►
From YouTube: SIG Instrumentation 20211209
Description
SIG Instrumentation Bi-Weekly Meeting Dec 9th 2021
A
Hello,
everybody
I
have
kicked
off
the
recording
and
today's
edition
of
sig
instrumentation
is
december,
9th
2021.
It
is
a
thursday
we
have
a
bunch
of
stuff
on
the
agenda
today
and
a
lot
of
people
attending
thanks.
So
much
for
joining
us.
A
You
can
follow
along
on
the
agenda,
which
is
both
in
the
invite
for
the
meeting,
and
I've
also
dropped
a
link
in
the
chat.
So
I
guess
our
agenda
was
supposed
to
start
off
with
announcements
and
action
item
follow-up.
What
did
we
have
for
action
items?
A
I
guess
I
don't
have
any
current
announcements
other
than
124
development
is
open.
Hurrah
and
unfortunately
I
don't
think
that
there's
a
schedule
up
for
124,
so
I
can't
tell
you
dates
for
anything.
I
can
go
and
check
really
quickly
and
the
release
repo
yeah.
They
don't
have
anything
up
for
124.
I
don't
think
they
have
a
release
team
assembled.
Yet
so
we
don't
have
a
schedule
yet.
So
I
can't
give
you
any
announcements.
There.
B
I
know
that
jordan
said
that,
like
there
is
a
calendar
for
the
from
the
release
team
and
they
mentioned
that
the
announcement
fees
will
be
late
january.
I
guess
yeah.
A
That's
what
I
would
expect
as
well.
They
did
promise
us
that
they
were
supposed
to
get
a
schedule
out
before
the
release
actually
started
and
technically
like
we're
in
the
124
cycle.
Now
like
they
cut
the
alpha
yesterday.
So
I'm
not
really
sure
what's
going
on
with
that,
but
in
any
case,
yes,
I
think
that
late
january
is
a
fair
assumption.
A
Technically,
I
think
that
the
next
release
cycle
doesn't
actually
open
until
january
because
they
want
to
have
some
time
between
releases
and
the
last
couple
of
weeks
of
the
year,
people
tend
to
shut
down
anyways,
so
so
that
should
be
coming
but
yeah
we
don't
have
any
dates
yet
action
items
from
our
last
meeting.
A
I
guess
we
had
a
follow-up
on
the
resource
metrics
api,
but
we
have
that
on
the
agenda.
So,
oh
and
I
was
supposed
to
remove
an
outdated
readme,
which
I
think
happened
so
we're
all
good.
A
So,
yes,
next
topic,
selecting
a
new
tech
lead,
so
at
our
last
meeting
frederick
announced
that
he
will
be
stepping
down.
He
meant
to
send
an
email,
but
that
email
has
not
been
sent
with
that
announcement.
So
the
rest
of
us
leads
now
take
on
the
responsibility
to
do
that
and
to
announce
a
new
tech
lead
to
replace
frederick.
A
So,
if
you're
interested
in
doing
that,
please
let
us
know
we
will
probably
select
someone
and
then
there
will
be
a
lazy
consensus
period.
So
to
give
folks
a
heads
up
about
that.
I
don't
think
that
we
can
announce
anyone
today
because
han
isn't
here.
So
consider
this,
I
guess
more
notice,
but
I
will
try
to
make
sure
that
we
have
at
least
a
written
email
go
out,
because
that
should
have
happened
already
so
action
written
email
on
rhetoric,
archer,
okay,
I
will
at
least
make
sure
it
happens.
A
Okay
and
then
I
guess
we
have
a
few
things
on
the
agenda
that
are
kind
of.
I
guess
either
related
to
kept
discussion
or
part
of
kept
discussion.
So
let
me
just
I
realize
that
the
doc
isn't
in
the
notes.
So
let
me
link
that
and
we
can
get
into
that.
A
Nods:
okay,
let
me
just
pop
this
out.
A
Okay,
so
this
is
the
dock
which
I
linked
here
and
it's
mostly
just
sort
of
a
copy
paste
carry
forward
from
123..
A
So
this
is
the
list
plus
there
might
be
some
a
few
new
things
on
this
list.
We
want
to
make
sure
all
the
new
things
are
on
this
list.
So
do
let
us
know
if
you
missed
anything,
but
this
is.
This
is
basically
the
list
of
stuff
from
the
last
release,
and
oh,
this
is
now
out.
So
I
guess
we
want
to
see
if
we
want
to
start
graduating
more
things
and
it
would
be.
This
is
a
preliminary
like.
A
Obviously
the
release
cycle
hasn't
opened
up
yet,
but
it's
good
to
have
at
least
sort
of
advanced
notice
and
make
sure
that
anything
we
want
to
plan
to
do
this
cycle.
