►
From YouTube: 2023-01-02 meeting
Description
cncf-opentelemetry meeting-2's Personal Meeting Room
C
This
time
I
had
to
work
so
I
was
sort
of
Uncle.
Oh
I,.
B
B
What
about
you,
I,
took
two
weeks
off
and
I
was
with
family
first
and
then
so
with
some
friends
after
that.
A
C
B
So
it
seems
it
was
very
quiet
on
open
Telemetry
this
this
winter.
C
Yeah,
it
was
very
quiet.
No
no
activity
and
I
didn't
find
time
to
also
contribute
to
the
issues
that
I
have.
D
B
B
Said
that
well
we
we
decided
last
time
in
in
December
that
we
would
cancel
two
meetings.
So
I'm
hoping
there
is
a
meeting
today
then.
B
D
C
B
B
C
C
So
let's
go
to
the
spec
changes.
It's
Jagger
duplication,
I.
Think
you
already
created
an
issue
for
that.
Yes,.
B
A
B
So,
from
a
technical
point
of
view,
deprecating
the
video
exporter,
which
would
be
easy
to
do
I,
guess
yeah.
The
question
I
have
is
on
in
general
for
deprecation.
B
A
C
A
D
B
Yeah,
especially
especially
if
there
is
a
long
time
between
the
announcement
of
a
deprecation
and
the
final
removal
like
if
we
say
we
duplicate
something
in
six
months,
people
should
know
that
and
they
should
not
have
to
read
every
change
log
for
for
six
release.
You
know
we
try
to
find
out
what
is
the
status.
C
A
B
C
Okay,
yeah.
We
also
need
to.
B
A
A
B
B
I
assigned
myself
to
to
that
already,
so,
if
it's
okay
I
can
propose
something
and
then
we
can
discuss
it.
Yeah.
C
Sure
sure
I
can
just
that's
PR.
A
A
C
C
So
the
next
one
shall
we
go
over
the
next
one
sure
yeah
and
internal
Matrix.
What
do
you
mean
with
the
internal
metrics.
B
So,
basically,
at
some
point,
I
will
need
to
go
in
production
with
traces.
So
assuming
you
have
open
Telemetry,
so
you
have
an
application,
sending
spans
to
open
Telemetry.
A
B
We
think
the
first
thing
that
may
happen
is
that
the
spine
is
either
collected
or
discarded,
because
we
have
some
Samplers.
We
have
always
owns
always
on
sampler.
We
have
some
sampler
sampler
with
a
probability
to
just
keep
the
keep
the
span
or
not,
and
then,
after
that
there
is
the
batch
processor
that
can
keep
spines
in
a
buffer
after
that,
it's
finally
exported
to
a
collectorally
the
downstream.
B
If
a
exporter
cannot
even
connect,
collect
sorry
connect
to
the
downstream
collector,
in
my
understanding
we
may
have
a
right
message:
envelopes
saying
it
cannot
connect
to
a
server,
but
the
span
is
discarded
and
from
an
external
point
of
view,
your
application
is
running,
but
you
don't
know
if
you
are
getting
all
the
traces
you
you
should
or
not.
B
How
many
time
we
connected
to
a
server
or
many
times
we
had
failures
to
connect,
for
example,
things
like
that,
so
that
those
metrics
themselves
can
be
exported
as
well,
so
that
you
know
when
monitoring
the
system
is,
at
least
you
know,
if
you
are
getting
all
the
choices
you
need
or
if
the
pipeline
is
broken.
At
least
you
know
where
it
is
popping
yeah.
C
Yeah,
yes,
yes,
exactly
I,
I,
think
I,
don't
know
if
it's
in
this
paper
or
not.
It's
probably
also
good
to
have
some
way
to
get
information
about
it.
Yes,
collector
backend
like
if
it's
actually
up
or
down
and
we
are
creating
spans
and.
B
Yes,
so
in
my
understanding,
it's
not
in
respect
and
also
I'm
suspecting
that
this
is
not
specific
to
C,
plus
plus,
so
any
language
and
every
any
exporter
will
face
the
same
issues
like
show
your
export
spans
in
in
Java.
But
how
do
you
know
we
are:
we've
actually
sent
Downstream
and
not
discarded.
For
example,
yeah.
