►
From YouTube: SIG Instrumentation 20210708
Description
SIG Instrumentation Bi-Weekly Meeting July 8th 2021
A
A
A
Okay,
I
can
try
to
repeat
what
you
just
said:
you
might
want
to
try
a
headset,
so
I
think
han
just
attempted
to
say
that
there
is
a
conflict
right
now
with
the
current
triage
session,
which
is
9
a.m.
Pacific
time
on
wednesdays,
I
believe
it
conflicts
with
an
open
telemetry
meeting
or
something
like
that,
something
upstream
that
is
not
a
cube
meeting,
and
so
we
are
proposing
to
move
it
to
the
same
slot
as
this
meeting.
A
So
this
is
9
30,
pacific
time
on
thursdays
and
so
right
now
we
currently
are
alternating
meeting
wednesdays
and
thursdays
and
instead,
we'll
just
be
in
the
same
slot
every
week,
it'll
just
be
alternating
triage
in
our
normal
meeting
and
we
wanted
to
get
folks
feedback.
I
think
han
updated
the
invite
already,
but
wanted
to
make
sure
that
that
still
worked
for
everyone.
B
A
A
Okay,
so
then,
I
don't
think
we
have
anything
else
on
that,
one
if
you
do
have
any
concerns-
and
you
don't
want
to
like
mention
it
on
the
recorded
meeting-
feel
free
to
reach
out
to
any
of
us
by
a
pm.
A
Let's
see
reviewers
updates.
This
was
also,
I
think,
han
item.
I
think
we're
sharing
this
one,
but
I
can
kick
it
off
so
first
off
congrats
to
lily
for
becoming
an
approver
in
kubernetes
kubernetes,
and
I
think
we
want
to
like
add
more
reviewers.
Was
that
the
gist
of
this
item
han.
A
B
A
B
A
B
B
B
So,
who
here.
B
A
A
Me
pull
up
I'm
going
to
share
my
screen.
Let's
take
a
look
at
the
list
because
I
think
a
lot
of
people
here
are
not
currently
on
the
review
list
and
then
I
think
we
also
have
some
people
on
the
list
who
we
have
sort
of
kept
on,
but
maybe
aren't
active
reviewers
and
we
shouldn't.
So.
Can
everyone
see
my
screen
yep
great?
So
here
is
the
owners
aliases.
A
Let
me
find
instrumentation,
so
here's
our
current
list
of
reviewers,
which
is
mostly
like
the
current
list
of
approvers
plus
a
few
other
people.
So
I
think
we
have
coffee
pack
andy,
who
is
not
active
sergio
and
really
like,
not
a
lot
of
other
folks
who
are
getting
pinged
first
instrumentation
reviews.
So
we
would
like
to
add
people.
Let
me
see
if
I
can
find
the
reviewer
there's
the
contributor
ladder
thing
there.
A
We
go
community
membership
that
will
take
me
to
approver,
but
so
for
those
of
you
who
are
not
currently
familiar.
Kubernetes
has
a
community
membership
ladder,
and
so
there's
this
sort
of
process
going
from
like
member
to
reviewer
to
approver
to
sub
project
owner
and
the
these
like
reviewer
to
sub
project
owner
steps
are
mostly
reflected
in
what
are
called
owner's
files.
So
this
is
an
example
of
an
owner's
aliases
file.
Let
me
go
find
one
of
our
owner's
files
in
kubernetes,
kubernetes,
probably
from
component
base,
or
something
like
that.
A
A
Who
can
review
them
and,
like
you
know,
a
label
gets
auto
applied,
so
we
know
that
it
applies
to
us,
and
so
this
list
is
pulled
from
this
list
over
here
and
the
owner's
aliases
file
at
the
root
of
the
repo
and
right
now
we
don't
have
a
lot
of
people
in
that
list
outside
of
the
current
approvers
list,
and
some
of
them
are
also
not
super
active,
so
we
want
to
add
more
people
to
that
list.
So
what
does
it
take
to
get
there?
A
So,
in
order
to
become
an
a
reviewer,
you
must
first
be
a
member.
So
if
you're
not
already
a
kubernetes
org
member,
you
need
to
go
through
and
meet
all
the
requirements
and
become
one.
So
if
you
are
missing
sponsors
and
whatnot,
you
can
reach
out
to
us
and
we
will
make
sure
that
you
know
you
get
your
plus
ones
for
your
sponsorship
and
then,
after
three
months
you
will
be
able
to
apply
to
become
a
reviewer.
A
A
So
that
is
not
a
really
reasonable
like
guideline
for
us,
so
I
think,
and
other
approvers
feel
free
to
jump
in
if
you
disagree,
but
typically
like
you
know,
if
we
see
like
five
high
quality
reviews,
we're
not
gonna
like
wait
for
you
to
get
20
or
something
like
that.
