►
From YouTube: 2023-03-16 meeting
Description
Open cncf-opentelemetry-meeting-3@cncf.io's Personal Meeting Room
A
A
B
E
C
C
F
Yes,
all
right,
I
have
this
thing
where,
whenever
I
get
off
a
plane,
I
just
get
terribly
sick,
so
go
in
a
cubecon.
It's
gonna
be
interesting.
E
It's
something
I
think
Alex
said
he
would
join.
F
C
E
So
you'll
try
to
make
it
if
you'd
like
to.
C
E
This
Splunk
surprise
receiver
that
you've
had
on
your
thing
for
a
while,
so
you
said:
okay,
do
you
want
us
to
you
want
us
to
help?
It
is
actually
it
is
a
vendor
component
right
I
mean
it's
kind
of
silly
that
you
would
mention
that,
especially
because
it's
related
to
a
proprietary
product
like
it's
kind
of
hard
for
us
to
I've,
seen
this
before,
where
we
have
customers
that
would
actually
build
some
pretty
good
Solutions
based
on
our
stuff
and
then
the
rest
of
them
are
like.
E
Oh
thank
you,
but
I
mean
this
one
seems
like
so
easy
to
maintain
compared
to
some
of
the
stuff
we've
had
to
deal
with
that
we
should
definitely
just
do
it
so
yeah,
I'm,
happy
I,
know
that
the
account
manager
like
I
know
that
you
went
through
the
couple
of
channels
because
I
got
the
like
even
internally
like
a
salesperson
like
the
account
for
relationship
comes
back.
How
can
we?
How
can
we
help
them
like?
Well?
E
E
F
F
F
So,
but
my
initial
thoughts
well,
if
we
make
this
a
primary
first
party
receiver
in
the
collector,
we
could
get
splunk's
help
and
backing
to
make
these
dashboards
alerts
and
like
fine
tune
everything
and
just
cut,
and
have
this
shared
understanding
of
how
we
do
it
all.
E
No,
no
for
sure
I
am
completely
in
agreement.
I
I
will
definitely
make
sure
we
we
get
that
the
thing
I
wanted
to
do
is
to
get
as
much
help
and
love
as
possible
right.
So
that's
not
something
we
do
because
we
we
have
to,
but
because
we
see
a
value
in
that,
and
so
we
have
internal
teams
I
suppose
they
could
also
use
this
stuff.
So
anyway,
it's
a
little
bit
of
Eternal
kitchen.
E
Sorry
I
can
we
can
dig
that
off,
but
yeah
for
the
for
the
draft
PR
that
you've
had
open,
maybe
I
can
help
move
it
forward
a
little
bit.
I
think
it
was
in
a
good
place.
What's
going
on
with
it.
F
Yeah,
so
the
last
like
so
I
wanted
to
use
just
a
simple
endpoint
on
Splunk
and
price,
just
to
be
able
to
check
the
logic
of
like
being
able
to
fetch
the
data
grabbing
it
and
then
passing
it
through
through
the
receiver.
But
the
last
place
I
got
up
to
was
I,
wasn't
sure
if
the
m
dotogen
was
concern,
concurrency,
safe
or
would
be
if
I
added
all
the
data
points.
F
If
there
were
like
data
points
that
shed
attributes
would
I
get
to
two
data
points
that
share
the
same
attribute
like
for
counters,
or
would
they
just
be
rolled
up
into
one?
F
Yeah
yeah,
it's
basically
one
massive
scraper
that
just
starts
a
worker
group
that
pull
that
will
respective
deployments
endpoints.
So
it's
like
for
those
Searcher,
the
cluster
master.
E
It
doesn't
have
to
work.
That's
the
thing
right.
The
stuff
is
in
development,
we're
not
supposed
to
come
up
with
the
perfect
solution
and
I.
Think
the
when
I
talked
to
when
I
talk
to
people
I
spoken
like
well,
you
know:
what
do
we
do?
How
do
we
get
started
and
I
think
we
need
to
get
started
in
a
way
like
we
need
to
get
a
bit
even
if
it
doesn't
work?
Actually,
it's
even
better
if
it
doesn't
work
right,
yeah
that
I
heard
a
joke
recently
about
you.
