision,
so
three
web
missing
and
at
the
same
time,
basically
the
resource
Group
had
all
the
VMS.
So
I
was
wondering
if
there
is
any
known
issue
that
the
number
of
VMS
in
the
resource
Group
might
differ
from
the
nodes
in
the
cluster,
because
I'm
seeing
basically
this
Behavior,
especially
when
they
create
big
clusters.
B
D
So
yes,
I
think
that
would
be
first
good
step.
So
the
thing
is
that
happens
when,
when
so,
when
VMS
get
created
in
order
for
them
to
join
the
kubernetes
cluster
as
nodes,
if
you're
using
cabsy
with
the
example
templates
you're
most
likely
using
the
cubadian
bootstrap
provider,
and
so
the
way
that
nodes
join
is
each
VM
in
the
cloud
in
its
run
will
run
a
cubadian
join
command
which
joins
the
node
to
the
cluster.
D
B
D
Should
lead
you
to
any
number
of
reasons
like
a
network
like
usually
it's
Network
could
be.
The
API
server
has
like
an
issue
at
that
moment
where
it
tries
to
join,
and
so
it's
very
likely
that
those
notes
that
are
not
showing
up
the.
If
you
went
on
the
cloud
and
it
logs,
you
would
see
cubadian
join
exited
non-zero
and
so
the
VM
failed
to
join
the
cluster
as
a
node.
D
So
it's
a
little
more
resilient
and
so
you'd
normally
be
able
to
like
have
a
higher
success
rate.
I.
H
D
G
B
D
So
that's
two
things
you
can
do
another
thing
you
can
do
also,
if
that's
still
not
enough.
Cluster
API
has
a
concept
of
machine
health
check
and
those
machine
health
checks.
What
they
do
is
they
watch
the
machines
and
if
they
see
that
a
machine
is
unhealthy
and
we'll
delete
the
machine,
so
a
new
one
gets
created.
B
D
B
A
A
A
So
Mike
you
had
your
and
Teresa.
Yes.
A
Problem
so
with
this
I
think
figure
at
the
end
of
the
agenda,
so
they
think
we
said
to
a
milestone,
review
mapped
so
so.
I
I
would
propose
going
through
the
pr
triage
because
you
have
39
PRS
up
there,
but
I'm
open
these
folks
have
something
else.
They'd
rather
do.
E
A
E
Actually
I
do
have
something
real
quick.
Is
there
a
way
to
get
a
like?
You
know
a
60-second
tldr
update
of
the
migration
to
ASO
how
that's
going
I've
been
way
out
of
the
loop
just
curious.
How
that's
been
going.
A
Right
John
has
the
same
address.
C
Yeah
I
can
comment
on
that.
So
I
think
there
are
a
couple
outstanding
items
left
before
we
can
kind
of
comfortably
turn
on
So
within
cap
Z
and
have
cap
Z
manager,
resource
groups
with
ASO
I.
Think
that
will
be
kind
of
the
first
step
into
using
ASO
is
just
to
use
or
just
to
manage
resource
groups
with
ASO
and
we're
trying
to
get
all
those
items
cleaned
up.
I
J
A
Cool
okay:
let's
move
to
PR
files,
then,
hopefully
my
screen
is
visible.
A
Okay,
so
how
do
we
go
through
this?
It's
like
one
by
one's
like.
I
Yeah,
so
I
would
there's
one
I
I
just
updated
a
bit
of
the
authors
and
stuff
and
some
of
the
statuses
there's
one
which,
at
the
very
end,
with
no
status,
which
is
Matt's.
I
I
F
Yeah,
it's
it's
weird.
One
of
the
windows
related
tests
has
failed
three
times,
but
I
can't
imagine
how
this
change
has
anything
to
do
with
it.
In
fact,
I'm
sure
it
doesn't
so
because
this
is
just
switching
to
a
newer
version
of
a
package
that
some
unit
tests
use
so
I,
I,
guess
that
requires
separate
investigation
at
this
point,
but
this
is
kind
of
like
a
this
is
a
prerequisite
for
using
async
services
in
SDK
V2.
So
this
was
something
I
could
separate
out.
A
By
the
way
which
which
PR
or
line
item
are
we
talking
Cecil
also
has
this
question
number
106.,
26.
I.