A
We're
happy
to
add
you
to
become
a
reviewer
right
at
that
point,
because
we
don't
have
nearly
the
pr
traffic
that,
like
I
think
these
guidelines
were
written
in
mind
for
and
then,
if
you've
been
doing
a
great
job
as
a
reviewer,
then
eventually
you
can
become
an
approver
after
three
months.
So
I
just
wanted
to
make
sure.
Let
me
copy
a
link
to
this
file
into
our
notes,
so
people
can
take
a
look
at
it
if
they
haven't
looked
at
this
file
before.
Are
there
any
questions.
D
A
I
think
that,
basically
anybody
who
wants
to
be
active
in
kubernetes
kubernetes
and
wants
their
name
on
an
owner's
file
like
here
is
a
great
opportunity.
We
would
like
to
add
people,
and
so,
if
you
are
doing
the
work,
we
will
add
you
as
a
reviewer
is,
I
think,
our
tldr.
A
So
these
are,
you
know,
project-wide
standards.
So
if
you
have
any
questions,
now
is
the
time
to
ask.
C
I
have
one
question
in
that
regard:
do
we
need
to
like,
if
you
want
to
be
a
regular,
do
we
need
to
create
the
pr
ourself,
or
should
we
ask
first
an
approver-
or
I
don't
know
in
the
sig
meeting,
to
do
that.
A
No,
you
don't
need
to
create
the
pr's
yourself.
In
fact,
this
is
all
about
just
reviewing.
So
like
you
can't
review
your
own
pr,
you
have
to
review
other
people's
pr's,
so
you
can
become
a
reviewer
without
having
contributed
to
this
code
base.
You
simply
need
to
make
sure
that
you
are
doing
reviews
and
then
we
will
check
to
make
sure
that
the
reviews
are
high
quality.
A
So
you
know
if
you're,
just
like
throwing
lgtm
on
everything
without
making
any
comments
or,
like
you
know,
sort
of
deeply
looking
at
that
code,
then,
maybe
like
that's
not
enough
to
become
a
reviewer,
you
need
to
actually
go
through
and
review
that
code
and
you
may
ask
well
how
do
I?
How
do
I
do
a
good
review?
We
have
a
guide,
and
so
this
is
linked
in
the
requirements
for
reviewers.
A
There
is
this
little
code
reviews
doc,
which
I
will
also
paste
into
our
meeting
minutes,
and
these
talk
about,
like
you,
know,
things
that
you
should
be
looking
for
in
reviews
and
what
the
expectations
of
reviewers
are
and,
like
you
know,
sort
of
slas
and
that
kind
of
thing.
A
Yeah
so
typically,
typically,
the
person
who
wants
to
become
the
reviewer
is
the
one
that
will
create
the
pr,
if,
like
we
as
sig,
leads
notice
that
someone's
been
doing
a
lot
of
work
but
hasn't
brought
themselves
forward
like
we
are
also
allowed
to
nominate
somebody
assuming
that
they
are
okay
with
that,
and
so
basically
you
just
have
to
like
create
a
pr
with
like
this
checklist,
and
you
know
like
check
off
each
box,
you
must
be
sponsored
by
a
sub
project
approver.
A
A
That's
normally
how
we,
how
we
add
a
new,
both
reviewer
and
later,
if
you
want
to
become
an
approver
same
similar
thing
for
approver,
just
the
bar
is
a
little
bit
higher.
C
A
So
han's
audio
is
kind
of
breaking
up,
so
I'll
repeat
it
for
the
recording.
If
you
want
to
become
a
reviewer,
it
is
very
highly
recommended
that
you
come
to
our
bi-weekly
triage
meetings,
because
we
don't
have
a
lot
of
traffic
in
terms
of
like
pr's
that
we
need
to
review.
A
There's
not
necessarily
a
lot
of
stuff
for
you
to
like
sort
of
jump
onto
and
add
reviews
to
so
make
sure
that
you're
attending
those
meetings-
and
we
will
help
find
you
things
to
review
so
that
you
can
be
the
primary
reviewer
on
these
things
and
like
work
towards
that
goal.
A
Any
other
questions
that
sort
of
answered
the
question
that
I
was
going
to
ask,
which
is:
how
do
I
find
things
to
review
and
I
think
the
answer
is
you
could
look
at
prs
that
have
the
sig
instrumentation
label
on
them
in
kubernetes
kubernetes,
but
if
you're
not
really
sure
where
to
start
jumping
in
on
that,
not
feeling
super
confident
come
to
our
triage
meetings.
We
will
help
find
you
things
to
work
on,
just
show
up
and
be
like
hey.
I
want
to
review
things
and
we
are
very
happy
to
share
work.
