►
From YouTube: SIG Instrumentation 20210218
Description
SIG Instrumentation Meeting: Feb 18th 2021
B
Yeah
I
got
it
okay,
so
today
is
february
18th.
This
is
the
bi-weekly
sig
instrumentation
meeting
yeah.
We
just
did
intros,
so
we
did
not
record
the
intros
so
we're
starting
10
minutes
late.
But
elena
do
you
want
to
do
you
want
to
kick
off
the
agenda.
A
Yeah
we
had
a
short
agenda
today,
so
I
figured
we
could
start
with
intros.
I
don't
know
before
I
jump
into
things:
there's
a
dot
dot
dot
on
the
agenda
for
any
other
business.
I
just
want
to
remind
people,
you
know
if
you've
got
something
you
want
to
discuss,
we've
got
20
minutes.
I
have
two
things
which
are
not
going
to
be
20
minutes,
so
you
will
have
your
chance
to
jump
in
so
think
about
it.
So
reminders
we
are
in
enhancements.
Freeze
now,
code
freeze
is
march
9th.
A
I
am
always
amazed
at
how
little
time
there
is
between
enhancements,
freeze
and
code
freeze.
So
if
you
have
a
cap
that
you're
working
on,
we
really
only
have
like
a
little
over
two
weeks
to
get
like.
A
We
only
have
two
full
work
weeks
left
in
the
release
cycle
to
get
all
this
feature,
work
in,
which
is
really
not
that
much
so
just
a
reminder,
you
know
if
you
have
a
cap,
if
you're
trying
to
graduate
something
make
sure
you
take
a
look
at
it,
you
know
who
you
are
so
any
questions
about
that
or
deadlines.
All
of
the
deadlines
are
in
the
agenda
so
march.
A
Great
okay,
no
questions
about
that.
I
don't
know
deadlines,
pretty
straightforward
annual
owner's
files,
org
cleanup,
so
one
of
the
things
that
we
are
doing
right
now
as
chairs
and
leads
primarily
driven
by
the
chairs,
but
also
with
participation
from
the
tech
leads,
is
our
annual
report.
And
so
the
annual
reports
are
due.
A
Basically
at
the
same
time
as
code
freeze,
I
think
march,
8th
and
so
han,
and
I
have
been
working
on
the
checklist
that
we
need
to
fill
out
for
the
annual
report
and
as
part
of
that,
one
of
the
things
that
we
need
to
do
is
owners
files
clean
up
and
making
sure
that,
like
you
know,
our
list
of
sig
members
is
up
to
date
and
whatnot.
So
I
have
already
gone
ahead
and
submitted
some
pr's
to
remove
inactive
people.
A
So
I
basically
took
a
look
at
like
you
know:
have
these
people
contributed
anything
to
cube
in
the
last
year
and
if
no
or
if
under
a
certain
threshold,
I
removed
them
from
both
our
like
reviewers
list
and
also
from
the
sig
org
list,
and
I
also
consolidated
we
had
like,
I
think,
eight
teams
or
something
on
github.
Now
we
just
have
three
so
there's
a
leads
team.
A
There's
an
approvers
team
and
there
is
a
members
team,
so
I've
reached
out
to
a
few
of
you
offline,
and
this
will
be
a
little
bit
more
difficult
for
the
people
who
are
not
yet
kubernetes
members.
But
if
you're,
not
a
kubernetes
member
and
I
reached
out
to
you
being
like
hey
you're,
coming
to
all
of
our
meetings
and
doing
things
and
participating,
please
join
our
sig
as
a
member.
Would
you
like
me
to
add
you?
A
We
can
find
ways
to
help
with
you
know,
finding
new
contributions
to
make
it
whatnot.
I
think
there's
lots
of
work
to
be
done
with
structured
blogging.
For
example,
it
doesn't
have
to
be
prs.
You
can
review
things,
you
can
file
issues.
You
can
comment
on
issues
all
of
those
things
count,
so
you
can,
you
can
make
doc
changes
like
it
doesn't
have
to
be
code,
and
all
of
that
stuff
is
valuable.
A
You
can
help
with
you
know
kubecon
stuff,
although
that's
kind
of
not
in
person
right
now,
all
those
things
count.
So
if
you
have
any
questions
about
like
where
to.
A
How
to
get
involved
that
kind
of
thing
very
happy
to
help
out
with
giving
pointers,
and
I
think,
because
we've
removed
some
of
our
inactive
reviewers.
We
don't
have
a
lot
of
people
on
the
instrumentation
review
list.
So
if
you'd
like
to
become
a
reviewer,
it
does
involve
some
responsibility.
