►
From YouTube: 2023-03-07 meeting
Description
cncf-opentelemetry@cncf.io's Personal Meeting Room
A
A
C
C
Yeah
the
one
I've
been
getting
lately
is
like
it
just
says,
like
you
can't
unmute,
the
host
doesn't
allow
it
like,
which
is
bizarre,
because
I've
gotten
in
and
I've
like
found
the
host
key
and
I've
like
assumed
the
host
and
I've
been
like
okay.
Now,
let's
allow
people
to
unmute
and
it's
like.
No,
you
still
can't
do
it
so
like
there's
just
something:
there's
a
bug:
I,
don't
know
what
it
is.
C
C
Yeah
I'm,
looking
at
the
agenda,
it
looks
like
there's
a
few
things
on
there
right
now
and
looks
like
because
I
can't
actually
attend
usage,
section
yeah,
okay,
yeah.
We
should
talk
about
that
yeah,
okay,
we
could
probably
just
jump
in
I
guess
for
three
minutes
past
the
hour
I
see
Dinesh
is
here
as
well
yeah.
Let's
do
that!
So
welcome
everyone.
C
If
you
haven't
already,
please
add
yourself
to
the
attendees
list
and
if
you
have
other
things
you
want
to
talk
about
on
the
for
the
agenda
today,
please
Adam.
We
can
also
pause
at
the
ends.
C
Okay,
there
we
go
okay,
cool,
so
the
first
thing
Tyler
helmuth.
He
actually
came
to
the
go
Sig
meeting
this
past
week,
there's
an
open
issue.
He
has
here
to
publish
a
Docker
container
for
this
project.
I.
Think,
let's
see,
we've
got
I
think
the
requisite
number
of
approvals
here.
The
only
open
question
that
he
had
was
naming
I,
don't
know
if
it's
been
resolved
yet
looks
like
this:
okay
Auto
instrumentation
go
yeah,
it
does
look
like
it
was
actually
resolved.
C
Okay,
cool
yeah,
I
think
the
only
other
thing
was
just
to
write
it
by
this
community
as
to
like
what
we
wanted,
the
docker
name
to
be
I.
Don't
know
the
full
process
of
publishing
this
I
think
it's
all
set
up
given
like.
D
Yeah
I
opened
an
issue
for
T
Grant
I,
think
from
the
technical
committee.
D
The
required
credentials
to
the
repository,
so
it
should
work
right
now,
I
think
that
I
already
tested
it
out
yeah.
It
looks
good
to
me
I
think,
like
they
had
some
discussion
here
about
whether
they
want
to
push
it
to
Docker,
Hub
or
GitHub
container
registry
I.
Don't
really
know.
D
Yeah
I
think
it's
like
about
the
like.
The
other
instrumentation
like
I
think.net
is
already
doing
it,
and
the
collectors
and
stuff
I
guess
it's
already
published
Docker
hub,
but
maybe
it
will
be
a
good
idea
to
like,
like
set
some
standard
policy
across,
like
all
other
organizations
to
like,
did
we
like?
Where
should
we
push
our
images
and
like
what
should
be
the
name
but.
C
Yeah
yeah
I
think
you're
right.
I
know
that
there's
currently
a
few
there's
like
I
mean
obviously
I.
Think
the
collector
is
the
one
that
stands
out
as
like
the
most
important
and
that's
in
Docker
Hub.
It
does
look
like
they
put
it
to
GitHub.
Actually,
if
I'm
reading
this
right,
okay
I
feel
like
this
might
be
a
situation
where
we
could
do
both
but
I,
don't
know
the
implications
of
that.
That
doesn't
seem
I.
C
Think
I
would
have
rather
got
it
the
one
that
we've
already
all
used,
but
it
looks
like
they've,
got
GitHub
tokens.
Okay,
maybe
this
should
work.
We
could
try
it.
We
could
always
move
it.
I
guess
the
other
side
of
things,
yeah,
yeah
well,
wait.
I
might
be
doing
both
I,
don't
know,
and.
A
I
don't
know
but
anyways.
C
I
think
this
looks
good
I.
Think
Robert
has
the
only
other
outstanding
issue
here:
give
us
a
pin
the
tag,
I,
think
that
seems
fine,
okay,
but
other
than
that.
I
think
that
we
could
probably
get
this
merged
soon.
If,
if
not
right
away.
C
Oh,
that,
whatever
okay
cool
so
all
right
action
item
on
that
I
think
it's
just
emergent.
I
think
it's
looks
ready
to
go
cool
next
item
on
agenda
Nicola.
You
had
something
here.
It
was
pop
head
over.
You.
B
Yeah
thanks.