E
You
know
you
can
you
can
get
Chad
GPT
or
you
can
just
post
the
wrong
solution
to
a
problem
on
the
internet
and
10
people
will
rushed
to
help
you
so
yeah.
I
am
a
strong
believer
that
being
wrong
is
the
best
outcome.
Sometimes
because
then
people
come
and
help
you
yeah.
So
how
can
we?
How
can
we
land
this
PR
at
minimum
so
that
you
can
also
kind
of
get
the
pressure
off
of
having
to
deal
with
this
and
kind
of
step
out
more
than
an
expert?
E
F
E
F
Think
I
just
need
to
get
that
Branch
into
a
state.
That's
passing
the
builds
and
it
should
be
relatively
simple
to
pass
through
I
I.
Think
I,
just
I've
left
enough
comments
to
like
describe
the
intent
and
the
idea
yes,
I,
think
you've
seen
a
lot
of
my
PRS
I.
Try
to
make
sure
that
they're
coherent
enough.
F
Sorry
with
the
recent
internal
turmoil
and
all
of
that
I've
just
like
been
pulled
into
other
things
and
I'm
like
I,
really
want
to
do
it,
but
my
ability
to
have
time
is
becoming.
E
F
Yeah
I
will
see
if
I
can
get
it
in
a
place
today,
where
it's
mergable
and
then
I
will
we'll
use
that
meeting
and
catch
up
next
week
to
talk
about
next
steps.
C
E
I'm
I'm
pleased
with
yours
right
now,
I'm
seeing
multiple
PR's,
where
the
workflows
are
awaiting
approval.
It's
kind
of
sad
because
this
is
probably
like
some
good
stuff.
But
how
would
you
know
the
P,
the
the
work
for
David
run
so
I,
just
unblocks
three
PR's
just
now.
It's
kind
of
it's
kind
of
annoying
having
to
do
that
by
hand
for
the
new
contributors.
E
The
you
can
share
screen
for
a
second
show
you
this.
This
works
right
here,
page
workflows,
approval
this.
This
is
boring,
but
I
mean
security
right.
You
gotta,
pay.
C
F
Yeah
I
I
get
it
that
I
wish.
There
was
a
way
to
flag
that
on
the
pr
to
be
like
this
is
a
new
contributor.
I,
don't
know
if
there's
a
workflow
for
that,
because
it
means.
F
E
E
I
mean
I,
don't
know,
maybe
you're
there
isn't
also
like
you
feel
you
can't
keep
up
with
email
is
because
last
since
Thursday
of
last
week,
I
went
on
a
on
a
weird,
binge
and
I
who
took
all
the
issues
that
said,
needs
triage
and
I
triage
them
all,
and
then
I
went
to
the
backlog
of
all
those
issues
and
I'm
trying
to
find
everything,
I
could
fix
I.
Could
you
know
that
was
just
waiting
for
to
be
closed?
E
I
closed
them
all
when
I
did
that
I
realized
that
some
issues
had
not
been
tagged
properly
because
of
some
issue
with
maybe
I,
don't
know
you
know,
GitHub
actions
didn't
work
at
the
time
or
sometimes
they
don't
get
picked
up
properly.
We
go,
but
so
I
had
to
take
manually.
Some
that
end
up
what
they
end
up
doing
is
tagging
you
right.
F
Yeah
I
need
one
things
that
I
would
love
more
contribution
to
is
the
Kafka
receiver
and
exporter,
mostly
because
I
know
that
in
the
streaming
exporters
at
least
or
the
restraining,
the
streaming
receivers
and
exporters,
there
is
an
issue
when
you
do
when
you
switch
versions
of
The
Collector.
So,
for
example,
if
you
go
from
like
version
a
to
version
B,
but
there
was
a
wire
change
in
your
encoding
format
between
that
oh
version,
B
will
break
because
it
reading
from
the
existing
topic
or
Kinesis
stream
or
pulsar.
F
F
Yeah,
so
we
it's
it's
a
pain
in
the
ass
that
I
know
a
few
users
have
been
caught
out
by
it
and
we
don't
there's
no
nice
way
that
I've
seemed
to
be
able
to
help
document
that.
