►
From YouTube: 2022-09-08 meeting
Description
Open cncf-opentelemetry-meeting-3@cncf.io's Personal Meeting Room
B
So
how
is
what
are
all
the
things
that
I
missed.
B
I
was
like:
okay,
that's
yeah,
I
haven't,
you
know
done
all
the
things
that
I
was
planning
to
do,
which
I
need
to
do
that.
Compiler,
docker
image
and
update
yeah
did
that
protocol
thing
get
updated
because
I
saw
you
I
had
some
remember.
There
was
a
again
version,
let's
mention
the
collector
and
yeah
yeah.
So.
A
B
B
B
B
Around
that,
but
but
I'm
thinking
you
know,
this
is
a
problem
across
so
many
language
things,
so
many
hours
are
spent.
Trying
to
you
know,
keep
this
thing
yeah,
which
seems
there
should
be
someone
you
know
contributing
or
something
some
kind
of
a
tool
right,
yeah
yeah,
like
even
the
docker
tool
that
they
have
is
somewhere
hidden,
and
I
had
to
ask
the
you
know
the
collector
person
and
they
send
me.
You
know
the
link
for
that.
B
A
B
It
should
be
like
blinking
yeah.
A
B
So
so
do
you
guys
have
a
collector
that
a
version
of
character
that
does
something
of
the
sort
or
by
the
way
I
I
just
shared
my
okay-
that's
why
I'm
just
still
getting
used
to
it,
so
I'm
keeping
my
hat
on
okay.
B
So
what
am
I
saying
yeah
so
so
from
what
I've
you
know
talked
to
all
these
companies.
Everyone
has
a
connector
and
yeah.
A
B
Yeah,
so
the
collector
has
yeah
collector
has
a
endpoint.
It
takes
multiple
endpoints
yeah.
Now,
whatever
is
going
to
ask
was
so
that
version
that
tlp,
so
the
otlp
has
got
the
data
and
then
the
data
model,
and
then
the
buyer
protocol
right.
So
if
the
wire
protocol
doesn't
change
and
the
data
model
changes,
it
should
still
be
able
to
send
across.
A
B
So
you
you,
your
data
dog
agent,
you
know
the
team.
Actually
that
maintains
that
a
collector
endpoint,
you
know
make
sure
that
you
know
it
maintains.
A
B
A
B
Point
is
so
the
the
collector
endpoint
datalog
agent
supports
has
to
be
compliant
with
the
let's
say:
python
javascript
java,
whatever
sdk,
that
is
sending
those
emitting
those
signals
how
to
be
compliant
with
the
data
model
and
the
wire
protocol
that
the
agent
collector
supports
agent.
Otrp
endpoint
supports
yeah.
A
A
A
B
I
mean
that
is,
that
is
essentially
the
problem
right,
because
you
have
to
have
a
version
that
is
tested,
yeah
yeah
compatible
according
to
the.
B
B
Yeah,
so
that's
where
I'm
you
know
struggling
to
find
the
right.
B
Because
otherwise
you
know
you
have
people
who
talk
high
level,
but
they
cannot
execute
right.
So
that's
so
yeah
that
is
but,
like
I
said
right,
I
applied
a
data
doctor,
so
you
know
it's
a
sales
job
which
becomes
not
necessarily
sometimes
just
a
pure
sales
is
not
very,
you
know
yeah.
So
that
is
so.
B
Then
it
becomes
more
of
a
you
know,
presentation,
demo
right
and
that
I
mean
I
can
do
that,
but
I
want
to
do
more
right,
so
that's
kind
of
where
things
are
so
I'm
still,
you
know
looking
for
the
right
effect,
but
yeah
once
that
is
set.
Then
I
can,
you
know,
get
back
into
yeah
focus
more
on
yeah
I
mean
I,
I
definitely
have
been
yeah
yeah.
A
B
It's
a
matter
of
time,
so
yeah.
A
B
So
I
saw
app
dynamics.
You
know,
folks
are
also
using
this
yeah,
so
yeah.
A
They
yeah,
I,
yes,
they
have
joined
the
team
and
they
they
have
come
to
some
meetings
just
at
the
before
starting,
but
you
were
in
the
first
meeting.
I
think
that
when
they
appeared-
and
he.
A
He
asked
some
stephen
some
question
on
the
channel
about
url
instrumentation
how
to
remove
the
url
or
filter
that
you
are
editing
on
the
library,
and
I
answer
him
on
the
channel
for
the
rest
yeah.
