►
From YouTube: 2022-06-01 meeting
Description
cncf-opentelemetry@cncf.io's Personal Meeting Room
B
B
Yeah,
I
so
I
think
we
already
have
corner
today.
I
know
that
robert
is
in
pto.
Raj
also
is
out
of
fast,
so
yeah.
I
think
we
can
get
started.
B
All
right,
starting
with
the
prs
yesterday,
I
opened
one
I
was
trying
to
get
familiar
with
the
metrics,
so
I
start
to
set
up
the
example
to
run
the
metrics
basically
noted
that
we
didn't
have
the
export
and
we
are
going
to
have
follow-ups.
I
still
I
thought
that
people
and
chris
had
feedback
on
the
pr.
I
still
didn't
have
a
chance
to
look,
but
at
least
there
will
be
some
follow-up,
like
the
configurational
settings
of
configuration
environment.
Helper
actually
doesn't
have
tests
for
those.
B
We
just
have
tests
dealing
with
the
the
password
for
the
enumerations,
and
I
just
build
on
top
of
that,
but
we
need
to
build
tests
for
the
other.
I
create
three
items
to
follow
up,
but
it's
basically
just
adding
the
samples,
and
I
think
we
can
discuss
more
when
we
get
to
the
issues
that
I
open
related
to
this
pr,
the
other
one.
I
think
I
think
robert
asked
he
put
the
template
for
dprs
and
I
think
zach
he
asked
it
to
everyone
to
approve.
B
I
think
you
were
the
last
one.
So
when
you
approve
merchants,
if
you
approve
it,
unless
you
have
something
to
try
there.
B
So
mateos
actually
pointed
this
to
me
that
there
was
actually
during
the
review
of
one
of
the
pr
zac
already
had
pointed
this.
But
anyway,
when
I
was
trying
to
do
the
example,
I
thought
that
the
things
were
not
consistent.
You
know,
and
I'm
it's
not
clear
to
me.
For
instance,
sometimes
metric
is
singular.
Sometimes
it's
metrics
and
plural
meter
sometimes
feels
like
it's
the
tracer,
but
sometimes
feels
like
a
logical
grouping
from
metrics.
B
So
I
think
we
should
just
carefully
review
and
separate
this.
A
related
thing
is
that
there
are
common
settings
that
right
now
are
on
the
tracer,
but
I
think
we
should
separate
from
the
signal
itself.
You
know,
let's
say
I
think,
flush
on
exit.
I
don't
think
we
should
have
different
settings
for
metrics
traces
and
eventually
logs,
but
I
think
they
are
generic.
They
apply
to
all
the
signals.
So
this
is
something
that
I
think
we
should
review.
B
Perhaps
we
can
get
back
today
and
kind
of
have
a
quick
discussion
about
those,
but
that
is
the
the
one
of
the
issues
that
I
opened
yeah,
and
this
is
what
I
just
mentioned
like
the
flushing
on
on
exit.
I
think
at
this
moment
I
think
we
should
have
just
one,
and
that
applies
to
any
signals
that
are
being
used
by
the
application
you
know
emitted
captured
by
us.
I
don't
see
in
principle,
I
don't
see
a
reason
to
have
this
separate
for
each
signal.
C
That
flush
on
on
exit
or
flush
on
on
unhandled
exception,
the
the
primary
concern.
There
is
there's
a
problem
and
we
don't
want
them
to
have
to
set
multiple
settings
to
to
control
that.
B
We
we
separate,
for
instance,
stuff
like
the
export
right,
but
then
I
I
I
the
question
becomes
for
other
things
like
that,
for
instance
the
autop
exporter,
to
go
without
unencryption
to
go
unencrypted,
it's
a
setting
that
applies
to
the
application.
B
So
does
it
make
sense
for
us
to
say
that
that
is
related
to
tracing?
I
I
don't
think
so,
because
the
whole
application
the
setting
is
going
to
be
applied,
even
if
the
other
exporter
is
prometheus.
You
know
so
that
in
principle
for
me
should
be
generic
independent
of
the
signal
you
know
I.
I
know
that
you'll
be
a
lot
of
cases
like
that.