A
D
I'm
sorry
I
could
probably
ask
this
in
the
pr
review,
but
has
have
any
other
kubernetes
project
switched
to
this
work
are,
do
we
know
if
it's
safe
for
use
license
wise.
F
It
is
a
straight
up
Fork,
so
it
has
the
same
license
and
the
official
repository
now
literally
points
to
this
I
I
went
I
went
over
to
look
at
Cappy,
but
Cappy
doesn't
actually
use
mock,
gen
and
I
gave
up.
So
there
probably
are
other
kubernetes
projects
using
it.
I
don't
know
who.
A
Thanks:
Matt,
okay,
so
how
do
we
want
to
like
go
through
this
items
like
sell?
We
go
into
like
the
leads
review,
one
or
the
weight
on
author
or,
if
anybody.
G
Sorry,
I
raised
my
hand
too.
Early
I
was
gonna
talk
about
the
first
one
go
ahead,
but
yeah
for
mate
skill
set,
reconcile,
slash,
delete,
async
I,
think
it
has
lgtm
for
me.
I
think
it
just
needs
another
round
of
review.
G
Jonathan
has
been
because
it's
like
the
bottom
of
the
pr
view.
So
a
lot
of
people
haven't
seen
it.
So.
Oh
sorry,
Cecile
I
think
you're
also
reviewing
it.
D
Oh
yeah,
no
you're,
good
I
just
was
gonna,
say
I
added
a
few
comments
yesterday
that
are
still
open,
I
think.
D
Think
we're
good.
Then
it
could
definitely
use
more
eyes
if
people
are
able
to
I
think
I
cc'd
you
met
for.
There
was
a
specific
thing
about
Flex
versus
uniform
that
we
weren't
sure
about.
F
H
Yep
I
also
started
on
a
it's.
The
scale
set
VM
refactor
as
well,
so
so
I
replaced
it
off
that
branch
and
then
I'll
get
to
these
comments.
H
Next
one
yeah
yeah
vm1
doesn't
have
a
PR
yeah,
it's
just
just
on
my
local
machine,
but
yeah
we're
good
to
move
on.
Okay,.
A
Thanks
folks,
I'll
also
like
give
it
that
look
this
piano
today.
The
next
one
is
again
Billy.
You
want
to
talk
about
this,
about
free,
existing
network
interfaces.
G
G
No
worries
yeah,
so
I
think
Dane
is
working
on
this,
but
he
can
kind
of
out
I
think
so.
This
is
kind
of
just
stale
at
the
moment,
but
I
think
there
are
a
few
comments
that
you
need
to
address.
So.
C
Yeah
I
can
comment
on
this
one.
So
I
think
that
proposal
is
mostly
complete.
I
think
Benny
just
needs
some
more
eyes
on
this
one,
including
my
eyes,
so
I
need
to
take
another
look
at
this
but
yeah.
Obviously
you
can
see
from
the
149
commits
that
there's
some
rebase
and
squashing.
We
need
to
do
on
that
yet,
but
otherwise
I
think
all
the
content
is
I,
don't
think
there's
any
changes
planned
there
so
yeah.
C
A
I
So
I
would
probably
put
this
as
weight
on
author,
then
I
I
would
just
but
I
would
because
they
he
needs
to
just
rebase
and
squash
commits
and
then
theoretically,
when
lazy
consensus.
But
obviously
we
also
open
open
that
up.
If
anyone
wants
to
review
so,
but
at
least
that
way
we're
not
continuing
to
check
it
over
and
over
again
so
yeah,
maybe
change
the
weight
on
weight
on
author
until
he
does
the
rebase.
A
Sure
let
me
do
that.
A
D
Yeah
Jonathan
I
have
been
reviewing
this
back
and
forth
and
I
think
it's
on
us
to
go
back
and
check.
I
was
going
to
do
another
review
today,
but
I
think
it's
ready
and
all
the
comments
are
addressed.
So
waiting
on
review
is
good.
Awesome.
D
Yeah
I
think
this
is
like
we
need
to
decide
if
we
actually
want
to
do
this
waiting
on
comments.
I
guess-
and
there
were
a
few
comments
against
doing
this,
so
it's
not
clear
that
we
want
to
move
forward
with
this,
as
is
God.
A
A
C
Yeah
I
think,
overall,
we
still
need
to
do
a
little
bit
of
planning
as
to
how
we're
going
to
start
merging
some
of
Benny's
work.