A
B
Now,
but
I'm
suspecting
it
might
be
a
no
type
because
just
to
export
metrics,
for
example,
we
need
a
semantic
conventions
to
name
those
metrics
like
for.
If,
if
a
span
is,
is
sampled,
if
you
have
a
counter
for
the
number
of
spans
which
are
collected
even
about
spends,
which
are
sampled
and
and
whatnot
resolve
metrics,
which
have
semantic
conventions.
A
C
A
B
So
I
can
I
can
create
an
issue
and
start
to
add
some
some
notes
and
details.
I,
don't
know
the
so
I'm
assuming
we
need
to
discuss
that
with
Tom
and
editor
as
well.
But
the
main
thing
is
whether
we
can
do
that
only
in
in
open
Telemetry,
C,
plus,
plus
or
if
it
needs
to
be,
it
needs
to
have
a
bigger
scope
so
that
it's
consistent.
D
C
C
B
And
we
have
to
expose
something.
So
where
is
this
as
well?
In
my
application,
in
fact,
I
would
like
both
I
would
like
to
have
some
apis,
so
that
I
can
expose
voice
metrics
in
my
own
logs.
First
of
all,
and
once
we
have
that,
then
we
can.
We
can
expose
the
data
in
metrics
as
well,
so.
C
C
C
B
Think
we
should
go
in
that
direction.
One
one
good
thing
also
was
revert
is
that
it
will
be
also
very
good
to
test
metrics,
because
we
will
have
something
to
test
metrics
on
if
we
Implement
that,
yes,.
C
B
C
Yes,
yes,
that
that's
also
but
I,
think
the
I
mean
for
me.
The
mindset
is
that,
like
the
Telemetry
stuff
is
another
another
burden,
so
we
should
be
how
to
say
we
should
leave
the
list
cost.
C
B
C
C
C
Shall
we
go
for
the
next
one,
and
this
is
I?
This
is
a
PR
that
I
raised
in
country
people,
and
it
would
be
great
if
I
think
you
also
have
a
brewer.
B
Yes,
I,
so
I
spend
most
of
the
time
in
in
CPP
and
rarely
go
to
CPP
country,
but
maybe.
A
C
B
C
So
I'm
working
on
some
I'm
trying
to
open
a
new
tab.
Okay,.
C
Oh
yeah,
there's
yeah.
C
So
I'm
implementing
this
feature
for
Ingress
nginx
Depot.
So,
basically
anything
that
is
behind
Ingress
can
be
traced,
and
so
this
is
a
issue
for
a
very
long
time.
C
It's
open
since
September,
okay-
and
there
were
several
issues
and
several
PR's
on
this,
so
I
had
to
make
the
bills
with
engineers
and
stuff,
and
there
is
an
image
and
for
that
and
so
on
what
it
comes
at
the
end
is
you
you
get
the
Telemetry
for
anything
that
is
behind
Ingress
and
for
that
one
I'm.
Using
this.
C
This
is
barely
maintained
now
and
but
it
should
be
good.
I
I
think
that
there
should
be
no
issue.
Just
no
new
features
are
coming
and
until
now
that
there
is
the
issue
that
came-
and
that
says
these
attributes
are
not
populated
correctly,
when,
when
the
new
grpc
is
used,.
C
I,
don't
think
so.
These
are
developed
very
long
before,
okay,
the
semantic
stuff,
but
people
seems
to
like
it.
They
won't
have
it,
and
it's
also
semantic
ones,
also
comes
in
I,
think
and
anyway.
So
to
fix
this
I
found
that
I
need
to
change
the
build
to
not
to
use
app
sale.
C
So,
basically,
the
issue
in
country
people
I'm
trying
to
fix,
is
I'm
trying
to
enable
this
population
of
attributes
correctly.
Okay,
it's
shown
here.
Well,
it's
it's
like
this.
With
the
current
build,
I
mean
not
build
with
the
build
that
I
have
in
Ingress
repo,
and
the
reason
that
I
I
cannot
change.
That
is
because
of
the
app
sale,
a
failure
in
grpc.
A
C
So
what
I'm,
basically
I
need
is
I
just
need
to
move
this
header
to
to
the
first
line,
so
grpc
is
included
before
any
other
API.
Oh
yeah,.