Let
me
first
introduce
myself
I'm,
actually
new
to
grafana
started
a
week
ago
and
we're
creating
a
new
team
and
grafana
that
would
be
primarily
working
on
instrumentation
for
evpf
or
open
source,
fully
committed
to
this
and
Mario,
as
he
actually
submitted
a
pull
requested
too
I.
Think
of
stuff
that
he's
been
working
on
he's
the
other
team
member
that
works
with
me.
B
Unfortunately,
for
him,
it's
late
in
the
evening
he's
in
the
European
Time
Zone,
so
Central
European
Time,
so
he
can't
attend,
or
maybe
he
will
one
day
but
I
decided
to
come
in
and
just
ask
the
basic
question
we
like
to
participate
and
contribute
to
this
project,
but
we
also
found
out
that
hewell
who
actually
have
originally
donated
this
still
can
Telemetry.
You
have
your
own
repo,
where
things
have
kind
of
progressed
a
little
bit
faster.
B
B
How
can
we
contribute
and
help
and
so
on?
And
you
know
to
avoid
duplicating
any
work,
that's
sort
of
like
a
newcomer
question.
If
you
will.
D
Yeah
sure
I
can
take
this,
so
I
figure
like
this
whole
project,
started
from
the
nation
from
Cuban
to
open,
Telemetry
community.
D
So
like
the
reason
for
the
keyword,
repository
is
mostly
because
they
think
we
are
still
a
new
group
and
I
was
still
trying
to
figure
things
out
together
and
like
we
wanted
to
continue
the
development
while
sorting
things
out
but
okay
like,
but
eventually
everything
should
come
down
to
the
open,
Telemetry
Rapunzel
like
it's
just
like
like
for
now
that
we
there
are
some
things
in
Cuba
that
are
not
in
the
open,
Telemetry
a
repository,
okay,
hopefully
I'm
hoping
to
actually
do
it
like
this
week,
like
I,
think
the
main
things
that
is
in
keval
and
not
in
the
open,
Telemetry
repositories.
D
I
think
it's
the
arm
support
for
one
and
like
the
the
new
way
to
do
like
a
location
without
the.
D
Walk
around
that
we
had
in
the
past,
so
these
two
things
they
it's
it's
unmade,
hopefully
like
like,
like
I,
think
the
end
of
the
week
I'll
be
happy
to
I,
hope,
I'll
be
able
to
match
them
to
to
the
upper
Telemetry
repository,
but
as
a
general
I
think
like
most
of
the
work
and
like
like,
is
just
like
being
done
in
the
up
until
I'm
free
positive.
That's.
B
That's
just
wanted
to
find
out,
because
we've
had
this
question
and
it's
like
yeah
yeah.
We
saw
some
fixes
over
there.
It's
like,
oh,
that
seems
to
be
like
fix
this
issue
that
we
found
as
well
and
what's
up
it's
great
and
we
still
have
to
figure
out
our
plan
and
whatnot.
We're
still
very
new
group,
especially
both
of
us
grew
new
to
propana,
but
also
group
new
to
to
this
as
well
so
but
yeah
we
intend
to
participate
and
I
contribute
this
message
again:
yeah.
C
Yeah
that
sounds
good
thanks,
Nicola
and
welcome
by
the
way.
Thank
you,
yeah
yeah
I
would
also
add,
like
the
the
more
contributions
we
can
get
into
the
opens
on
the
true
side
of
things,
I
think
the
the
more
confidence
we
have
in
speeding
up
development
there,
like
you're,
just
gonna,
you
know
help
the
process
of
moving
everything
over
quicker,
so
I
think
that's
a
great
great
thing
to
see
like.
C
We
definitely
welcome
the
contributions
and
I
definitely
know
that,
like
Dinesh
has
done
a
lot
of
work
in
trying
to
just
self-evaluate
I
think
that's
a
great
place
for
you
all
to
start
as
well.
Just
so
I
could
you
know,
put
it
out
there
just
take
the
thing
and
try
to
run
it,
because
it's
still
like
it's
still
prototyping
it
so
I
think
that's
a
great
place.
Yeah.
B
Yeah
Mario's
been
doing
a
lot
of
that.
Like
I,
said
I
started
a
week
ago,
not
even
control
a
week,
but
yeah.
B
Yeah
there's
another
one:
I
think
he
he
I
remember:
there's
there's
definitely
one
where
I
think
around
the
launcher.
That's
for
some
parts.
The
launcher
is
not
required.
The
keymon
launcher
so
I
think
he
might
pull
request
around
that
I,
don't
know,
I
think
he's
made
another
one
on
my
page.