B
A
lot
of
know
a
few
cases
like
that,
but
I
think
we
have
to
kind
of
have
one
place
very
specific
for
these
global
settings,
independent
of
signal
and
the
ones
that
are
very,
very
signal
specific.
B
So
I
I
jumped
kind
of
fast
to
the
two
first
items,
but
actually
the
important
thing
is
to
decide
when
we
want
to
target
those.
You
know
this
one.
I
would
like
to
target
to
the
next
beta.
B
So
since
we
are
gonna
move
that
one
I'm
gonna
move
to
the
same,
so
we
try
to
sort
all
this
to
the
same.
B
B
This
is
an
interesting
one.
I
think
yesterday
I
gave
a
run
of
this,
just
to
be
sure
that
I
was
understanding.
So
there
is
one
exception
here
that
is
kind
of
expected
in
the
sense
that
they
didn't
have.
B
C
I
took
a
quick
look
and
I
I
mean
something's
timing
out
and
so
what's
going
on,
I
mean
it
seems
like
a
networking
issue
at
first
glance
and
I
don't
know
how
they're
running
zipkin
or
where
they're
running
zipkin,
but
the
export.
The
zipkin
exporter
can't
seem
to
talk
to
zipkin
at
the
url
that
they
configured.
B
Did
you
try
because
of
the
example
we
can
just
to
do
a
sentence
check
because
our
test
used
the
marketing,
so
I
think
the
zipper
king
should
be
fine,
but
just
as
a
senate
check
can
you
run
the
example?
Just
change
the
exporter
to
be
zipking
and
see
if
it's
working
just.
B
B
We
should
try
to
figure
out,
but
it
just
seems
a
timeout
yeah,
and
this
is
our
default.
Timeout
seems
a
bit
big.
B
B
B
A
B
Yeah,
but
this
default
timeout
seems
pretty
bad
right
100
seconds
for
the
default.
We
can
change.
Yes,
yes,
but
perhaps
the
fix
is
on
the
sdk,
but
so
that
we
live
as
it
is.
That's
my
question.
A
B
Okay,
actually,
chris
I'm
gonna
be
doing
some
stuff.
With
the
example
I
can,
I
can
try
to
run
this
using
the
zip
can
export.
I
I'll
do
that.
D
C
Well,
no,
no
more
specifically,
it
has
to
be
an
asp.net
core
application.
You
probably
have
to
use
the
asp.net
core
iis
module.
B
Okay,
that's
fine,
the
the
setup
that
I
have.
You
typically
have
that
ready,
but
okay,
but
or
or
do
you
mean
that
they
are
not
using?
No,
it's
just
a
asp.net
core
application.
B
That's
fine!
That's
right!
Okay,
I'll
give
a
try
just
send
check,
I'm
not
gonna
kind
of
look
much
deeper
because
it
seems
that
I'm
out
seems
like
you're
just
not
reaching.
Oh.
A
C
And
if
I
remember
right,
the
the
sdks
documentation
for
shows
how
to
use
code
to
change
some
of
the
settings
and
replace
your
hp,
client
factory
or
something
like
that.
B
A
C
C
C
That
all
of
them
expose
that
that
same
technique,
zipkin
might
be
unique
because
jaeger
has
its
own
protocol.
But
zipkin,
I
think,
was
the
simplest
export
protocol.
B
Okay,
so
because
either
is
gonna
require
code
right
now
or
change
on
the
sdk,
but
it
looks
like
code
via
the
factory
spr
to
explain
it.
What
milestone.
B
B
Oh,
but
let
me
let
me
because
this
is
going
to
confuse
with
the
the
issue
itself,
so
so
sophia.
So
you
may
end
up
not
creating
the
issue
right,
because
if
that
is
the
the
way
to
configure
by
the
sdk
and
then
we
have
to
write
code,
then
we
can
create
a
issue
on
our
repo.
So
that's
the
path,
and
then
you
put
the
milestone.
As
I
said,
please.
B
Yeah,
so
for
this
one
right
now,
I
will
leave
it
open
because
we
don't
know
yet.
I'm
gonna
try
the
repro.
If
doesn't
reproduce,
we
have
to
decide.