We
have
resourcing
on
and
sort
of
everything's
in
place,
because
there
are
potentially
a
lot
of
things
that
we
could
move
forward.
So
first
item
merrick,
deprecating,
k-log,
specific
flags.
I
think
that
one
went
alpha
and
123.
do
you
want
to
move
it
to
beta
and
124.
A
We
don't
have
to
decide
today
it's
just
kind
of
a
havoc
cooking,
so.
A
So
apparently,
our
goal
for
beta
are
go.
Runner
project
is
well
maintained
and
documented
documentation
on
migrating
off
k.
Log
flags
is
publicly
available
and
the
flags
are
marked
as
deprecated.
D
Yeah,
patrick,
do
you
have
any
so
I
can
start
at
the
bottom.
The
k
log
flags
are
marked
as
deprecated.
Okay,
that's
good.
We
already
did
that
123.,
okay,
because.
D
A
Okay,
then,
that
sounds
like
a
pretty
easy
thing
for
us
to
bring
to
beta
it
shouldn't
require
too
many
resources,
and
so
I
guess
we
just
need
to
ensure
that
the
cap
is
updated.
A
D
There
was
one
beta
criteria
where
I'm
not
sure
goal
runner.
Oh
well,
go
runner
itself.
The
tool
is
fine.
Wasn't
there
some
other
tools,
but
that
we
wanted
no.
A
A
Cool,
so
it's
good
to
confirm
that
okay,
so
metric
cardinality
enforcement,
do
we
have
anybody
who's
available
to
pick
this
up.
B
I
can
look
into
it
like.
I
don't
think
there
is
much
to
do
last
time.
I
checked
okay,
because
I
I
think
the
only
item
from
alpha
to
beta
was
to
support
like
giving
a
file
as
a
parameter
for
this
particular
like
feature,
but
I'm
not
sure
like,
because
it's
not
yet
documented.
So
I'm
not
sure
where,
when
we
want
to
do
that.
A
Okay,
well,
you
can
take
a
look
and
we
don't
have
to
decide
now,
but.
C
A
Will
see
this
one
we
might
ga
it.
I
think
I
need
to
reach
out
to
clayton.
I
don't
think,
there's
anything
that
really
needs
to
be
done
at
this
point.
We
may
consider
promoting
some
of
these
metrics
to
stable,
but
I
don't
even
think
that
was
criteria
in
the
caps.
A
A
Yes,
that
was
definitely
similar
to,
I
think,
some
of
the
stuff
that
they
were
interested
in.
I
don't
think
we
have
that
on
our
list,
yet
so
I'll
make
sure
to
add
that
at
the
end,
because
I
saw
you
put
that
on
the
agenda,
so
structured
logging,
I'm
assuming
we're
not
doing
anything
because
it
went
beta
and
123.
So
typically,
we
don't
ga
for
another
release.
A
D
Question
yes,
from
a
technical
point
of
view,
our
procedural
point
of
view:
do
we
need
to
have
it
marked
for
the
release
to
actually
get
features
accepted?
No,
no.
You
just
need.
A
Basically,
the
things
that
you
would
need,
like
specifically
a
cap
like
the
reviewers,
will
take
like
a
closer
look,
is
if
you're,
making
any
changes
and
package
features
like
you're,
changing
the
maturity
of
a
feature
gate,
or
something
like
that
or
if
you're
changing
an
api
state.
A
Those
are
the
sorts
of
things
that
we
would
expect
there
to
be
an
enhancement
for,
but
if
it's
just
like
we're
continuing
this
migration
and
we're
not
planning
on
graduating
this
thing,
then
there's
no
specific
need
to
have.
D
A
A
We
can't
do
any
work
between
releases
and
we
will
have
to
not
graduate
things
when
we're
not
planning
on
graduating
them.
So
I
will
follow
up
with
the
release
team,
yeah.
A
Well,
the
difference
in
that
case
was,
I
think
we
did
want
to
actually
like
declare
it
as
beta
last
release,
because
we
had
like
slipped
over
a
few
times.
A
Yeah,
but
that
was
to
change
the
the
maturity
not
just
like
do
some
interim
work
and
like
in
between
I'm
not
sure
so
anyways
I'll
follow
up
with
them
api
server
tracing
david.
How
are
we
going
to
beta.
E
I
think
we
should,
I
think
we
we
just
need
to
have
a
release
of
fcd
in
order
to
update
the
dependencies
to
stable,
which
is
one
of
the
requirements,
and
I
also
need
to
do
some
work,
at
least
at
google,
to
shop
it
around
and
get
feedback
on
it.
E
E
A
That
would
be
awesome,
but
yes,.