D
Yeah
I
reviewed
this
book
because
I
think
that
the
main
change
that
he
does
as
it
is
mostly
just
not
panic
because,
like
we
replace
like
the
like
before
this
pull
request,
if,
if
the,
if
the
Venture,
if
the
launcher
did
not
like
ran
and
executed,
the
target
process
like
the
instrumentation,
would
panic
and
he
added
that
it
would
not
panic.
It
would
just
affect
the
effect
of
running
the
instrumentation
without
running
the
launcher
before
it
is
just
at
the
context.
D
Propagation
will
not
work
because,
like
that's
that's
the
reason
we
need
to
launch
upon
doing
the
context
propagation
but
like
yeah,
but
hopefully
like,
like
I,
got
a
much
better
way
to
do
it.
If
you
wanted
to
okay,
I
can
share
the
screen
and
show
it
to
you,
but
yeah,
but
I
saw
like
we
have
a
couple.
Other
open
pull
requests
that
maybe
we
should
review
and
add
marriage.
A
C
I
was
kind
of
thinking
that
as
well.
Maybe
we
can
just
kind
of
go
through
that
right
now,
yeah.
C
Cool
boy-
let's
see
here
so
I,
think
starting
at
the
bottom.
We
have
this
automatic
offset
generator
for
the
CI
job.
I
thought
that
this
is
oh
yeah
Damien.
This
looks
good
yeah.
This
is
a
good
merge.
So
that's
on
me.
So
that
looks
great
next
up.
I
think
is
this
remove
depends
in
the
key
value
launcher.
So
this
is
the
one
we
were
just
talking
about
right,
yeah,.
B
D
C
Yeah,
okay,
I,
don't
know
if
we
can
split
this
to
have
the
demo
be
its
own.
Pr
would
help.
But
I
don't
know.
B
C
C
Okay,
cool
and
then
they're
just
gonna,
add
that
other
feedback,
but
other
than
that
like
this
did
look
pretty
simple
outside
of
the
example.
So
yeah
I
think
that
I'll
try
to
review
this
actually
afterwards,
it
doesn't
look
too
bad.
C
Cool
this
is
the
pr
for
the
Docker
image
we
just
talked
about
I.
Think
Heather
did
actually
respond
to
this,
but
yeah.
This
looks
ready
to
merge.
C
Up
this
is
results.
C
Actually,
I
think
we
could
just
do
this
really
quick
I,
don't
think
Tyler's
gonna
care
that
it's
pinned
to
a
particular
version.
C
Branch,
okay,
yeah,
this
one
I
think
looks
ready
to
go.
It
will
then
merge
that
after
it
finishes
its
CI
job,
next
up
is
the
fix.
An
HTTP
instrumentation
for
register
based
ABI
is
another
one
for
Mario
I.
Think
right.
Yeah.
C
Has
anyone
had
time
to
take
a
look
at
this
one?
Yet?
Okay,
we
don't
need
to
I
think
do
code
reviews
in
the
meeting,
but
yeah
90
lines.
I
think
this
is
actually
pretty
straightforward.
Okay,
oh
it's
just
offsets.
C
Yeah,
okay,
so
yeah
Eden
walking
through
the
timeline
on
that.
So
what
are
we
waiting
for
on
the
offset
generation
like
the
automatic
side?
Are
you
just
waiting
to
open
a
PR
still
or
no.
C
B
C
Let
me
share
again
here:
yeah
I
thought
that
it
well
I
mean
it
was
merged,
but
yeah.
This
might
just
need
to
I,
think
I
think
the
assets
might
actually
might
be
able
to
be
removed
from
this
right
away.
So
I
think
that's
that's
fine,
yeah
cool
and
then
this
one
I
just
saw
this
morning
him
in
not
exactly.
That
is
what
this.
D
D
He
found
that,
like
in
previous
Go
version
like
you
had
this
go
mode
vendor
we're
gonna
use
like
two
longer
dependencies
into
a
vendor
directory
and
when
you
do
that,
like
the
name
of
the
functions,
are
not
the
names
of
the
instrumentation
like
currently,
we
just
do
3D
quality
between,
like
what's
the
name
of
the
function
in
the
binary
to
what's
the
name
of
the
functions
that
we
expect
from
the
offset
Json
file
and
when
you
use
the
go
mode
vendor
mode
like
the
name
of
the
function
contains
like
the
whole
path
to
the
vendor
directory.
D
C
C
Yeah,
okay,
all
right,
yeah
I,
think
this
I
think
this
looks
great
too.
It
just
needs
some
more
eyes
on
it.
I
think
at
this
point,
yeah
cool,
well,
I.
Think
with
that
yeah.
This
looks
good
to
go
as
well.
C
I
think
that's
all
the
poll
requests
that
are
open,
so
it
just
needs
more
eyes
at
this
point.