B
C
So
with
this
change
I'm
enabling
shutting
down
the
upsell
from
from
from
opentility,
so
we
don't
need
to
use
upsell
anymore,
okay,
so
you
can
build
it
and
at
the
same
time,
I
also
updated
the
grpc
version
here
and
the
old
open,
Telemetry
version.
So
we
we
could
use
the
latest
one,
which
is
then
this
one
goes
to
the
CI.
So
the
latest
versions
are
tests
and
all
these
attributes
that
I
showed
you
they
are
tested
in
the
in
the
CRI
okay.
C
A
B
B
On
on
the
side,
notes:
I,
really
like
your
your
theme
and
color
screen
as
colors.
It's
much
easier
to
read
compared
to
when
lalit
is
sharing
his
screen.
C
You
mean
the
font,
size
or
called
I.
Don't
know
the
colors.
C
B
Not
really
well,
no,
nothing
new.
The
only
thing
I
just
added
is
for
the
verification,
but
we
just
discussed
that
already.
B
C
C
A
B
A
Future
because
next
comparison
practice
like.
B
C
Progress
yeah
it's
in
progress
and
it's
only
opened
mostly
I.
Think
yes,
so
we
are
a
bit
slow
there
and
we
also
have
many
issues
on
metrics
and
the
logs
aspect
is
experimentaling.
It.
A
B
C
C
And
don't
react
on
the
others,
so
this
one
should
also
go
to
me.
No,
you.
B
One
is
about
basically
making
a
super
project
using
open,
telemetry
I'm,
still
struggling
with
that
in
with
cmeg,
because
my
own
application
is
also
super
projects,
including
a
lot
of
things,
including
open,
telemetry
and
I've,
not
seen
anywhere
some
clear
guidelines
about
whoever
we
should
do
add,
subdirector
or
NC
make
or
whether
it
should
be
an
external
projects.
A
C
Yes,
this
is
the
Zone
I
think
again,
I
I
this
here,
I
am
I,
was
almost
there
with
the
super
project.
B
I
think
I've
seen
I
think
I
went,
has
done
something
or
so
I,
don't
remember
if
the
URL,
but
is
he
has
done
something,
but
those
are
built
with
all
the
dependencies
of
job
PC
and
open
Telemetry
itself.
C
A
C
C
C
D
B
We
were
looking
at
and
luminations.
B
Do
you
know
on
on
the
next
one
shared
pointer?
Do
you
know
the
full
story
about
the
ABI
compatibility
and
why
we
use
STD
sometime
and
no
STD,
sometimes.
C
B
A
C
I
could
go
to
the
next
one.
Please
add
example:
how
to
use
package.
A
B
Oh
so
this
one
is
mine,
so
the
story
is
when
doing
some
tests
on
the
so
I'm
working
to
implement
SSL
on
the
HTTP
Explorer
and
when
testing
Buds
and
using
the
HTTP
examples.
B
I
noticed
some
crashes
all
that,
due
to
the
order
in
which
we
construct
single
terms
and
the
order
in
which
we
delocate
them-
and
it's
it's
a
bit
unclear
what
the
proper
fix
is,
whether
we
just
do
some
manual
cleanup
and
manual
well
manually,
initialization
and
manual
Cleanup
in
all
the
or
the
application
side,
including
an
example,
or
if
we
should
have
a
more
controlled
way
to
initialize,
open
Telemetry,
elaborate
itself,
broken
Telemetry
Library,
so
other
PR,
but
I
just
changed
it
to
drafts
because
it
needs
to
have
a
description.
B
B
A
typical
application
will
initialize
the
trace
provider
at
the
beginning
of
the
main,
but
if
you
leave
it,
as
is
the
Singleton
for
the
retrace
provider,
is
still
pointing
to
something
preventing
the
allocation
so
on
shutdown,
what
I
added
is
some
code
to
reset
the
Tracer
provider
to
null
so
that
when
everything
is
located
properly
before
the
application
should
read
returns
from
the
main
itself,
it's
it's
one
way
to
do
things,
but
it's
it's
more,
a
workaround
to
avoid
the
12
over
drops
inside
for
contemetery
level
itself.
A
A
B
Audio
resume
work
on
that
I
just
did
that
in
mid-December.
Also
I
will
merge
with
the
latest
code
and
and
take
a
look
at
that
again.
