►
From YouTube: 2022-07-07 meeting
Description
Instrumentation: Messaging
A
B
A
You
in
a
recording
studio,
I'm
in
a
co-working
space
and
they
have
those
like
little
rooms
for
meetings.
Oh
cool.
B
That's
a
yeah:
it's
been
a
while,
since
I've
been
in
an
office
like
that,
but
those
are
great
little
places.
B
So
anthony
I'm
pretty
sure,
aaron
had
said
that
he
wasn't
going
to
be
able
to
make
it
right.
C
B
Okay,
yeah.
I
think
we
can
probably
get
started
here,
so
we
had
to
start
it
off.
Welcome
everyone.
My
internet
connection
is
a
little
shaky
today.
I
just
remembered
I
want
to
start
off
by
that.
So
if
I
do
drop
off,
maybe
anthony
can
just
pick
up
wherever
I
leave.
Hopefully
it
works.
B
Okay,
so
if
you
haven't
yet,
please
add
your
name
to
the
attendees
list
and
anything
you
wanted
to
talk
about
to
the
agenda
and
we
can
jump
in
here
just
to
start
it
off
also
another
announcement
I'll
be
out
next
thursday,
friday,
so
I'll
miss
the
meeting
anthony.
Are
you
okay
with
running
the
meeting
next
week.
C
Yeah,
I
will
be
around
next
week,
the
week
after
I
may
or
may
not
be
around
okay
to
be
decided.
B
Well,
I
will
be
so
that
should
work,
okay,
perfect,
we
will
keep
it
on
the
books.
Then
then,
next
up,
I
wanted
to
jump
into
the
metrics
sdk.
I
went
through
the
project
board
today
and
tried
to
clean
some
things
up.
It
looks
like
we're
progressing.
B
I
I
think
we're
actually
getting
a
more
accurate
number,
as
I
think
is
always
the
case.
These
numbers
like
right
about
this
point.
I
think
you
have
a
pretty
accurate
number
and
then
you
get
to
the
end
and
then
that
last
issue
takes
you
way
longer.
You
think,
but
other
than
that
I
think
we're
doing
pretty
good
the
project
board
itself.
We
can
take
a
look
at
it.
I
closed
a
bunch
of
issues
this
morning,
just
because
our
design
of
the
package
structure
and
layout
has
changed
from
the
demo.
B
Essentially,
so
we've
got
a
fair
amount
closed
that
were
essentially
documentation
or
related
to
old
package
structure.
This
is
the
current
work.
That's
in
progress.
Aaron
definitely
has
this
export
pr
opened.
This
telemetry
pipeline
structure
is,
I
think
I
was
looking
at
this
one
before
I
left
it's
going
to
be
a
big
one,
but
the
other
side.
B
This
is
essentially
like
the
you
come
in
from
the
the
meter
and
how
does
the
meter
keep
track
of
all
the
instruments
and
how
those
instruments
get
routed
back
to
the
readers
themselves?
So
there's
a
lot
of
complexity
there.
I
think
that
we
can
probably
do
a
little
better
job
explaining
that
in
this
ticket
it's
it's
got
a
lot
of
conversation,
but
I
think
that
there
was
like
a
left
off
yeah
like
this
pipeline
thing.
So
there's
I
think
some
more
design
work
to
go
here.
B
This
is
probably
an
area
that
you
could
see
a
little
bit
more
explanation
and
breaking
out
of
actual
work
into
the
project,
but
other
than
that,
I
think
that
it's
a
pretty
good
overview
of
what
we
actually
currently
have.
The
other
thing.
That's
in
progress
is
now
that
we've
merged
this,
like
internal
aggregator
package
or
aggregator
type.
We
can
start
implementing
these
aggregators.
B
I
picked
up
the
sum
portion
right
here,
but
we
still
have
the
last
value
of
the
delta
histogram,
the
cumulative
histogram,
the
histograms
are
going
to
be
probably
kind
of
complex.