I
released
some
versions,
one
that
someone
with
the
otp
fixes
also
previously
supporting
the
latest
xcode
and
also
yeah.
Also
bryce
was
working
on
supporting
custom
metrics
as
he
was
he.
A
He
was
working
on
a
project
that
was
using
apple
metrics
about
the
apps,
so
they
they
don't
give
you
live
reporting,
but
they
give
you
reports
like
the.
A
Data,
so
the
thing
was
that
how
he
could
send
all
that
metrics
into
open
telemetry
without
going
through
the
metrics
api
itself,
because
metrics
api
suppose.
So
I
mean
you
get
metrics
and
you
are
calculating
things
like,
let's
say
with
the
data
points
and
that
stuff,
so
he
has
been
working
on
supporting
providing
all
the
metrics
and
already
calculated
on
the
with
the
times
and
everything.
So
you
are
not
calculating
on
live
but
just
fill
in
the
metrics
data.
So
you
can
report
that
through
open
telemetry
and
that's
up.
A
A
A
A
B
B
A
Based
on,
what's
reported,
he's
he's
doing
everything
yeah,
that's
the
point.
The
time
time
is
not,
I
mean
it's,
not
life
yeah,
so
you
have
to
put
the
time
standard.
There
was
in
the
past
also,
so
he
is
providing
a
way
to
to
fill
the
metric
data
without
using
a
metric,
an
open,
telemetry
metric,
but
just
filling
the
data
so
that
that
that's
almost
done.
A
Currently,
it's
not
compiling
because
some
he
did
some
tests
with
the
frames,
but
it
should
be
so
the
yeah.
B
No
apple,
a
bunch
of
stuff
right,
we
should
try
to
use
as
much
as
you
know
what
they
have
right.
I
think
they
have.
They
have.
A
Yeah
yeah.
The
thing
is
that
if
you
want
to
collect
metrics
about
your
app
some
internal
metrics,
especially
about
the
cpu
memory
and
things
like
that,
you
cannot
do
that
live
during
your
app
run.
The
only
way
to
get
that
information
is
through
apple
metrics.
That
reports
you
when
apple
wants
so.
B
A
Get
that
later,
so
you
want
to
use
that
in
a
in
a
in
a
way
that
you
can
compare
with
others
and
things
like
that.
You
he's
using
open
telemetry
for
that,
and
it
makes
all
sense,
but
the
api
was
not
prepared
for
that
kind
of.
B
But
this
is
the
add-on
it
doesn't
have
to.
We
don't
have
to
mess
with
the
otrp
or
open
telemetry
spec
for.
A
We
metrics
api,
it
has
not
been
updated.
B
No,
but
for
for
using
the
old
matrix
you
could
you
could
fit
it
into
the
model
that
that
open
telemetry
has.
You
didn't
have
to
yeah.
A
B
Now
that
that
makes
yeah,
this
is
similar
to
the
instruments.
Right
I
mean,
if
you
look
at
get
the
dsim
file
open
up
in
instruments
and
you
get
lots
of
data,
but
then
yeah,
but
that
is
all
when
there
is
a
crash,
or
you
know,
yeah
after
the
fact
so
yeah
do
we
still
have
any.
Is
he
working
on
anything
with
the
crash
crash.
A
Do
you
mean
support
for
grass
yeah
yeah,
you
know
pentel?
No,
we
don't
have
it,
but
I
don't
think
there
is
a
model
in
open
telemetry
for
reporting
grasses
correctly.
I
mean
you
can
do
that
you
can.
I
am
doing
that.
I
am
collecting
crisis,
but
not
using
open,
telemetry
and
reporting
them
in
in
some
tasks.
A
It
has
some
attributes
like
yeah,
like
the
logs
and
the
apple
log,
for
the
crash
and
things
like
that,
but
using
other
tools
yeah,
but
you
I
mean
you
can
do
that
and
record
as
an
error
and
provide
some
attributes
that
you
use
for
that.
But
I
don't
think
there
is
a.
There
is
a
respect
for
that.
B
No
but
again
this
should
be
across
right.
The
other
other
applications
can
also
crash,
and
you
know
maybe
things
like
node
or
you
know,
java
or
those
things.
Maybe
the
crash
files
are
probably
huge.
You
know,
but
at
the
same
time
you
know
there's
a
lot
of
information
in
that
as
well.
So
so
I
think
yeah
you
should
look
at
that
spec
as
well,
but
yeah.
I
think
so.
There
is
nothing
in
your
open
telemetry.
This
is
all
custom.