B
This
was
the
one
that
I
looked
at
a
little
bit
and
actually
there
are
two
things:
there
are
two
errors:
first,
this
symbol
of
cup
and
then
he
fix
it
by
updating
the
image,
and
I
think
I
understand
why
I
will
get
that
and
then
he
had
the
error
because
of
not
having
the
the
the
right
version
of
system.
B
Diagonals
piotr
pointed
that
by
adding
the
package
that
should
fix,
I
give
a
try
to
use,
because
that
first
error
is
just
us
trying
to
use
some
dependence-
that's
not
present
on
the
system.
B
So
what
I
noticed
based
on
what
the
person
that
reported
said,
is
that
if
we
use
a
more
recent
image,
then
the
error
is
gone.
You
know,
so
the
way
that
we
are
building
probably
are
linking
to
a
kind
of
more
up-to-date
dev
library,
and
when
we
go
to
these
older
versions
it
breaks.
B
My
guess
is
that
this
is
not
something
that
we
should
be
fixing,
because
five
is
already
out.
Dotnet
3.1
is
going
to
be
out
by
the
end
of
the
year
and
we
have
a
workaround.
So
I
intended
to
think
hey.
If
you
want
to
use
with
the
net
core
app
3.1,
I'm
tending
to
think
hey
use
bullseye
instead,
you
know.
Otherwise
you
need
to
look
at
the
build
to
be
sure
that
we
link
with
older
versions
of
the
the
libraries.
C
I'm
I'm
wondering
going
forward
if
we
should
be
expanding
some
of
our
test
cases
to
to
cover
some
some
of
this.
B
I
think
it's
a
good
idea.
We
should,
we
should
add
a
test
skills
to
start
with
these
images.
The
images
that
we
support.
B
C
Because
it
seems
like
a
gotcha
that
we
may
not
notice
when
we
up
update
a
native
dependency
and
all
of
our
tests
pass.
So
we
think
okay
looks
like
we're
good,
but
our
tests
ran
against
a
specific
version
of
a
of
a
container
and
it
didn't
work
against
some
of
the
older
ones
that
are
kind
of
one
of
the
standard
ones.
That
people
would
use.
B
Yeah
we
we
should
be
having
a
test
that
we
are
building
the
container
itself
that
we're
gonna,
because
right
now,
when
we
run
the
docker
composer
stands
for
the
stuff
that
you're
gonna
talk
more
would
be
red
just
this
kind
of
stuff,
but
you're
right.
We
don't
build
kind
of
hey.
Let's,
let's
try
to
run
our
beats
on
3.1
docker
image.
You
know
we
need
one
issue
for
that.
Okay,
are
you
open
the
issue.
B
Yeah,
so
I
think
right
now
we
are
building
linux
actually
on
this
version
here
and
probably
this
has
more
up
to
date,
libraries
for
the
compiler
and
then
when
we
go
to
3.1
5.0,
it.
C
C
And
you
have
to
specifically
call
add
more
more
to
the
name
if
you
want
a
specific
ubuntu
version
or
and
so
on,
or
even
a
specific
debian
version.
Yes,.
B
Yes,
so,
but
but
the
rpm
package
should
use
the
same
libraries,
so
the
package
that
you
build
should
use
the
same
libraries,
but
because,
like
3.1,
the
default
is
stuck
sometime
in
the
past.
Now
that
we
build
against
20.04
the
dependence
that
we
need
is
not
there.
You
know.
D
B
We
can
just
cross
in
the
next
meeting
the
milestone
for
that,
depending
on
the
cycles
that
people
have,
but
I
have
opened
the
issue
and
we
decided
from
there
and
perhaps
that
what
we
should
add
here
is
kind
of
the
matrix
of
images
from
the
dotnet
that
we
test
against.
You
know
in
that
issue
that
are
you
open
just
to
be
sure
that
we
are
covering
people
can
expect
hey
they
test
against
this
image.
You
know.
A
B
I
think
should
be
actually
because,
even
if
it's
let
me
even
if
it's
just
code
on
our
side,
I
I
think
it's
going
to
take
a
little
bit
to
to
get
to
that.
I
kind
of
I
think
for
the
sdk
perspective,
it's
pretty
good,
that
they
have
this
password
with
the
factory
to
create,
but
I
kind
of
if
I
was
a
api
user,
it's
such
a
common
case
to
change
that
default,
because
the
default
is
kind
of
bad
that
I
would
like
a
quick
configuration
for
that.