I
don't
know
you
know
when
we'll
get
to
them,
so
I
think
it's
it's
a
it's
a
good
overview.
I
think
where
we're
at
there's
documentation
work
for
the
views
and
the
metrics
package
to
be
done
this
one.
B
Probably
we
could
wait
a
little
bit
just
because
I
think
it's
not
really
complete,
but
I'm
sure
we
could
add,
you
know
always
add
things
and
then
again
kind
of
like
with
this
pipeline,
these
stub
async
and
synchronous
instruments.
We
need
to
like
do
some
sort
of
registry
where
we
keep
track
of
them
and
have
them
work
back
into
this
pipeline.
So
there's
a
connected
nature
to
these
three
issues,
so
I
think,
having
a
good
layout
is
probably
a
good
next
step
coming
up
in
the
future,
but
other
than
that.
B
B
So
I
can
pause
here
and
see
if
there's
any
comments.
B
It
doesn't
look
like
it
okay,
so
we
can
jump
back
in
to
the
agenda
next
up.
I
saw
this
in
the
slash
channel
anthony,
I
think
we're
gonna
bring
it
up
as
well.
There's
this
question
of
a
release,
the
main
branches
of
both.
That's
not
what
I
wanted
to
do,
both
the
contrib
repo
and
the
hotel
repo
have
additions
that
could
be
released.
I
think
that
it's
probably
worthwhile
that
it's
been
a
few
months
since
we've
actually
done
the
release.
B
So
I,
if
there's
any
other
opposition,
I
think
that
we
should
try
to
prioritize
that.
C
And
yeah,
I
agree:
it's
been
a
little
over
two
months
since
we
did
a
release.
There
wasn't
a
whole
lot
that
had
been
done
since
then,
because
a
lot
of
the
work
has
been
happening
on
the
the
metrics
alpha.
Obviously,
but
I
did
merge
in
a
number
of
outstanding
pr's
yesterday.
So
there's
there's
a
few
more
things
queued
up
for
release
now,.
B
Yeah,
I
agree.
Additionally,
I
think
yeah
auto
prop
is
the
thing
that
was
under
question:
damian,
we
just
merged
in
a
pr
for.
B
Yeah
not
on
the
contrib
repo,
it's
a
little
bit
looser
here,
but
we
should
probably
add
something
here.
I.
A
B
Yeah
that
sounds
good
we've
kind
of
just
ad
hoc,
taking
it
as
a
responsibility
of
the
releaser
to
try
to
do
a
little
audit,
but
that
doesn't
always
happen.
So
I
think
that
your
suggestions
are
are
welcomed,
so
yeah.
That
sounds
good.
I
will
do
that
tomorrow.
Okay,.
C
C
B
Okay,
okay,
cool
yeah:
I
will
take
a
look
at
this
as
well.
B
C
There's
so
much
going
on
that
rebasing
or
merging
in
change.
Log
conflicts
becomes
a
problem
for
people
with
prs.
C
B
C
It's
it's
been
a
thing
that
we've
run
into
in
the
past,
and
the
discussions
come
up,
but
it
wasn't
enough
of
a
problem
for
us.
I
think
if,
if
this
works
out
for
the
collector-
and
they
go
down
this
path,
doing
it
for
consistency
and
to
be
prepared
for
a
future
where
it
may
become
more
of
a
problem
might
be
good.
B
I
am
worried
because
they
do
have
the
distributed
make
targets,
but
we
can
address
that
when
we
get
there,
I
guess
okay,
cool
yeah,
thanks
for
the
suggestion
on
that
one,
I
think
that's
a
something
I
think
also
the
ownership
structure
and
the
collector
contrib
is
something
we
need
to
look
at
for
the
arkansas
as
well,
but
I
think
that's
a
topic
that
just
needs
somebody
to
find
the
time
to
do
it,
because
I
don't
have
that
time
right
now
and
I
think
nobody
does
so.
B
Yes,
okay,
that
is
up
for
the
release,
so
anthony.