E
F
A
F
Way,
if
you
created
a
topic,
you
could
either
append
the
version
to
that
topic
so
that
once
you
build
you
like
have
a
version
topic.
That's
like
metrics,
otlp,
metrics
underscore
0.19.0,
so
that
you
can
only
reading
from
that
semantic
version.
Topic,
yeah
I,
think
some
cues
support
the
notion
of
metadata
I
think
at
least
Kafka
does,
but
we
don't
use
that
correctly
or
I.
Don't
think
we
utilize
it
at
all,
but
I'm
like
it's
been
a
while
since
I've
looked
at
that
code,.
D
E
Yeah
we're
we're
still
getting
new
contributions
on
some
weird
exotic
stuff,
like
the
Azure,
even
Hub,
there's
some
stuff,
that's
still
percolating
from
Chris
Perkins.
He
he's
adding
supports
for
metrics
right
now,
which
is
and
then
I
didn't
even
know
you
could
do
that.
Okay,
I'll,
take
it
so
apparently,
Azure
is
exposing
those
metrics
in
some
format
that
we
can
tag
along
and
pick
up
so
yeah.
Why
not
right
yeah.
F
The
one
nice
thing
is:
if
we
had
this
encoding
package,
we
could
have
a
generic
HTTP
exporter
and
just
provide
enough
conflict
so
that
vendors
don't
have
to
build
their
own
exporter
like,
for
example,
like
you
could.
If
we
had
a
signal
effects
encoding
like
encoder
for
metrics,
we
could
just
provide
that
to
the
HTTP
used
encoding.
Here's
the
endpoint
send.
E
F
Yeah
so
like
heck
or
even
just
like
cloudwatch
metrics,
if
you
wanted
to
pipe
them
somewhere
else
to
S3
and
like
you
can
pull
that
into
the
file
file,
receiver
or
Explorer.
C
F
E
I
mean
I've
had
this
issue
with
the
pocket
exporter
the
market
exporter.
Actually,
really
do
you
need
to
have
some
pretty
strong
opinions
about
how
you're
going
to
export
your
data
to
map
it
to
a
certain
Market
schema
and
initially
what
I
was
trying
to
do
is
to
bulge
the
ocean
and
say:
oh
there's,
probably
Universal
Market
exporters
going
to
work
for
any
open
Telemetry
data
based
on
the
protofiles
right.
E
And
I
kind
of
like
the
idea
you're
having
here
where
maybe
what
I
need
to
do
is
offer
a
developer.
Api
that
says:
build
your
own
pocket
exporter,
insert
your
pocket
translation
layer
here,
just
like
HTTP
some
idea
and
then
build
your
and
then
every
vendor
is
going
to
have
their
own
flavor
of
how
to
export
departed
using
the
schema,
That
Matters
to
them
yeah
in
general.
They
would
that
would
help.
F
I
think
yeah
I
think
it
just
makes
the
experience
slightly
better
and
just
allows
us
to
track
like
who's
using
what
encoding
a
lot
easier,
because,
like
one
thing,
I
really
hate,
is
that
we
have
of
vendor-specific
things
of
Open
Source,
because
I
can't
remember
what.
But
there
was
someone
wanted
to
add
a
proprietary
vendor
coding
to
one
of
the
exporters,
but
because
it
was
vendor
vendor
specific.
It
was
pushed
out
of
the
the
agnostic
firm
component
and
made
into
their
own
yeah.
F
C
E
And
so
you
would
do
like
extension,
like
kind
of
a
yaml
exposing
coatings.
F
F
Because
yeah
I've
from
my
perspective,
I,
think
that
reduces
a
lot
of
the
vendor-specific
instrument.
Exporters
or
receivers
that
may
or
may
not
need
actually
need
their
own
dedicated
one.
F
E
I
mean
we're,
certainly
there's
a
race
for
everybody
to
standardize
on
the
TLP
tool
right
rather
than
creating
your
own.
So
hopefully
the
best
way
to
win
also
is
to
yeah.
We
are
going
to
make
it
so
much
convenient
for
vendors,
but
why
don't
you
go
with
adopting
your
telepians,
then
right.