A
I
would
be
very
happy
to
walk
you
through
that
and
talk
about.
You
know
like
what
the
requirements
are
and
in
order
to
become
a
reviewer,
you
basically
just
need
all
of
the
approvers
to
sign
off
so.
B
A
Must
become
a
kubernetes
org
member
once
you've
been
a
kubernetes
org
member
for
at
least
three
months
you
can
become
a
reviewer
and
then
once
you've
been
a
reviewer
for
at
least
three
months
you
can
become
an
approver.
I
think
we're
okay
in
terms
of
approvers
right
now,
like
we're
not
in
any,
I
think
actually
sig
instrumentation
really
punches
above
its
weight.
A
That's
a
problem
that
I've
seen
in
some
other
groups,
but
not
so
much
here,
but
you
know
having
more
reviewers,
so
we
can,
for
example,
assign
new
things
during
triage
meetings,
and
that
kind
of
thing
is
always
super
great,
so
I
actually
put
together
something
this
is
technically
for
node
not
for
instrumentation,
but
it
is
equally
applicable
for
sort
of
trying
to
learn
how
to
become
a
like
an
org
member,
how
to
do
pr
review,
how
to
approve
things
for
instrumentation,
and
it
has
a
bunch
of
links
to
other
documents.
A
So
I
will
paste
that
in
the
chat
and
also
in
the
agenda-
and
you
can
take
a
look
through
that
it
talks
about
you
know,
for
example,
if
you
want
to
do
a
review,
what
do
these
labels
mean
like?
How
do
I
become
an
org
member?
What
can
org
members
do?
What
can
reviewers
do?
What
can
approvers
do?
So?
A
Everyone's
shaking
their
head
cool,
I
mean,
I
hope,
that's
clear,
there's
like
nothing,
hopefully
too
mind-boggling
there
other
than
come,
do
work
for
us.
We
like
it
when
people
do
work
with
us,
we're
very
friendly.
We
promise
what
else
do
we
have?
I
guess
I
have
the
annual
report.
It's
in
progress.
I
will
send
out
a
draft
copy
to
the
mailing
list
before
which
is
part
of
the
requirements
before
the
deadline
and
before
we
like
officially
submit
it
because
we're
supposed
to
do
it
collaboratively
with
the
sig.
A
A
B
Yeah,
so
we
we
do
have
code
freeze
in
two
weeks
and
we
do
have
a
couple
things
that
we
need
to
get
out
before
then
right,
because
we
still
do
have
a
couple
of
caps
and
clay.
We
have
the
tracing
thing.
We
have
the.
We
have
we're
promoting
a
couple
metrics
to
stable.
A
And
we've
also
got
lots
of
structured
logging
stuff.
I
think
I've
been
behind
on
those,
but
I
think
at
least
one
of
the
big
pr's
for
for
cubelet
merged
and
we
have
the
what
looks
like
good
beginnings
of
a
static
analysis
tool
to
prevent
it
from
reverting
so
yeah.
I
think
we're
making
good
progress
on
do
we
have
like
a
road
map
or
anything
like
tracking
all
of
the
feature
work
that
needs
to
be
done
for
the
stability
stuff.
B
A
B
Have
dele
we
have
the
things
partitioned,
so
I
think
ue
is
gonna.
Work
on
the
the
storage
object,
counts,
yeah,
that's
gonna,
be
exciting,
we're
going
to
yeah
for
stable
metric
and
then
the
request
we're
going
to
we're
going
to
promote
the
request
once,
but
I'm
going
to
I'm
going
to
shed
the.
I
think,
the
client,
client
and
maybe
content
type.
B
That's
you
chen.
If
she
has
a
bandwidth.
If
not,
I
will
do
it.
I
will
talk
to
her
today
and
if
she
doesn't
have
the
bandwidth
I
can
do
it.
I
think
ding
hwa
might
be
interested,
so
I
can
talk
to
dinghua
afterwards.
A
Sounds
good
and
then
merrick?
How
are
you
feeling
about
the
structured
logging
stuff?
I?
I
have
been
kind
of
underwater
with
other
things,
so
I've
been
trying
to
review
things
if
they
come
in
but
and
and
flag
approvers,
because
cubelet
we
can't
approve.
We
need
to
get
node
approvers
for
those
things.
A
D
C
A
D
Laptop
laptop
is
swapping
yeah
yeah,
so
inconsistency
of
keys
in
review
makes
it
slower.
So
definitely
it
would
be
something
for
if
we
would
migrate
more
repo,
that
would
be
more
code
based.