That
he's
done
so
far,
so
I
wonder
if
a
good
steps
for
this
one
would
be
in
progress
since
I.
Don't
know
if
we're
ready
to
merge
this
stuff.
Quite
yet.
C
A
It
makes
sense
anybody
has
thoughts
on
this
one
before
I
move
to
the
next
title.
Okay,
so
the
next
one
is
service,
cidr,
Edition,.
A
C
C
I
Can
we
can
we
ask
somebody
to
these
two
PRS
because
we
have
nobody
assigned
to
either
of
them.
F
I'll,
take
that
too
I
mean
I
did
a
little
review
this
morning,
okay,
but
it
could
use
more,
buys
I.
Suppose
it's
not
very
big.
F
Big,
so
this
is
not,
strictly
speaking,
a
prerequisite
for
SDK
V2,
but
if
you
start
transitioning
things
to
SDK
V2
from
SDK
V1,
you
start
to
see
patterns.
One
of
the
big
patterns
is
hey.
We
changed
this
field
to
a
pointer,
or
this
used
to
be
a
pointer
to
an
array
of
strings.
Now
it's
an
array
of
string
pointers,
so
those
are
two
transitions.
They
just
changed
everywhere
and
once
I
started
doing
the
refactoring,
it's
very
difficult
to
work
with
constants
and
treat
them
as
pointers
tldr.
F
This
Library
solved
that
problem
and
then
once
I
waited
in
here.
It
really
wants
you
to
start
using
this
approach
everywhere,
because
if
you
continue
using
the
old
package,
you
get
deprecation
warnings
unless
you
sign
them.
So
this
is
just
mechanical.
Switching
of
all
our
uses
of
the
old
pointer
package
to
the
new
PTR
package.
F
A
F
But
that's
a
huge
it's
a
huge
PR
and
you
know
I,
don't
know
if
it'll
make
it
harder
for
us
to
cherry
pick
things,
probably
not
that
consequential.
A
I
So
I
think
do
we
have
do.
We
have
a
assignee
for
the
ASO
pause
logic
from.
A
A
Yeah,
so
this
review
Saxon
is
done.
Do
you
want
to
like?
Shall
we
go
through
this
weight
on
order
or
we
are
good.
A
I
also
see
like
there
are
four
blocked
PRS
like
David.
Do
you
think
like?
Is
it
good
to
all
of
this.
J
I
So
I
don't
think
we
need
to
cover
those
unless
I
mean
the
only
one
that
I'm
kind
of
curious
on
is
why
can't
we
move
to
ASO
operator
V2,
because
I
know
that
was
something
we
actually
really
needed
because
of
the
the
bump,
with
the
fix
they
had
in
there.
So
yeah
go
ahead,
I
don't
know
who's
first
matter,
John.
C
I
I
think
we're
probably
going
to
say
the
same
thing,
but
I
think
all
those
depend
about
PR's
are
kind
of
blocked
on
the
Cappy
one
five
stuff
since
the
Cappy
one
five
comes
with
the
controller
runtime
zero
one
five,
then
the
ASO
also
needed
to
kind
of
move
with
that
in
sync,
so
yeah,
the
Cappy
V
1.5,
PR
kind
of
involves
a
lot
of
significant
dependency
bumps
yeah
Matt
did
you
have
anything
to
add.
F
Yeah,
that's
exactly
it
I
guess,
probably
that's
that
PR
should
be
and
needs
a
review
status,
so
I'll
go
make
that
change.
It's
another
really
big
PR!
That's
mostly
mechanical
stuff,
although
there
are
a
few
interesting
changes
and
I
have
two
questions
about
how
we
handle
things
in
the
tests,
so
so
yeah.
I
I
just
moved
it
needs
review.
Can
we
can
we
have
somebody
assigned
to
review
that
to
help
out,
because
that
sounds
like
that's
a
domino
once
you
get
that
one
that
we
can
basically
just
immerse,
all
of
the
independent
Bots
or
just
because
there's
one
that's
Azure
cloud
provider
to
126
to
127.
Is
that
one
fair
game
two.
F
A
I
There's
no
there's
no
definite
set
of
things.
I'd
actually
vote
for
a
bug.
Triage
would
be
the
next
one,
because
there's
a
lot
of
old
bugs
there.
That
I
think
could
use
some
eyes
but
I'm
again
I'm
open.
If
people
feel
like
Milestones,
more
important.