B
So
we
discussed
this,
I
I'm
gonna
open
the
the
follow
up
for
this
one,
and
I
don't
know
if
we
should
track
it
for
our
ceo
beta.
Let's
after
hoping,
we,
we
discussed
that
in
the
next
meeting.
C
Yeah,
my
guess
is
either
a
future
beta
or
the
rc
simply
because
it's
not
something
that
we're
directly
claiming
that
we
support
it's
really
expanding
our
support,
matrix
and
safety
net.
B
C
No
zero
to
zero
is
zero.
D
B
A
C
Yeah,
we
might
just
push
it
out
a
week
for
now,
maybe
because
I
think
we've
got
it
for
the
seventh
if
we
bump
it
to
the
14th
that'll
give
us
another
week
to
reevaluate.
C
I've
started
to
look
at
how
to
do
the
integration
tests
and
do
it
in
smaller
steps,
but
there's
some
things
that
they'll
probably
be
questions
about.
C
For
example,
we
would
need
to
add
in
some
some
of
the
protobuf
support
and
bringing
in
the
otlp
protobufs
into
our
repo,
because
we
can't
simply
extract
the
the
protobuf
stuff
from
the
otlp
exporter.
All
that's
marked
internal,
so
it
would
be
something
we'd
have
to
version
separately.
B
C
B
Yeah
pushing
two
weeks
out
from
the
seventh
to
the
21st,
so
we
have
time
to
really
finish
this
stuff
regarding
matrix.
That
is
the
big
addition
for
the
next
beta
release.
C
Yeah,
so
there
is
something
of
interest
on
the
20th,
I
believe
so
I
want
to
say:
that's
the
open,
telemetry
community
community
day
conference,
I'm
not
attending
in
person,
but
it
could
be
interesting
for
somebody
to
bring
up
during
that
conference.
But
I
I
don't
know.
B
Yeah,
I
I
have
the
feeling
that
we
may
be
not
able
to
to
have
these
three
by
then
that
were
committed
here.
I'm
not
gonna
commit
anything
else.
Unless
there's
a
bug
fix
that
we
discover,
but
I
think
it's
more
realistic
on
the
21st.
B
D
D
B
Yeah,
so
I
think
the
project
board
looks
good.
We
don't
have
anything
we
are
going
to
keep
the
committee
they'll
move
the
dates
yeah.
I.
B
Do
you
guys
want
to
bring
anything
that
passed
not
discuss
it.
A
A
B
Yeah
so
good
good
good
that
people
to
remember
that
I
had
forgotten
about
that.
So
this
friday
8
a.m,
pacific
time
in
the
calendar,
it's
for
anyone
interested
in
profiling.
It's
a
it's
a
good
start,
it's
right
the
beginning!
So
it's
a
good
chance
to
get
involved
on
that.
C
B
Yeah
yeah,
I
I
I'm
curious
about
those,
but
I
wait
that
you'll
fry
that
you're
here
there.
C
Yeah
I
mean
one
example
is
there's
a
group
working
on
jfr
support
in
open
telemetry,
and
so
that
involves
some
profiling
data.
Similarly
browser
the
client-side
stuff
has
some
profiling
information,
but
they're
still
sorting
things
out.
B
Yeah,
I
I
think
the
the
big
thing
that
crossed
my
mind
is
that
I
tend
to
think
that
people
doing
these
languages
are
gonna
kind
of
because
it's
what
they
are
doing,
so
they
are
gonna
approach,
a
lot
from
their
language
perspective,
and
I
think
what
I
eventually
seek
for
profiling
should
be
a
lot
about
a
format.
That's
general
enough
that
you
can
submit
data
in
any
language.
You
know
and
yeah.
B
Very,
like
you'll
be
prof,
because
google
already
offered
to
to
put
that
so
very
likely
or
bp
profit.
If
you
don't
like
pre-profi,
it's
your
chance
to
go
there
in
the
beginning
and
say
no
before.
C
B
Yes,
yes,
that
that
I
think
there
is
a
red
one
speck
that
up,
because
it
allows
a
kind
of
this
binary
data
into
the
logs.
So
people
could
go
to
that.
You
know.