That
would
be
really
useful
to
just
automate
this.
So
just
if
you
have
a
set
of
like
transformations
like
someone
writes
pods
with
the
with
using,
I
know
not
enough
like
with
bad
capitalization,
we
should
be.
We
should
not
review
this.
We
should
have
estimated
analysis
for
that.
D
So
it's
not
a
schema.
It's
a
set
of
like
prefer
like
easy
transformations
that
are
trivial
to
to,
and
should
always
be
correct.
Yeah.
A
That's
something
yeah,
I
I
put
like
slash
conventions
because
I
mean
at
some
point
we
may
end
up
wanting
to
have
like
at
least
like
if
you
use
any
one
of
these
things
they
have
to
conform
to
this
schema
but
like
for
other
things,
there's
more
flexibility.
So
I
don't
know.
I
think
that
that
sounds
totally
reasonable.
I
agree
adidi,
who
wrote
some
of
the
like
analysis,
tooling
already
to
prevent
the
regressions.
A
I
haven't
had
a
chance
to
take
a
look
at
it.
It's
on
my
backlog,
but
I
think
you
gave
it
a
look
and
said
it
looked
good
like
I
can
maybe
ask
her.
C
A
D
D
D
Say
what
we
plan
in
static
analysis?
We
should
have
something
that
explains
why
some
transformation
are
as
they
are
and
like
why
we
are
doing.
Why
are
we
picking
up
those
keys
so
like
it's
not,
I
know
my
opinion
on
how
what
keys
should
be
and
what
we
had
yeah.
We
should
we.
A
Have
like
a
list
or
something
because
that
would
be
very
helpful
if
we
could
just
start
like
if
you
want
to
like
start
a
thread
and
pound
in
sig
instrumentation
or
something
like
that
and
just
talk
about
like
we
want
to
see
these
things.
We
don't
want
to
see
these
things
and
then
maybe
we
can
work
on
updating
the
migration
guide
so
that,
because
that
we
can
update
that
quite
quickly
and
then
people
will
hopefully
conform
to
it.
D
Yeah
so
like
I
was
thinking
that
statistic
that,
like
we
could
start
from
static
another
like
I
could
do
static
analysis
and
cutter.
What
we
have
currently
and
then
yeah
split
them
by
groups,
because
there
are
lots
of
keys
that
we
or
like
we.
There
are
lots
of
low
level
of
things
that
we
didn't
expect
or
we're
not,
and
not
that
obvious.
A
Sorry,
I'm
typing
and
muting,
so
I'm
I'm
putting
an
action
in
the
minutes
for
you
to
do
that.
When
do
we
want
to
sync
up
on
that,
because
I
think,
like
our
next
meeting's,
not
for
two
weeks
and
so
that
might
be
too
long
like
do
we
want
to
sync
up
on
that
asynchronously,
basically,
like
what
deadline
should
I
put
on
this.
D
I
I
I
will
spend
some
time
tomorrow
tomorrow
for
for
sure,
so
I
think
online.
We
could
start
looking
into
monday
like
to
have.
A
Something
so
I'll
put
on
in
the
action
and
the
notes
that
we
will
asynchronously
check
in
on
monday,
which
is
the
22nd.
A
A
Okay,
anything
else
on
any
topics
for
the
current
release
features,
because
I
see
that
we
have
something
that
has
snuck
onto
the
agenda.
E
Yeah
I
mean
this
is
just
kind
of
gauging
what
people
think
I
don't
know.
If
there
are,
there
are
any
other
people
except.
I
know
that
dave
david
lilly
and
I
have
been
joining
the
open,
telemetry,
prometheus
working
group,
and
I
I
felt
like
kubernetes
was
always
a
large
topic
and
at
the
very
least
I
wanted
to
get
everyone's
kind
of
opinion.
E
I
mean
I
I
I
feel
like.
I
know
the
answer
to
it,
but
I
feel
like
we
cannot
ever
not
have
our
like
in-process
metrics
endpoint
right.
That's,
basically,
an
api
that
we're
offering
to
two
users
right
and
part
of
open
telemetry's
discussions
that
are
happening
right
now
is
like
should
kind
of
the
open.
E
Telemetry
libraries
have
an
option
for
something
like
this
as
well
right,
as
opposed
to
always
pushing
to
a
collector,
and
yes,
I
feel
like
we
would
be
like
we
were
essentially
we're
already
even
stabilizing
metrics
within
that
endpoint
right.
So
I
feel
like
we,
we
won't
ever
have
the
option
of
not
having
that,
and
I
think
this
would
be
a
really
good
data
point
to
take
to
the
open,
telemetry
working
group.