B
I
I,
don't
think
you
can
order
based
on
date,
unfortunately,
with
the
this
is
the
projects
board
which,
which
is
which
is
a
flaw
that
has
a
filed
bug
if
we
want
to,
for
that
reason,
revert
to
the
actual
issues
and
then
filter
on
bugs
that's
fine.
The
only
advantage
of
going
with
this
view
is
that
you
can
pretty
easily
see
all
of
them
and
change
things
with
the
status,
but
but
yeah
to
find,
which
is
the
oldest
one.
I
There's
no
unfortunate
way,
though,
to
like
sort
by
date
of
issue
and
that's
that's
a
limitation
of
the
of
the
thing
yeah
so.
I
Wanted
to
go
by
that,
like
there's,
there's
one
1876
I
think
that's
the
oldest
one,
which
is
number
10
on
the
list.
If
you
wanted
to
go
with
the
oldest
first,
you
could
also
you
could
also
filter
on
the
life
cycle
so
that
you
could
look
at
the
life
cycle
label
and
there's
some
that
are
life
cycle,
rotten
or
life
cycle
Frozen.
D
Yeah,
for
some
reason,
I
don't
see
the
like
newest
issues.
There
are
bugs
in
the
issue
like,
for
example,
there's
one
from
three
days
ago:
code,
job,
failing
with
token
error,
that's
labeled
as
kind
bug.
I,
don't
see
that
one
in
here.
I
Okay,
why
don't
I
look
at
that
separately?
But
if
we
want
to
spend
spend
time
on
reviewing
the
actual
bugs
so
that'd
be
good,
because
I
think
that's
that's
an
automation
problem
that
I
can
look
into.
D
A
Okay,
Nawaz,
you
have
your
Androids
yeah.
J
I
I
could
speak
on
the
first
issue
that
depend
about
code
gen1.
C
J
Was
the
one
who
had
opened
it
up?
It
has
been
in
the
back
burner
it's
more
of
a
observation
which
I
found
that
we
have
a
GitHub
workflow,
which
is
supposed
to
update
the
two
links
present
in
our
hack
tools,
which
also
has
a
Go
Mart,
but
it
is,
it
is
not
really
doing
anything.
I
haven't
seen
even
a
single
comment,
so.
J
I
think
I
need
to
sit
with
that
code
gen
and
see
if
the
main
targets
were
invoking
are
even
correct
or
if
there
is
anything
we
have
to
do
with
like
if
we
should
even
run
that
GitHub
workflow
every
time
depend
on
whatever
you
disappear.
That
is
what
the
issue
is
trying
to
look
into.
Go
ahead.
Succeed.
D
I
was
gonna.
This
is
kind
of
a
question
for
everyone,
but
also
David.
Would
there
be
value
in
US,
maybe
like
as
we're
going
through
this
adding
priority
label
to
these
so
like
either
like
urgent
or
long
term
or
me
like
I,
don't
know
what
the
others
are,
but
because
I
think.
Ideally
we
want
to
fix
all
these
bugs,
but
it's
just
that
some
of
them
are
more
urgent
than
others
like,
for
example.
This
one
is
not
super
high
priority.
It's
something
that
we'd
want
to
fix.
D
I
Yep
yep
yeah
I
do
I
think
that
is
a
great
plan
as
we
triage
these,
then
at
least,
if
we
put
a
priority
label
on
them.
We
know
we've
at
least
looked
at
them
and
given
in
a
rough
priority.
So
with
this
one
look
at
the
existing
labels
that
we
have
there
we
have
prior,
if
there's,
if
you
just
type
priority
in
the
label
list.
Well,
if
you
click,
you
have
to
click
on
the
labels,
so
go
ahead,
and
if
you
click
on
the.
I
The
to
the
or
see
where
those
the
label
yeah
right
there
yep
and
if
you
just
type
priority,
those
are
the
those
are
the
tiers
that
we
have
so
kind
of
the
awaiting
more
evidence
is
kind
of
like
you're,
not
sure
that
you
even
want
to
fix
this
at
all.
Then
the
next
lowest
one
would
be
priority
backlog
and
then
they
kind
of
go
various
directions.
After
that,
there's
a
critical,
urgent,
important,
long-term
importance.
Soon,
I
feel
like
with
this
one.
I
It's
probably
just
priority
backlog,
because
it's
like
we
validated
that
it
is
something
we
want
to
do.