A
So
I've
taken
off
the
question
mark
in
the
hopes
that
we
are
certain
on
that
so
redesign
event
api.
A
I
asked
around
a
little
bit
about
this
one
because,
like
the
api
is
graduated,
which
is
good,
the
feature
is
not
technically
marked
as
stable,
because
we
still
have
a
lot
of
stuff
on
the
old
api
and
like
in
theory,
we're
never
going
to
get
rid
of
it,
but
jordan
suggested
I
maybe
add
a
to
update
the
cap
to
say
that
the
goal
is
that
we
don't
want
any
entry
components
using
it.
So
I
don't
think
we
have
anyone
to
own
this
one
still.
A
I
think
it
was
owned
by
chelsea,
but
I
don't
know
that.
There's
anyone
staff
to
work
on
this,
so
should
we.
E
A
Yeah,
that's
part
of
the
problem
too,
like
we
went
to
all
this
trouble
of
making
this
new
api
that
has
a
lot
less
performance,
scale,
issues
and
whatnot
and
then
nothing's
using
it
so
great,
because
I
think
that
will
help
with
some
scaling
things:
okay,
kubernetes,
dynamic
components,
log
sanitization!
A
I
would
like
to
deprecate
this.
I
know
that
I
think
some
people
at
google
are
using
this,
but
as
far
as
I'm
aware,
it's
not
being
used
anywhere
else.
We
had
a
bunch
of
reservations
when
we
first
introduced
the
feature
and
as
far
as
I
can
tell
like,
I've
tried
to
email,
the
owner
and
get
feedback,
and
I
just
I
haven't
heard
anything.
So
I
would
like
to
announce
this
for
deprecation.
A
A
I
will
mark
this
as
deprecated.
I
will
send
the
email
I
do
have
to
touch
base
with
the
sig
security
people,
because
we
technically
moved
this
cap
to
them,
because
it
was
one
of
a
couple
of
like
security
related
caps.
We
tried
to
hand
over
to
them,
but
it
turns
out.
There
is
some
code
that
sig
instrumentation
actually
owns
for
this
one,
and
I
know
they
were
kind
of
like
well.
What
do
we
do
with
this?
A
So
we
might
offer
to
take
it
back,
and
I
need
to
follow
up
with
them
on
that.
So.
A
Cool
yeah,
because
this
one
has
been
in
like
perma
alpha
for
like
three
plus
releases,
so
I
think
it's
it's
time
to
wish
it
good
night.
A
I
haven't
heard
anything
from
han
about
metric
stability
classes.
I
know
that
there
was
discussion
about
targeting
this
for
124,
but
has
anyone
seen
a
cap.
A
I
don't
think
he's
had
the
bandwidth
to
work
on
this
also,
let
me
start
tagging
people,
because
I
realized
I'm
not
doing
that.
A
I
don't
know
if
this
helps
anybody
else,
but
I
find
it
super
useful
to
get
an
email
being
like
you
need
to
do
all
these
things
now.
A
Okay,
does
anybody
want
to
look
into
this
or
pick
this
up
or
pin
ping
han?
Should
I
just
tag
han.
A
Okay
and
then
I
think
this
one
is
schwei
graduating
the
metrics
api
to
ga,
so
I
put
sort
of
like
beta
and
ga
in
question
marks.
So
I
don't
think
we
had
a
cap
for
this
one
before,
but
it's
already
beta
right.
A
But
there's
no
associated
cap,
so
I
guess:
do
you
want
to
go
ahead
and
just
like
write
like
us,
you
have
like
a
skeleton
cap
for
like
it
doesn't
have
to
have
all
the
details
or
whatever
also
there
might
be
design
proposals
if
it's
five
years
old
in
the
archive
design
proposal
repo.
So
you
can
totally
just
like
reference
that
as
well
and
don't
have
to
let
go
and
rewrite
everything.
F
A
A
On
the
yes,
I'm
on
the
production,
readiness
team,
so
I
can
help
you
with
that.
We
have
a
little
bit
of
documentation
on
it,
but
there's
like
some
comments
in
line,
if
you
actually
edit
the
markdown,
what's
your
email,
so
I
can
assign
you
to
this.
B
If
I
can
just
like
one
at
one
thing
regarding
that
topic,
so
I
got
like
three
metrics
api
that
are
still
not
graduated
to
gi,
even
though
they
are
in
there
like
for
a
while
now
so
the
matrix
api.
We
have
also
the
custom
matrix
on
the
external
matrix
one
and
they
all
suffer
from
the
same
problem
that
they've
been
beta
for
like
too
long.
B
So
I'm
wondering
like
if
we
should
also
like
do
the
same
for
the
two
other
resource
metrics
api.