Okay,.
D
B
A
B
So
this
one,
so
we
are
waiting
on
Syria
I
added
that
tag,
because
the
Siri
is
not
signed.
We
change
it
in
C,
make
sense
to
me:
I
mean
if
we
say
we
have
a
required
package,
then
what
is
the
point
to
to
find
to
test?
If
it's
found
or
not
it's
it's.
C
B
Also
need
to
do
windows
build
myself
and
I've,
not
even
started
yet
on
that
I
don't
quite
understand
what
we
this
vpkg
is
I
mean
at
work.
We
also
have
Windows
build,
but
we
don't
do
it
the
same
way.
So
I
have
to
investigate
exactly
what
other
dependencies
were.
C
B
Specification
this,
so
it's
a
big
one
from
moment:
Freddie
files
changed.
A
D
A
A
A
C
So
I
mean
we
don't
have
access
to
registry,
okay,
so
I'm
trying
to
somehow
store
a
Docker
image.
That
has
everything
that
we
need
for
build
and
the
way
I'm
doing
it
is
I'm
using
this
artifacts
from
GitHub
actions,
so
I
store
the
docker
image
as
entire
file
in
GitHub
actions
artifacts
and
in
the
next
build
you
just
download
that
artifact
and
on
turret
do
you
load
the
image
and
everything
is
there,
and
this
makes
the
builds
with
grpc
like
two
weeks
faster.
C
So
this
is
the
whole
thing
that
this
PR
trying
to
do
I
mean
this
PR
prepares
everything
so
and
if
this
PR
got
merged
with
the
next
PR,
I
can
start
using
that
image.
B
C
A
B
A
A
B
The
HTTP
exporter,
we
don't
support
SSL
yet
so
this
is
an
attempt
to
to
fix
that.
C
B
I
mean
overall,
it
looks
okay,
given
what
you're
trying
to
do,
but
as
far
as
detailed
code
review
on
every
line,
I'm
not
familiar
enough
with
we've
based
on
and
how
it
works.
B
Pushed
yeah
he
woke
up
and
did
a
merge,
but
nothing
happened
after
that.
B
It's
your
well
me,
but
it
was
a
minor
comment,
but
the
main
thing
about
similarly
to
object
to
as
well
is
he's
using
the
portable
buffers
as
a
way
to
expose
an
interface
to
the
application
to
store
things
in
memory.
So
it
worked
complexity.
There.
B
A
D
B
Here's
well
not
only
that,
but
Jan
was
I
mean
he
attended
to
C
plus
first
meeting
at
some
points,
and
it
was
nice
talking
to
him,
but
when
he
fell
off
right
or
for
three
months
or
at
least
three
months
I,
don't
remember.
Also
it's
it's
not
hard
to
tell
whether
he
will
continue
to
work
on.
B
B
D
C
I
think
one
way
is
I,
don't
know
if
you
should
reach
them
directly,
I
mean
if
they
attend
the
meetings.
We
could
ask
them.
C
That's
that's
a
good
point.
Maybe
we
we
could
think
about
asking
some
of
these
people
or
filing
issue
very
often,
but
I
I
think
Planet
already
doing
that
somehow
like
he
encourages
raising
PRS
and
so
on.
B
Well,
it's
I
mean
compared
to
all
the
other
projects
in
in
open
Telemetry
like
Java
or
Google
or
The
Collector
I
mean
I.
Guess
we
have
less
people
involved
in
in
the
C
plus
plus
part.
So
it's
more
a
niche,
but
the
people
who
are
interested
in
and
including
the
the
people
who
file,
defects
and
debug
things.
They
seem
to
be
very
competent,
competent
as
well
in
general,
so
I'm
hoping
we
can
still
attract
some
some
people
who
are
motivated
to
to
contribute.
C
Yes,
yes,
yes
and
let's
let's
discuss
this
in
the
next
meeting-
maybe
we
can
sure
definitely
I
I
think
there
are
two
three
people
that
are
creating
really
good
issues
with
all
description
and
so
on,
and
they
should
be
very
knowledgeable
and
you
could
benefit
from
there
contribution.
Definitely,
but
how
we
approached
him,
I
I,
don't
know.
C
Maybe
we
can
also.
We
could
also
use
the
slack
Channel.