Is
that
something
that
you
could
do,
or
should
I
plan
to
do
this
tomorrow?.
C
Yeah
I
can
take
that
off
tomorrow.
Okay,
let
me
yeah
my
calendar's
fairly
open
tomorrow:
okay,.
B
Okay,
perfect,
I
will
be
online
as
well
to
do
reviews,
okay,
so
another
thing
that
came
up
was
recently:
we
saw
evpf,
auto
instrumentation
here
in
the
sig
meetings,
a
few
months
back
and
it's
being
proposed
to
submit
it
as
a
donation
to
the
open,
telemetry
community
at
large.
I
think
it's
more
of
a
standardization
similar
to
what
other
languages
have
like
java
or
python
without
an
interpretation.
B
So
that
being
said,
there's
a
community
proposal
here
I
have
not
been
able
to
read
through
it
yet
there's
some
good
comments.
That's
not
true!
I've
read
through
it
at
a
high
level
and
there's
some
comments.
One
of
the
things
that
really
struck
me
was,
I
think
it's
this
comment
similar
to
what
david
ashbold
was
saying
last
time.
This
came
up
to
the
meeting
is
how
do
you
track
input
and
output
of
a
particular
process?
It's
not
necessarily
like
an
easy
thing.
I'm
also
not
an
expert
on
it.
B
So
I
wanted
to
read
a
little
bit
more
into
what's
going
on
here,
but
I
do
want
to
encourage
anyone
who
is
interested
in
this
project
and
having
auto
instrumentation
for
goat
to
go.
Take
a
look
at
this
and
dive
into
some
of
this.
It
needs
communication
and
I
think
it's
a
it's
a
benefit
to
the
community,
but
I
also
want
to
make
sure
that
it
gets
it
right.
The
last
thing
we
want
is
to
have
a
buggy,
auto
instrumentation
to
frustrate
the
users,
but
I
haven't
validated
that
it's
possible
yet.
B
A
Yes,
so
a
few
months
ago
we
were
running
performance
tests
on
github
actions,
runners
and
we
removed
fat
because
noisy
neighbors
meant
that
pretty
often
they
were
fading
without
any
good
reason.
A
And
so
we
had
fading
tests
that
were
false,
alarms,
tyler
helmuts
and
has
been
asking
the
cncf
to
give
me
access
to
an
equinix
metal
account
that
the
php
sig
is
already
using
for
where
we
basically
can
set
up
virtual
machines
that
could
be
used
as
either
for
something
kind
of
low
level
or
we
could
use
it
just
to
set
up
gita
custom
runners.
A
I
know
I've
been
talking
with
the
person
who
I
don't
remember
his
name,
but
I
could
find
out
from
the
php
sig
who
created
the
machines
they're
using
it,
looks
like
they
want
to
set
up
a
full-blown
kubernetes
cluster
for
something
I
don't
think
we
need
that
much,
and
so
I
guess
my
main
question
is:
do
we
want
to
set
up
those
machines
manually
or
do
we
want
to
set
up
a
terraform
repository?
And
if
we
do,
how
do
we
automate
running
the
terraform
scripts?
A
And
are
we
happy
with
just
having
custom
runners
or
do
we
want
something
different?
And
finally,
because
there
was
the
concern
of
running
custom
runner
from
forks?
So
I
guess
for
now
at
least
we
could
run
the
performance
test
only
on
merges
on
main,
so
that
forks
are
not
involved.
B
Yeah,
that's
a
that's
exciting
to
hear
that
we
got
some
access
to
a
bare
metal
machine.
I
know
that
it
was
in
the
wings.
A
And
I
think
I
could
get
access
for
more
people.
If
we
need,
I
mean
I
got
access
because
I
was
involved,
but
I
guess
all
maintainers
could
could
get
access.
B
I
think
that
might
be
important
in
the
future
if
we,
if
we
have
something
running,
but
at
this
point
in
time
I
don't,
I
don't
necessarily
see
it
as
I
think
the
people
working
on
the
project
should
probably
be
just
limited
people
having
access
so
yeah.