A
I
think
that
makes
sense.
I
have
tagged
you
as
a
reviewer
and
perhaps
once
we
have
a
design
doc
up.
We
can
discuss.
A
Cool
and
then
I
know
that
sally
had
wanted
to
propose
this
cubelet
open,
telemetry
chasing
for
alpha
this
release
and
she
had
started
doing
a
bunch
of
work
and
short
presentation
previously.
Does
anybody
have
any
concerns
about
pulling
this
into
the
release.
A
It's
up
here
I'll,
make
sure
you
get
your
email.
A
Okay,
look
at
all
those
action
items.
I
feel
great
about
this,
we're
so
prepared,
and
then
patrick,
you
had
one
more
thing
that
I
think
you
maybe
want
to
add.
D
Yeah,
that's
the
contextual
logging,
so
just
to
give
everyone
some
feedback.
Some
some
background.
This
has
been
in
very
early
drafts
of
structured
logging
and
then
was
pulled
out
as
being
too
complicated
as
a
first
step,
but
I
think
we
should
bring
it
back
now,
because
there
are
use
cases
in
six
scheduling,
for
example,
where
they
ask
for
features
that
exactly
depend
on
this
functionality.
D
Being
able
to
attach
values
to
all
lock.
Calls
that
happen
in
a
certain
call
chain
is
one
example.
My
my
personal
pet
peeve
is
automated
testing
and
having
a
test
output
associated
with
individual
failed
test
cases.
D
How
we
want
to
make
a
logger
available
in
a
certain
call
chain
without
falling
back
to
a
global
logger,
which
is
what
what
we
currently
do
with
k-log,
and
that
is
that
is
mostly
the
technical
question
that
we
need
to
solve,
where
we
need
to
get
agreement
among
kubernetes
developers,
because
the
cap
proposes
to
use
for
context
the
go
contact
structure
and
attach
a
logger
to
it
for
those
functions.
That
already
have
a
context-
and
I
know
that
some
people
have
concerns
about
that.
A
A
That
makes
sense-
maybe
let's
get
like
a
an
actual
presentation
to
the
full
sig
in
the
new
year
and
we
can
discuss
more
than
I'm
gonna
just
put
this
sort
of
as
question
mark
for
now,
but
I
think
we
need
more
information
and
yeah.
We
can
discuss
further.
D
Okay,
so
right
now
I've
been
discussing
it
with
six
scheduling
and
with
the
folks
who
had
open
use
cases.
Four
ways
and
I've
been
pointing
them
at
the
camp,
and
I
know
that
six
scheduling
aldo
is
fully
behind
that
and
and
once
this
done,
because
they
they
want.
They
have
an
open
issue
about
enhancing
logging
in
cube
scheduler,
and
this
is
a
major
part
of
their
solution
that
they
they
want
to
do.
D
And
the
other,
the
other
thing
is
about
structured
logging.
One
observation
is
that
we
are
basically
now
touching
large
parts
of
a
code
base
to
replace
k
log
in
for
f
with
k
log
in
for
s
and
if
we
then
go
in
and
later
do
the
same
thing
with
contextual
logging,
it
touches
the
same
files
a
second
time
and
we
need
to
for
reviewers
again
when
we
make
that
second
change.
D
My
thinking
also
was
that
if
we
can
get
contextual
logging
infrastructure
in
place
in
124
and
perhaps
have
cube
scheduler
as
a
good
test
case
and
and
proof
of
concept
in
practice
that
this
works,
then
future
conversions
can
directly
do
contextual
logging
and
structured
logging
in
one
pr
for
for
certain
parts
of
the
code
base
and
overall
we
will
make
more
progress
towards
our
ultimate
goals.
If
we
combine
those.
A
That
sounds
great
to
me.
We
have
one
minute
left.
So,
let's
come
back
to
the
rest
of
our
agenda,
so
schwei
feel
free
to
ping
me
about
the
slo
stuff
offline.
I
can
help
you
with
that.
I
don't
think
we
have
francesco
today.
So
we
can
push
this
into
the
new
year.
Are
there
anything
else
that
you
would
like
to
discuss,
because
we
will
be
definitely
canceling
our
final
two
scheduled
meetings
of
the
year
and
I'm
not
sure,
what's
going
on
triage
next
week
as
there's
also
the
conflicting
contributor
celebration.
A
A
Oh,
that's
an
announcement.
You
should
register
for
the
contributor
celebration.
A
Sounds
like
no
well,
it
was
wonderful
to
see
you
all.
We
won't
be
meeting
as
a
full
sig
until
the
new
year,
so
if
you're
taking
some
time
off,
I
hope
you
have
a
restful
and
relaxing
break
and
see
you
all
in
2022
cheers.