But
it's
not
like
super
critical
to
to
fix
it
or
I,
wouldn't
say
it
soon
or
I,
wouldn't
say
it's
like
urgent.
So
party
backlog
feels
like
the
right
thing
to
me,
but
I'm
open
for
feedback
here.
I
And
that
would
just
kind
of
like
and
I
think
it's
a
little
false
misnomer
because
they
all
say
priority,
but
but
to
me
like
priority,
backlog
is
more
of
just
a
checkbox
that,
yes,
we're.
Not
we
we
validate
that
it
is,
is
a
thing
that
we
should
fix,
but
it's
not
urgent
or
critical
or
like
this,
you
know
we
have
to
fix
this
long
term.
A
Kind
of
make.
I
A
Awesome,
so
what
I'll
do
is
I'll
run
through
one
by
one,
real,
quick
I
think
we
have
10
more
minutes
and
I'll
pause.
If
anybody
has
anything
so
the
second
one
is
not
possible
to
run
camps
in
AWS.
A
A
E
So
this
one
looks
like
it
has
a
bit
of
a
workaround
as
well
as
it's
something
somewhat
temporary
as
we
move
to
work
with
identity.
E
You
know,
especially
since
you
know,
I
mean
from
my
experience,
I
mean
I'm
running
taxi
on
AWS,
along
with
Kappa,
and
it's
working,
fine,
so
I
think
it's
definitely
a
I
mean
I
have
run
it
and
run
into
this
problem
before,
but
I
I
do
think
that
there
are
ways
to
work
around
it.
So
I
don't
know
what
kind
of
priority
this
really
is
versus
it
being
a
problem.
D
Should
we
consider
closing
the
issue
seems
like
the
person
got,
the
workaround
thing,
you
didn't.
We
don't
plan
on
fixing
this
long
term.
So,
okay,.
I
The
only
the
only
thing
is
that
I
guess
Mike
if,
because
you've
got
experience
with
this,
I
mean
if
you
feel
like
the
workarounds
you
know
of,
are
represented
in
here
cool.
If,
if
you
wanted
and
you
and
you
have
a
pointer
to
some
other
workaround
that
you
know
of
that
would
be
helpful,
but
otherwise
I
I,
don't
think
it's
something
that
we
probably
want
to
try
and
fix,
because
there
are
workarounds
and,
like
you
said,
there's
the
update
to
work.
E
Yeah
I
mean
we
essentially
the
the
workaround
says
you
know,
don't
run
nmi
or
run
it
with
don't
intercept
or
whatever
it
is,
and
that's
the
way
that
we
run
it
is
we
just
we
don't
run
the
nmi
and
we
just
use
the
service
principles
for
authentication
until
workload
ident
or
to
the
has
been
working
on
is
fully
in
and
I'm
out
of
touch
if
that's
already
been
merged.
In
the
last
two
months
since
I've
been
on
vacation,
okay,
cool.
A
Awesome
so
David
used
to
want
to
say
something:
oh.
I
A
On
okay,
so
the
next
one
is
test.
Reconciling
bone
natural
I
think
it's
a
good
first
issue,
which
is
I,
think
really
has
commented
sometime
back
and
we
have
not
heard
from
the
author
I
think
it
makes
sense
to
like
put
it
into
backlog.
What
do
you
say.
J
Just
a
thought
on
top
of
my
head:
do
we
want
to
prioritize
or
assign
any
priority
to
good
first
issues,
because
it
could
give
a
a
sense
to
the
community
that
hey
these
are
not
of
priority?
Maybe
I
should
not
pick
it
up.
That's.
A
A
very
fair
point:
I,
don't
know
how
to
answer
this,
but
I'm
thinking.
I
So,
if
I,
so,
the
question
is,
is
there
a
way
we
can
help
people
prioritize?
The
good
first
issues
is
that
what
the
question
is.
J
No,
the
question
is:
if
we
assign
a
priority
to
help,
wanted
or
open
to
Community
first-timers
kind
of
an
issue,
it
could
give
out
a
meaning
that
hey
this
is
a
old
issue
or
a
backlog
issue.
We
don't
really
need
to
jump
into
it
and
work
on
it.
J
So,
for
instance,
the
test
reconcile
inbound,
not
rule
test
cases
cannot
be
run
in
parallel.