E
So
I
actually
wanted
to
bring
up
a
subject
to
spawning,
but
didn't
make
it
to
the
collective
meeting
this
morning,
I
have
a
controversial
take,
which
is
that
I
think
we
are
kind
of
lacking
a
little
bit
in
process.
E
We
we
have
processed,
don't
get
me
wrong
and
you
guys
are
actually
the
most
obedient
when
it
comes
to
like
getting
things
right
and
doing
the
tagging
and
being
good
about
process
and
but
when
I
was
doing
the
review
of
all
the
issues.
I
was
like
some
of
the
stuff,
like
I,
think
it's
probably
disconcerting.
E
When
someone
just
ring
the
repo
I'd
like
us
to
be
much
better
about
closing
things
and
know
that
I
know
there's
some
automations
supposed
to
close
up
style
issues
and
it's
just
not
working,
but
we
have
well
probably
600
or
so
studies
using
there
of
830.,
it's
kind
of
a
shame.
E
So
in
general
you
know
what
I
like
to
do
is
to
to
close
all
this
stuff.
I
also
would
like
to
probably
start
triaging
and
helping
out
the
core
collector,
which,
in
my
opinion,
gets
even
less
love.
E
I
mean
and
kind
of
applies
the
simacity
to
build
on
the
issues
and
start
closing
aggressively
stuff
I'm
a
little
worried
that
actually
well
here's
the
feedback,
maybe
that
I'm
trying
to
get
to
I'm
still
getting
a
lot
of
feedback
of
people.
That
say
it's
great
teacher
and
better
I
hate
it.
Can
you
please
go
to
stable
like
everybody
else?
What
are
you
guys
doing
and
you
know
like
a
dab
Dance
all
day
about
how
better
is
the
new
GA
and
how
great
things
are?
E
I'm,
sorry,
I
think
this
just
doesn't
work
like.
Maybe
maybe
we
need
to
stretch
stable,
maybe
using
international
component
States.
Maybe
we
need
to
do
better,
a
big
more
specific
about
things,
but
that's
kind
of
probably
gonna
be
my
new.
You
know
my
new
adventure
or
crazy
ideas
just
start
like
really
moving
on
quality
and
closing
down
all
sorts
of
stuff,
because
I
I
Feel
Like,
We
Are
we're
not
doing
a
good
job
of
servicing
the
community
in
the
sense
of
adoption.
In
that
sense,.
C
E
Think
there
are,
we
have
too
many
open
issues.
We
have
too
many
things
that
are
not
reached.
We
have
too
many
things
that
need
to
be
just
claimed
down,
and
we
should
just
be
good
about
closing
things,
and
you
know
I
try
to
move
components
to
stable
about
three
two
three
months
ago,
when
the
feedback
I
got
was
interesting.
Is
that
if
your
components
for
any
reason
depends
on
any
of
the
other
components
which
are
not
themselves
more
stable,
then
you
cannot.
You
cannot
Market
stable.
B
E
It
does
make
sense,
it
hurts,
but
pretty
okay,
that's
also
telling
me
very
clearly:
oh,
you
need
me
to
move
up
the
stream
to
find
like
up
the
dependency
tree,
what
else
I'm,
depending
on
and
what
those
things
so
I,
unfortunately
find
myself
having
to
go
like
way
up
to
like,
for
example,
depend
on
the
config
package.
Okay,
config
package,
see
if
it's
better,
it's
not
stable.
That
thing
depends
on
three
other
things
right,
so
it
keeps
going
yeah
I,
think
it's
gonna
be
a
fight.
F
E
F
Think
I
think
that
was
part
of
that
conversation,
but
so
I
think
it
was
like
in
two
parts.
Issues
are
not
being
closed,
so
we
can't
correctly
just
articulate
if
a
component
or
portion
of
the
collector
is
actually
being
effectively
used
and
without
issues
and
is
would
be
considered
stable.
Like
there's
no
outstanding
issues,
all
feature
development
is
stopped
and
that,
like
it
is
in
just
maintenance
mode
or
just
getting
feature,
updates
and
stuff
like
that,
the
other
part
is.