I
think
that
kind
of
comes
to
the
next
steps.
It's
like
what
was
the
workflow
going
to
look
like.
I
I'm
pretty
open
to
whatever
suggestions
you
have.
B
I
don't
know
if
tyler
helmuth
is
still
working
on
this
or
if
anybody
else
is
going
to
be
working
on
it,
but
I
you
terraform
scripts.
Are
that's
a
good
suggestion?
That's
you
know
not
my
favorite,
but
I
you
can't
complain.
We
don't
have
a
suggestion
to
replace
it
with
so
yeah.
I
mean
I've
definitely
worked
with
them
before
and
they're.
Not
they
work.
I
think
it
a
lot
of
they
work
on
like
many
different
platforms.
B
So
I
think
that's
a
really
good
idea
in
case
we
do
move
in
the
future.
B
Yeah,
I
think
that's
a
good
idea,
and
I
would
what
I
would
do
is.
I
would
probably
also
build
them
with
the
understanding
that
it's
going
to
be
a
fork
that
you're
going
to
run
against
and
maybe
even
just
have
it
as
your
fork
or
something
like
that.
So
it's
a
configurable
variable
or
something
like
that.
So
in
the
future,
if
we
wanted
to
have
like
an
official
fork
or
we
wanted
to
push
it
back
to
the
original
main
repository,
it
could
be
done
that
way,
but
yeah.
That's
how
I
would
build
it.
B
At
least
just
so.
You
don't
have
to
wait
about
anything
on
the
main,
and
so
you
can
get
it
into
a
state
that
works,
and
then
we
can
integrate
it
back
into
the
main
for
the
terraform
scripts
repository.
B
I
would
you're
gonna
need
to
create
a
community
issue
which
I
don't
know
if
you
okay,
so
I've
seen
over
issues
about
that
yeah,
yeah,
yeah,
okay,.
B
A
I
maybe
I
can
find
it,
I
don't
remember
from
head,
but
I
can
find
it.
B
Okay,
it'd
be
ideal
if
we
could
update
the
findings
and
what
our
plan
is
in
that
issue.
If
you
can't
find
it
ping
me
and
I'll
try.
A
To
find
it
as
well-
and
so
last
question
is:
do
we
are
we
okay
with
running
vtr
from
scripts
locally,
or
do
we
want
to
do
something
like
set
up,
terraform
cloud
or
or
something
like
that.
C
C
C
B
I
also
just
want
to
give
you
a
heads
up,
anticipate
tigran,
being
really
interested
in
this
and
probably
wanting
to
collaborate
with
you
on
it
sure
yeah.
I
think
that
they're
doing
a
lot
in
the
collector's
space
that
is
just
anthony
would
know
more
than
I
do,
but
it's
essentially
a
one-off
performance
and
benchmark
tests,
but
yeah.
C
Well,
I
think
that's
what's
driving
tyler
helmut's
involvement
as
well
he's
associating
this
for
the
on
behalf
of
the
collector,
because
there
are
load
tests
that
are
run
as
part
of
the
collector's
ci
but
they're,
frequently
flaky
because
of
the
variability
in
github
actions.
Runners.
A
Yeah,
actually
it
just
like,
I
just
thought
about
it,
but
maybe
that
will
end
up
being
a
an
orgs,
an
org
white
thing,
because
I
think
runners
are
at
the
organization
level
and
not
repository
level.
B
Yeah,
I
think
that's
the
case
yeah.
I
think
you're
right.
B
I
think
that
that's
something
to
keep
in
mind
when
you're
building
the
terraform
scripts-
I
guess,
is
all
I'll,
say:
yep
yeah,
okay,
cool
yeah,
please
just
keep
us
informed.
This
is
great.
I
really
appreciate
it
all
the
work
going
in
there
and
let
us
know
if
we
can
help
in
unlocking
anything
for
you.