A
D
Yeah,
so
the
word
that
we
have
planned,
I
think
that
one
of
the
major
points
to
graduate
the
structural
beta
was
providing
similar
level
of
features
and
as
k-log.
So
at
this
moment
we.
D
D
We
will
not
finish
graduate
the
cup
to
beta,
but
for
123
we
plan
to
either
to
to
have
the
work
for
the
and
prepare
the
cap
and
start
working
on
the
other
cap
or
decide
to
cut
the
scope
to
just
get
us
there.
D
D
A
Yeah,
I
think
we
can
move
it
to
the
working
group.
I'm
happy
to
like
approve
that
if
we
want
to
like
move
it
to
its
own
directory
or
whatever.
C
E
D
So
that
that
was
discussed,
I
think
on
121,
if
we
should
launch
graduate
studies
looking
to
beta
with
cube,
let's
migrate
it.
The
problem
back,
then,
was
that
the
the
original
cap
proposes
two
separate
things.
One
is
the
migration
itself
and
second,
the
the
json
as
an
alternative
format
to
really
consume
the
deluxe
and
problem
is
that
we
are
more
further
with
migration,
because
lots
of
people
are
contributing
and
it's
easier
and
it's
easier
to
to
do
small
changes.
D
D
That
that's
only
my
fault
that
I
don't
want
to
block
it
further,
because
I
think
I
I
would
also
want
to
I'm
planning
to
talk
with
spectra
if
and
the
second
lead
to
what
what
we
are
trying
to
actually
to
discuss
what
we
are
trying
to
achieve
and
like
what
would?
What
really
that
data
means
and
if
the
mind
definition
of
beta
for
json
should
be
maybe
separate
than
migration
for,
and
migration
of
locks.
E
Makes
sense,
thank
you.
So
if
I
understood
correctly,
that,
like
cutting
scope,
would
mean
potential,
none
parity
with
k-log.
D
E
That
seems
to
be
a
theme
for
us,
but
I
think
I
think
that's
not
not
all
too
surprising,
giving
given
the
like
cross
sick
nature
of
most
of
the
work
that
we
do
so,
but
at
the
very
least,
I'm
happy
that
we
found
this
out
and
I'm
actually
I'd
be
quite
happy
with
dropping
some
k-log
features.
D
Interesting,
if
you
have
any
ideas
about
alternatives,
because
now
we
are
like
we
we
are
we.
The
cap
is
mostly
about
giving
people
alternatives.
So
if
you
think
you
like
you,
have
an
ideas
about
how
people
can
do
it
better
without
those
skeletal
features
that
we
are
removing
them
would
be
great.
Currently,
we
are
yeah,
depending
on
some
some
tool
that
was
created
randomly
without
intervention,
so
it
would
be
great
if,
if
we
have
some
better
long-term
solution,
yeah.
A
Pond
you
guys
are
like
do
you
want
to
say
it
in
chat?
Maybe
oh
about
working
groups.
I
don't
know
about
like
so
the
working
groups
aren't
supposed
to
own
code.
I
don't
know
if
that
means
that
they're
not
allowed
to
own
caps.
B
A
Yeah
they're
they're,
temporary,
but
like
caps,
are
also
temporary
right,
like
they
start,
and
then
they
end.
So
I
mean
I
think
that
would
probably
be
a
question
for
the
the
enhancements
team.
I
think
that
we
haven't
had
a
like
it's
interesting.
The
way
that
working
groups
have
worked
out
in
the
past
we
haven't
actually
had
with
the
cap
that,
like
you,
know,
sort
of
the
the
the
working
group
got
disbanded
when
they
kept
merged
kind
of
thing.
A
I
think
we've
been
having
some
difficulties
with
that
sort
of
thing
as
a
project
with
like
the
scope
being
limited
and
working
groups,
actually
wrapping
up
when
they
finish
a
thing
so
yeah.
I
think
this
is
something
that
we
could
just
ask
them.
A
F
Thanks
we'll
try
and
get
there's
one
pr
that
I
really
would
like
to
get
in,
so
I
suppose
I'll
take
this
opportunity
to
post
it
here,
but
yeah.
We
have
the
primary
pieces
in
the
big
initial
pr.
Obviously,
and
then
a
couple
of
follow-ups
have
made
it
in
the
last
one
that
I
would
really
like
to
get
in
is
the
piece
that
connects
it
to
xcd
so
that
we
can
join
this
work
with
the
stuff
lily
already
did.
F
Actually,
if
you
look
at
that
to
get
everyone
to
click
on
the
link,
I
sent
there's
a
there's
a
picture
of
a
trace,
including
all
three
which
is
kind
of
exciting
nice.
It.
It
isn't
stackdriver
so
sorry
for
the
plot,
but
it
was
just
the
easiest
for
me.