We
need
to
actually
clearly
Define.
What
stable
is
I
know.
F
Hotel
has
a
definition
of
what
we
expect
stable
to
be
and
I
think
that's
three
years
support
for
like
backwards
compatibility,
I,
believe,
that's,
probably
the
hesitation
of
wanting
to
push
things
to
stable.
Just
because
of
that
guarantee
that
we're
trying
to
push
forward
with
open
telemetry.
F
E
F
This
going
to
work,
I
think
what
would
be
a
good
idea
is
we
need
to
start
having
cow
node
as
part
of
real
wanting
to
contribute
your
component
and
stuff
like
that.
You
need
to
provide
like
a
future
roadmap
of
like
these
are
the
stuff
that
I
foresee
to
getting
us
to
stable,
and
that
would
help
articulate
the
definitions
of
stable
for
us
and
that
solves
it
on
the
per
component
basis.
C
F
This
outstanding
feature
for
Kinesis
to
not
to
include
kpl
aggregation.
E
Yeah
yeah:
well,
let's
take
Mystic
exactly
right,
the
cafe
Receiver
right,
so
you
could,
you
could
say:
yes,
we
will
add
additional
protocols
but
really
to
get
stable.
E
B
F
Think
what
would
be
a
good
discussion
is
having
an
issue
that
said
like
how
defining
a
road
map
to
stable
and
then
start
having
code
owners
defining
what
it
means
for
their
for
their
component
to
be
stable,
yep
I
think
that
will
help
surface
what
needs
to
be
marked
stable,
like
the
priority
of
what
needs
to
be
marked
stable
in
the
collect
like
in
court
and
I.
F
E
Yeah
yeah
for
those
what
I'm
gonna
say
is
every
issue
in
the
core
needs
to
be
tagged
with
the
area
that
they
touch
right
and
some
of
them
are
tagged
like
area
Builder
area
config.
Maybe
we
need
to
go
and
make
sure
that
everything
is
direct
so
that
you
have
also.
These
are
like
roadmap
items
for
your
module
in
the
sense,
and
maybe
that's
how
you
build
a
roadmap
right.
You
take
a
look
at
everything,
that's
open
against
your
component
and
you
say:
oh
well,
I
see
you
have
tennis
shoes.
E
Eight
of
them
are
bugs
that
I
need
to
fix.
One
of
them
is
a
feature
that
I'd
like
to
really
get
to
before
Market
stable,
when
one
of
them
is
just
relevant,
I,
don't
care
for
it.
So
this
way
you
can,
you
can
build
a
roadmap
and
your
roadmaps
just
a
link
of
links
to
everything.
That's
opened
against
you
and
voila.
E
A
E
Yeah,
that's
a
good
point,
I
think
being
new
to
to
taking
the
role
of
treasure
was
going
to
be
interesting
for
me
on
the
contrib
to
kind
of
get
a
better
sense
of
what's
actually
happening.
It's
a
really
interesting
yeah.
Actually,
you've
got
a
nice
kind
of
tour
of
the
components
actually,
because
then
you
get
to
read
about
all
the
issues
really
realized
that
we
have
a
ton
of
Prometheus
issues
in
the
back.
That
I
don't
think
we
should
fix
because
they're,
not
ours.
There
are
Prometheus
things.
F
E
Yeah
for
sure-
and
this
is
now
in
a
good
place
where
we
could
do
that
because
we
have
like
so
many
of
those
are
tagged
properly
now
that
we
can
probably
do
that
for
Prometheus,
there's
still
Prometheus
tag,
that's
used
for
some
reason,
so
you
have
so
I
feel
like
43
of
them
are
tagged
as
Prometheus.
E
That's
good,
two
pages,
that's
the
most
most
ancient
ones,
and
then
you
know
around
and
there's
still
some
that
need
some
love
I'm,
seeing
some
that
that
could
really
be
I'm.
Gonna
have
to
study
them
again.
E
I
mean
we
know,
we
know
this
is
normal
right.
It's
a
time
like
there's
just
too
much
to
do,
but
yeah
we
we're
gonna.
Have
those
issues.