So.
E
It's
when
our
triage
meeting
is
every
hour.
I
I
forget:
if
it's
now,
every
week,
maybe
david
remembers
or
if
it
ever,
if
it's
every
other
week
is
every
week.
Okay,.
E
Yeah
yeah
definitely
so
like
what
what
do?
How
do
other
people
feel
about
this?
I
I
imagine
there
will
be
the
inevitable
issue
or
question,
maybe
even
if
we
even
already
have
it
that
people
want
kubernetes
to
implement
open
telemetry
support
right.
B
It's
a
weird
question,
though:
if
you
don't
support
like
being
able
to
have
expose
a
prometheus
endpoint,
then,
basically,
that
decision
is
going
to
inhibit
the
adoption
of
open
telemetry,
because.
E
A
A
C
C
No,
no
one's
speaking,
I
I
can
just
briefly
from
what
I
understand.
The
question
is
more
like:
does
kubernetes
want
to
migrate
to
open
telemetry
format
if
it
won't
be
a
prometheus
or
a
format
that
we
currently
support
right,
because
we
need
we're
basically
always
an
in
process
matrix,
endpoint
right,
and
do
we
agree
on
that?
This
is
essentially,
I
think
what
frederick
wanted
to
say
right.
C
E
Why
that's
why
I
wanted
to
bring
that
up
here
so
that
I
don't
just
like
reflect
my
own
opinion
there,
but
that
we,
as
as
our
as
our
group
kind
of
agree
with
this
right-
and
there
is
kind
of
also
the
the
the
point
in
in
that
group
where
people
are
saying
literally
a
prometheus
target
and
an
open
telemetry
target
should
be
indistinguishable.
E
You
shouldn't
even
be
able
to
tell
that
this
is
an
open,
telemetry
thing
or
a
prometheus
instrumented
target,
and
I
I
very
much
can
sympathize
with
this,
because
that's
exactly
how
the
adoption
would
work
right.
I
think
that's.
C
E
Of
what
you
were
saying
as
well
on
but
yeah
are
there
any
other
opinions
in
the
room,
because
essentially,
what
I
want
to
do
is
take
whatever
feedback
we
have
here
into
that
group.
C
Okay,
I
think
from
cube
state
metrics
side
as
well
like
there's
no
way
we
can
ever
not
do
that
so,
like
like.
First
of
all,
we
we
have
a
hard
time
even
coming
up
with
a
release,
and
I
I
don't
necessarily
see
that
we
can
even
feasibly
do
that,
but
yeah.
I
I
think
that
we,
we
all
sort
of
agree,
but
is
there
anyone
opposed
to
it
essentially
or.
A
Opposed
to
it,
but
time
check
we
are
at
the
top
of
the
hour.
So
I
think
where
can
people
send
feedback?
Slash
participate
since
we
don't
have
any
meeting
left
to
discuss
so.
E
I
I
would
suggest
that
we
actually
bundle
that
up
here,
because
there
are
already
like
every
meeting.
There
are
like
70
participants
because
everybody
feels
like
they
need
to
participate
in
this
thing.
So
I
I
think
it's
good
that,
like
lily
myself
and
david
as
representatives
of
instrumentation
are
there,
I
I'm
not
sure
it
adds
much
value
if
all
11
of
us,
oh.
A
E
A
Okay,
I
think,
probably
in
the
future
it'll
be
good
to
like.
We
can
have
that
as
sort
of
an
open
topic
on
the
agenda,
and
if
you
want
to
bring
anything
we
can
we
can
discuss
and
then
that
way
we
don't
have
to
worry
about
it.
So
that
sounds
good
and
it
sounds
like
yeah.
What
is
the
messaging
that
we
want
to
send
to
the
upstream
one
for
this
particular
question,
just
to
make
sure
that
it's
captured
in
the
minutes.
E
I
mean
it
was.
It
was
more
like
that
there
wasn't
an
a
question
that
that
that
group
was
necessarily
targeting
towards
us.
It
was
more,
I
feel
like
this
was
an
opinion.
That
was
the
case,
but
I
wanted
to
not
just
state
my
own
opinion,
but
like
have
our
group
design
insight
on
this
right,
so
just
what
what
we,
what
we
have
here,
essentially
that
it's
not
feasible
for
the
kubernetes
project
to
do
this
is,
is
enough
of
a
data
point
that
I
can
take
back
to
this
working
group.
A
A
Okay,
great,
we
are,
I
think,
at
times,
so
we
should
probably
wrap
it
up.