You
know
whatever.
A
Yeah,
this
is
custom
things
on
top
yeah
for
yeah
for
the
rest
of
things.
I
don't
think
there
is
much
yeah
much
happening
at
least
on
the
project.
Yeah
yeah,
the
I
can.
A
B
A
Yeah,
so
that's
what
it
does.
It
provides
some
updates
on
top
of
it.
So
you
add
to
the
meter
you
have,
you
can
add.
Instead
of
creating
the
instagram,
you
add
a
brow
double
instagram
that
includes
all
the
data
and
that's
the
the
change
on
the
api.
It's
it's
quite
yeah
just
that,
but
you
can
do
that
from
the
meta.
B
A
A
Yeah
I'm
having
like
lots
of
internal
workout
here
yeah
we
yeah
we're
working
on
some
things
that
are
presenting,
so
I
have
been
fully
dedicated
to
those
things
so
yeah.
I
have
not
had
time
for
for
adding
anything.
I
have
asked
for
time
for
working
on
open,
telemetry,
some
functionality
again,
but
probably
after
before
october.
A
Yeah
we
have
a
conference
in
october,
so
probably
until
then
we
I
won't
have
much
time.
For
I
mean
I
wouldn't
be
alone.
B
Yeah
anything
that
makes
sense
to
you
know
coded
in
open,
telemetry
right
that
makes
yeah
any
feature:
functionality
that
is
common
right.
A
Yes,
I
think
so
yeah,
I
think
so.
I
have
not
forgot,
I'm
using
still
the
good
one.
What
I
did
was
I
moved
to
another
exporter.
I
I
was
using
the
data
doc
exporter,
that
is
in
the
project
yeah,
but
I
have
to
move
to
another
exporter
because
I'm
using
other
things,
I
am
using
that
for
many
other
things
apart
from
open,
telemetry
traces.
Now
so
I
moved
to
I
mean
it's
open
source,
it's
based
on
the
same
source
code,
but.
A
Because
yeah
it
works
and
I
am
using
the
releases
from
there
and
it's
working
for
me.
Probably.
B
Yeah
yeah,
and
what
about.
B
A
Old
version-
I
don't
know,
okay,
I
I
think
we
can
see
here.
I
don't
know
if
he's
here
or
inside
or
something
like
that,
that
you
can
check
forks,
there's
wow,
we
have
52,
we
have.
A
A
A
Yeah,
I
don't
know
or
using
or
not
using
this
for
development.
I
don't
know
yeah
brace.
B
B
But
I
think
yeah,
as
we
add
more,
I
think
there'll
be
more
people
using
it.
A
B
A
Releasing
my
product
yeah
so.
B
No,
no!
No!
No!
No!
I
think
I
I
need
to
first
get
back
on
that:
okay
or
that
docker
image
right,
because
that
is
yeah.
That's
right!.
B
B
Yeah
yeah
and
then
yeah
then
at
some
point
we'll
add
logs.
A
Yes,
that's
right!
That
is,
I
mean
I
I
have
has
time
for
that
to
my
to
my
manager.
So
if
I
can
give
some
time
for
for
supporting
love,
I
know
that's.
There
are
many
features
working
on
logs,
especially
hotel
client
working
on
top
of
four
logs
with
the
spec.
So
that
would
be
great.
If
I
could,
I
mean
probably,
can
implement
that
in
a
week
or
something
like
that
now.
A
And
they
they
they
have
a
library
which
is
called
swiflox.
Okay,
that's
not
the
scene
that
uses
for
logging
and
the
system.
It's
something
crazy.
It's
like
the
apple
metrics,
it's
not
related
to
swift
metrics.
A
They
have
these
swift,
metrics
swift
logs,
probably
focused
on
the
server
side
of
swift,
and
they
have
something
that
is
totally
different
when
they
release
for
ios,
so
they
have
a
library
for
logs
the
unified
login
system.
They
call
it
like
that,
and
it's
not
related
to
suite
logs,
for
example,
so.
A
Very
powerful
you
can
capture
it,
you
can
send
it,
but
no
one
is
using
it.
So,
instead
of
the
server,
maybe
we
don't
have
any
user
who's
using
swift
on
the
server.
A
So
I
don't
know
if
we
should
really
spend
time
there,
because
people
is
basically
using
this
on
ios
yeah.
B
A
A
A
A
A
I
know
how
to
do
that,
because
I
am
really
doing
so.
I
could
write
that
easily
but
yeah
but
reporting
on
with
the
api,
something
like
that.