I
have
one
for
the
Kafka
exporter.
Just
now
use
exporters
Cafe
with
Z
series,
gets
an
error.
C
E
Sitting
there
for
a
while,
but
actually
it's
just
an
issue
with
the
customers-
configuration
that's
Issue,
four,
four,
nine
one
and
three,
so
it's
okay,
we
just
need
to,
like
you
know,
fix
up
all
that
stuff
and
I
agree
with
you.
We
should
just
do
a
massive
thing,
but
maybe
to
make
it
more
worthwhile
than
just
hey
guys.
You
have
work
to
do.
You
need
to
review
your
stuff
for
six
months.
A
F
Yeah
yeah,
okay,
let.
E
C
E
Righty
other
than
that
I
got
a
weird
receiver
for
you.
If
you
like,
yeah.
E
E
So
the
receiver
is
a
very
simple
one
that
is
looking
for
a
sponsored.
E
The
the
receiver
itself
consists
of
looking
at
files
and
Reporting
the
last
time
they
were
modified.
Their
size.
Customers
are
accessed,
okay,.
C
F
B
E
E
E
F
No
I
at
least
knows
a
tool
like
OS
query
could
like
would
help
discover
things
like
that.
Well,
that's
more
of
like
compliance
and
like
management
control
like
operation,
then
like
I,.
F
B
E
So
anyway,
I'm
I'm,
looking
for
maybe
someone
who
would
like
to
sponsor
this
I
mean
I,
don't
know
if
he
can
sponsor
my
own
stuff,
but
that
sounds
gross.
E
B
C
E
The
metrics
I
picked
might
not
be
the
metrics
that
the
names
might
you
know
naming
is
hard.
F
E
Try
to
get
them,
yeah
I
mean
it
keeps
professor
of
support.
So
it's
a
it's
an
actual
like
you
know,
customer
relationship,
and
they
need
this
urgently,
so
maybe
we'll
we'll
see
how
to
best
help
them
out,
but
right
now
this
that's
all
we
have
I
I
went
a
little
deeper
than
I
think
what
they
need,
what
they
wanted
to
make
sure
that
nobody
changes
the
file.
E
What
I'm
offering
is
you
can
see
the
timestamp
of
any
time
the
file
has
been
accessed,
modified
or
changed
permission
changed
right.
F
Okay
I'll
comment
after
the
meeting,
because
I
have
some
ideas
because
you
get
to
look
at
like
C
group
allocation
and
stuff
like
that
and
expanded
to
include
that
kind
of
information.
B
E
I
mean
definitely
I
have
a
very
partial
view
of
the
use
case
in
solution
here
so
I
mean
so
I
mean
really
all
this
is
the
output
of
os.stat
right
here
yeah.
E
F
F
Sorry
zq
I
realized,
like
we've
been
like
talking,
I,
don't
know.
If
you
have
anything
else,
you
wanted
to
bring
up
yourself.
F
D
F
I'm,
so
sorry
is
there
anything
that
you
want
us
to
repeat,
or
do
you
want
to
help
get
some
more
clarity
on.
D
F
Yeah,
that's
fine
again,
like
you
can
ping
us
on
slack
or
stuff
like
that
everything
don't
make
sense.
Just
if
you,
whatever.
D
E
Hey
so
maybe
a
point
for
that,
do
you
know
who
sets
up
those
Zoom
calls.
F
So
they're
set
up
that
by
the
government's
Community,
so
only
they
have
access
to
edit
permissions
on
the
calendar.
So
I've
I've
been
going
I've
been
asking
Morgan
to
do
that.
For
us.
E
Oh
yeah,
okay,
I
know,
there's
an
option
also
in
Zoom
to
have
captions
automatic
generated
I
mean
my
French
accent
gets
in
the
way
of
stuff.
So
if
that
might
be
a
thing
to
ask
at
the
same
time
that
you
asked
maybe
move
the
simple,
if
you
don't
mind
asking
Morgan
to
say:
hey,
can
you
click
the
checkbox
for
captions?
It
will
help
a
lot
people
who
might
not
be
like
me,
who
are
not
native
to
English
if
they
have
if
they
want
to
follow
along.