B
Yeah:
okay,
next
thing
on
the
agenda:
anthony
yeah.
Another
thing
I
wanted
to
add:
there's
some
dependable
scripts
that
need
some
reviews.
I'm
guessing.
C
Yeah,
I
think
I've
got
a
couple
reviews
on
them
now
I
just
wanted
to
raise
attention
to
them.
I
have
rather
shamelessly
stolen
these
from
the
collector,
where
alex
bowden
created
them.
Initially,
then,
I
made
some
improvements
by
one
of
the
things
I
was
noticing
when
I
tried
to
run
it
in
contrib
was
that
it
was
trying
to
update,
testify
end
times,
because
there
were
independent
years
merging
it
for
for
each
module.
C
So
I
do
a
two-pass
process
here,
where
I
build
up
a
map
of
module
versions
that
need
to
be
updated
and
then
use
that
so
it
dedupes
saves
times
running
it.
But
if
it's
run
as
a
github
action,
it
probably
doesn't
matter
too
terribly
much.
Rather
you
kick
it
off
and
walk
away,
but
the
idea
here
is
that
we
can,
after
a
depend
about,
does
its
weekly.
C
You
know
party,
where
it
creates
all
the
all
the
prs.
Ever
we
go
push
a
button
in
github
actions
and
this
bundles
them
all
up.
Does
all
of
the
go
mod
updates
itself
and
then
creates
a
new
pr
that
will
include
all
of
those
changes
and
once
that's
merged
dependable,
goes
on
and
closes
out
all
xprs.
B
Yeah,
I
think
this
is.
I
approve
both
of
them,
so
I
I
definitely
am
in
the
fan
table
of
this.
It
was
one
question
I
had
for
you.
Is
this
github
token?
Do
we
actually
use
it
anywhere.
C
It's
automatically
provided
by
github
actions.
It's
it's
used
to
push
to
the
origin
and
to
use
the
github
cli.
Okay,.
B
B
So
this
this
means
that
all
approvers
should
have
access
to
kick
this
off.
B
Okay
yeah,
this
sounds
good
to
me.
Love
me
some
some
solid
z,
shell.
Actually,
I
don't
think
I
think,
there's
some
things
that
bash
even
support
in
here.
B
C
B
B
Yeah,
I
agree,
I
think
there
it's.
You
know,
50
lines,
something
like
that.
Well,
cool!
That's
it
for
the
agenda
that
we
have
listed.
Does
anybody
else
have
something
they
were
going
to
talk
about.
B
Okay,
has
there
been
any
use
cases?
I
think
that's
another
thing.
I'd
love
to
hear
about
if
anybody's
been
using
hotel
in
the
wild.
B
I
was
excited
to
see
that
somebody
was
requesting
a
release
for
the
auto
prop
stuff,
which
shows
that
we're
actually
providing
value.
So
I
do
think
people
are
using
it
and
they
still
are,
but.
C
People
are
indeed
using
it
and
I
will
say,
I'm
getting
more
and
more
inquiries
from
internal
teams
at
aws,
not
necessarily
stuff,
that's
going
to
be
public
facing,
but
we
are
we're
using
it
more
and
more
internally
here,
even
from
teams
that
are
not
in
the
open
source,
observability
space.
B
Yeah,
that's
that's
a
really
good
sign.
I
think
over
at
splunk
we're
on
the
precipice
enjoying
very
similar
things
so
yeah.
I
think
I
think
once
you
see
a
lot
of
that,
it's
a
lot
easier
to
talk
to
customers
as
well
as
just
have
the
culture
around
using
open,
telemetry.
B
Well,
cool
yeah!
I
just
I
think,
that's
it
then
I
want
to
be
respectful
of
most
time
so
appreciate
everyone
for
showing
up.
If
you
have
anything
else
available
via
slack,
I
hope
see
you
next
week
anthony
will,
but
I
will
see
you
the
next
or
the
week
after
that.
Until
then
thanks
everyone.