E
A
E
Recording
elana:
did
you
see
my
comment
on
the
c
advisor
metrics:
rework
p
cap
that
was
merged.
A
E
Yes,
so
I
I
I
I
missed
two
like
comment
in
time.
I
commented
on
the
pr
after
it
was
merged.
It
was
one
of
the
caps.
I
think,
like
one
thing
that
was
unclear
to
me
was
what
this
means
for
like
c
groups
v2,
because
a
bunch
of
metrics
stuff
changes
from
v1
to
b2-
and
I
remember
this
was
being
discussed
in
the
like
general
c
groups.
V2.
A
That's
a
very
good,
so
actually,
I
think
hopefully
this
is
making
that
a
lot
simpler
and
easier
and
nicer,
because
right
now
we
kind
of
have
this
in
direction
of
like
cubelet
talks
to
c
advisor
to
get
the
metrics
which
talks
to
the
container
runtime
and
we
are
moving
instead
to
cubelet
directly
like
does
an
rpc
to
the
container
runtime
being
like
yo,
give
me
metrics,
and
so
it
basically
completely
delegates
it
to
the
cri
in
a
bunch
of
cases
rather
than
having
to
go
through
the
see
advisor
middleman,
which
also
means
that
we
don't
have
to
worry
about.
A
E
Right,
it's
more
of
a
it's
more
of
a
compatibility
problem,
because
almost
literally
everyone,
I
know
relies
on
these
metrics
and,
like
c
groups,
b2
doesn't
track.
I
don't
know
what
it
was
anymore,
whether
it
was
network
stats
or
cpu
stats.
I'm
not
I'm
not
sure
anymore
what
it
was,
but
the
the
recommendation
was
to
use
like
ebpf
to
collect
this
type
of
data
in
the
future,
and
I
I
feel
like
we
should
have
at
least
let
the
the
container
runtimes
know
well
in
advance
that
this
is
coming.
If
we're
doing
this.
A
It
because,
like
all
of
the
container
run
times
once
the
changes
get
made
to
the
cri,
they
have
to
go
and
implement
them.
So
it's
being
made
in
in
consult
with
them.
I
would
say
like
on
the
c
groups:
v2
side
of
things
c
group's
v2
is
still
an
alpha
and
so
like,
I
think,
until
that
graduates
to
beta,
we
don't
necessarily
have
like
the
big
same
stability
guarantees
that
we
might
need
there,
and
I
think
the
other
thing
too
is
like
until
we
start
telling
people
like.
A
Oh,
you
should
migrate
onto
c
groups
v2.
I
don't
think
that
that's
ever
in
the
short
term
going
to
be
a
thing
like
getting
rid
of
cruise
v1
support
in
favor
of
sea
groups,
v2
support,
it
would
be
nice,
but
like
that's
a
very
down
the
road
thing,
because
a
lot
of
people
are
running
with
kernels,
that
don't
support
c
groups
v2
and
they
are
supported
kernels
in
production.
A
So
until
we
kind
of
like
until
that
ship
has
sailed
or
we're
just
like.
Oh
yeah,
secretly
one
thing
of
the
past,
I
don't
think
that's
gonna
happen
in
the
short
term.
So
right
now
anything
in
node,
that's
touching!
The
cri
has
to
support
both
secrets,
v1
and
c
groups
v2.
So
I
think
that
it
will.
B
A
Fine,
I
I
if,
if
we
like,
you
know
if
we
get
to
the
point
where
it's
like,
okay,
we're
moving
to
cruise
v2
and
this
is
a
blocker,
I'm
sure
that
we
will
find
some
sort
of
solution
for
that.
Definitely
I
looked
at
the
pr
that
that
peter
was
working
on
peter
hunt
is
currently
the
one
who
owns
this
cap
and
it
had
a
bunch
of
network
stats
stuff
in
there.
So
I
it's
certainly
part
of
the
specification.
E
Yeah
I
just
wanted
to
have
mentioned
it
because
in
the
sea
groups,
v2
cap-
I
I
forget
who
the
person
from
redhead
was
who
was
working
on.
That
said,
we'll
figure
it
out
in
this
other
cap,
which
was
the
sea
advisor
removal
cap
and
then
the
sea
as
a
removal
cap,
we're
now
saying
we'll
figure
it
out
when
the
ziggurats
v2
stuff
happens.
So
it's
kind
of
you
know
I
just
wanted
to
have
mentioned
it
that
this
will
it's.
A
But
that
that
makes
sense
and
yeah.
I
would
just
suggest
to
maybe
comment
this
on,
like
the
pr's
in
flight.
A
Okay,
great
well,
there
you
go,
do
you
have
them,
so
we
can
add
them
to
the
meeting
minutes.
A
Okay,
we
are
at.