It's
a
good
first
issue,
meaning
it
could
be
for
anyone
who's
trying
to
get
into
Cappy,
and
since
we
have
assigned
a
priority
backlog
to
it,
if
a
first
timer
would
look
into
it,
it
should
be
like.
Oh,
this
is
a
backlog
could
be
super
old.
Do
I
really
want
to
contribute
to
it,
so
it
could
give
out
that
feeling,
but
I'm
not
sure
John.
C
Yeah
I
could
see
where,
like
if
a
issue
is
tagged
with
a
lower
priority,
then
that
might
encourage
people
to
do
it,
because
then
people
don't
necessarily
feel
rushed
to
do
it.
So
I
guess
the
point
I'm
trying
to
make
is
that
I
think
different
people
are
always
going
to
interpret
priorities
in
different
ways.
So
I
don't
know
if
we
should
create
any
rules
based
on
that
really
go
ahead.
Cecile.
D
Yeah
I
was
gonna,
say,
I.
Think
one
of
the
rules
for
good
first
issues
is:
they
have
to
be
low
priority
because
you
don't
want
newcomers
who
are
picking
up
something
for
the
first
time
to
be
on
a
critical
path
or
something
or
to
be
fixing
an
urgent,
critical
bug,
so
I
think
actually
does
I
agree.
I
agree
with
John,
I
I
would
think
the
other
way
where
it
might
give
people
peace
of
mind
that,
like
this
is
something
we
want
to
do,
but
there's
no
like
urge
on
it.
A
A
So
the
next
one
is
changing
VM
size.
It
is
not
assigned
to
anybody.
A
D
A
Awesome,
the
next
one
is
log
editor,
which
is
help
wanted,
so
it's
already
and
priority
backlogs
I
think
we
can
skip
this
one.
Unless
anybody
wants
to
talk
about
this
one,
the
next
one
is
make
as
over
an
async
spec
type
default.
It's
life
cycle,
Frozen
charm.
Do
you
have
any
comments
on
this?
If
you're
here.
A
D
A
D
I
think
it's
actually
the
previous
one.
It's
actually
maybe
mislabel
I,
don't
know
that
that
is
really
a
bug.
It's
more
of
a
code
cleanup
issue
right.
A
D
Sorry,
yeah
I
I
think
that
seems
are
pretty
important,
like
that
seems
like
more
than
backlog
to
me
just
because
we've
kind
of
ignored
it
so
far
it
doesn't
mean
it's
not
important,
but
I
don't
know,
makes.
F
I
think
it
could
be
important,
I,
don't
think.
They've
really
proven
the
case
that
there's
an
actual
memory
leak
and
then
somebody
followed
up
and
kind
of
gave
it
new
life
after
a
while,
but
they
don't
seem
to
be
referring
to
the
same
thing.
I
have
put
in
effectively
a
couple
of
days
trying
to
reproduce
this
and
working
with
Pete,
Rock
and
I
wasn't
able
to
reproduce
it
and
so
I'm
not
sure.
Obviously,
if
there's
a
real
memory
leak,
this
is
very
important,
but
I
don't
think
I've
had
enough
time
to
take
it.
There
really.
H
I
So
I
would
I,
because
we're
going
to
be
doing,
we
still
have
on
the
plan
with
ASO
to
do
a
bunch
of
performance
testing
which
should
include
memory
work
I
would
be
in
favor
of
actually
closing
this
out
with
this,
the
specula,
the
the
explicit
call
out
the
fact
that
we
are
going
to
test
for
and
evaluate
memory
leaks
as
we
do
scale
testing.
I
D
Yeah
we're
at
time,
but
real,
quick,
I
I,
don't
think
we
should
close
it
because
closing
it
sends
a
signal
of
like
we're,
not
we're
ignoring
your
report.
So
I
think
we
should
maybe
keep
it
open
but
put
Priority
more
evidence
which
is
more
like
we
try
to
reproduce
this.
We
couldn't,
but
if
you're
willing
to
prove
this
to
us,
we
would
love
to
fix
it.
I
Okay,
yeah-
that's
fine
I
can
also
put
I,
can
also
put
a
comment
in
there
as
such.
Thanks.
A
David
we
adopt
ours,
so
I
think
what
we
can
do
is
you
know:
I'll
take
a
look
at
the
the
stable
async
and,
if
there's
anything,
I
can
update
that,
but
I
think
that's
it.
For
today's
meeting.