I
guess
is
what
the
the
answer
there
is.
A
C
Okay,
cool
back
to
the
agenda
I
know
president
can't
attend,
but
he
was
asking
how
we
can
move
this
forward.
I
think
this
again
is
a
PR
in
the
contributive
Instagram
implementation.
Dinesh
is
taking
a
look
at
this.
I
still
have
taken
a
look
at
this
multiple
times,
but
I,
don't
I,
think
I
think
there's
been
a
update
recently.
E
C
Yeah,
yes,
I
agree:
I
am
as
well
also
for
Go.
Metro
kpi
is
solidifying
by
the
way,
just
totally
related
engine
yeah,
but
yeah.
Okay,
cool
I
think
that
that
sounds
good.
It
just
needs
more
eyes
on
this
as
well.
So
let's
I
think
plan
to
get
some
more
eyes
on
all
these
PRS
I.
Think
that's
the
only
action
item.
We
have
yeah
cool
all
right,
so
I'll
pause
here
see
if
we
want
to
talk
about
something
else,
that's
not
on
the
agenda.
D
True,
let
me
let
me
show
you
so
you
can
see
that
so,
basically,
the
main
changes
that
we're
doing
instead
of
like,
like
previously
like
the
instrumentation
agent,
looked
for
the
all
the
mapped
memories
of
the
target
process
and
looked
for
a
specific
areas
where
it
was
allocated
by
the
launcher
program
that
launched
the
original
process,
but
I
just
changed
it
to.
Basically,
there
is
an
option
to
do
like
a
remote
you
can
actually
allocate
if
you
have
enough
permissions
in
line
hotel,
we
are
because
we
are
running
vbps.
D
We
have
enough
permissions,
you
can
actually
do
like
a
remote
and
map
for
a
different
process
so
like,
as
you
can
see,
we
can
actually
be
unable
to
do
that
too,
and
and
like
that's
another
cool
thing,
because
it
actually
means
that
we
can
now
take
a
running,
go
process
and
instrument
it
without
having
to
restart
the
process.
So
it's
even
more
powerful
now
so,
basically
yeah
we
have
this
function.
We
know
that
map
which
basically
returns
us
the
memory
address
of
well.
D
We
should
like
just
to
recap:
we
need
this
area
of
memory
because
we
are
generating
basic
identity
from
the
Canon
from
evpf,
so
we
need
to
place
them
somewhere
from
the
so
they
go.
Runtime
know
how
to
read
them
and
put
them
in
you
know:
HTTP
headers
or
GPC,
headers
or
basically
for
doing
contact
propagation.
So
yeah,
that's
the
main
choice.
The
remote
function
itself
is
it's
based
on
like
it's.
It's
basically
like
a
it's
a
function
that
you
can
call
and
do
an
app
map.
D
D
To
this
area
of
memory,
so
it
will
not
be
cleaned
like
we
clean
it
ourselves
at
the
end
of
every
ebpf
program,
so
that
if
we
have
like
a
instrumentation
for
an
HTTP
call,
we
will
write
the
trace
ID
and
send
it
to
this
memory.
And
after,
like
the
after,
we
close
the
span
after
the
spend
it's
finished.
We
will
clean
this
president
from
the
memory,
so
we
we
actually
do
the
garbage.
It's
not
a
garbage
commercial
Pro,
Within
evpf.
C
Is
this
dedicated
memory
to
the
process
like
how
is
this
not
going
to
collide
with
the
memory
space?
That's
allocated
to
go.
E
Did
you
test
the
limits
of
like
if
you
it
was
threshold,
this
memory
might
need
to
be
increased
or
something
like
that,
because
you're
putting
a
span
ID
and
request
ID
and
I'm,
assuming
there
should
be
a
limit.
D
Yeah
so
yeah,
so
there
is
a
limit
right
now.
I
think
it's
like
again.
It's
it's
a
pretty
easy
calculation,
because
I
think
10
ID.
If
I
remember,
because
it's
like
fixed
in
bytes
and
Trace
ID
is
32
bytes
or
maybe
I'm
confusing
the
numbers,
but
yeah
we
can
take
those.
We
can
take
those
and
multiply
it,
but
basically
the
number
of
concurrents
spans
that
we
are
want
to
support
and
that's
that's
to
be.
That
would
be
the
amount
of
memories
that
we
need
to
allocate.
E
C
D
C
C
All
share
that
yeah
cool
all
right,
well,
I
think
with
that,
then
anything
else
we
want
to
talk
about
before
we
end
it.
C
Okay,
yeah
I'm
excited
to
go.
Do
some
reviews
so
yeah
we'll
end
here
and
we'll
see
you
all
in
two
weeks
or
a
second
or
stay
online
yeah
all
right.