►
From YouTube: 2021-04-15 meeting
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
A
Okay,
that's
because
yesterday
not
yesterday,
actually
the
last
week
we
are
going
to
release
a
dot
lambda
layer,
so
we
have
some
had
some
internal
discussion
and
we
found
the
the
current
upstream
repo
I
mean.
The
open-direction
number
still
is
hard
to
publish
london
here
due
to
some
reason,
for
example,
what
kind
of
third
party
exporter
can
we
bring
and
another
thing
is
same:
safe
aw
account
is
not
ready
yet,
and
you
will
bring
this
third
party
as
water.
We
still
have
no
ci
to
cover
them.
A
We
cannot
guarantee
their
quality
for
this
question.
Adult
london
already
has
answer,
but
we
have
no
answer
in
hotel
room,
so
I
want
to
post
the
document
in
channel,
so
please
go
through
it
in
10
minutes,
and
then
we
can
have
discussion
is
that
okay.
B
Yeah,
we
can
certainly
do
that,
but
I
alexa
thanks
for
joining
in
today,
we've
been
having
back
and
forth.
You
know
also
within
our
teams,
to
figure
out.
You
know
what
needs
to
be
in
the
open
telemetry.
You
know
public
layer,
lambda
version
that
we
want
to
add
versus.
B
You
know
what
becomes
customized
to
a
dot
and
is
available,
for
you
know
aws
specific
endpoints
as
as
a
second,
you
know
as
a
second
layer
right,
but
our
thinking
was
overall
that
could
we
actually,
you
know
and
first
of
all,
what
should
we
add?
You
know
in
the
public
layer,
you
know
what
makes
sense
as
a
basic
configuration
and
then
what
you
know
value
can
we
add
from
a
configuration
perspective
that
can
be
turned
on
for
other
third
parties.
B
B
It
could
be,
you
know,
made
as
easy
to
use,
as
you
know
anybody
to
be
able
to
instrument
with
so
that
said
again,
you
know
we're
trying
to
figure
out.
First
of
all,
what
goes
into
the
hotel,
lambda
extension
or
layer
and
and
what
you
know.
We
have
a
good
idea
of
what
goes
into
the
aws
a
dot
version
but
like
to
clearly
define
you
know.
B
What
are
the
benefits
we
want
to
ensure
we
support
with
it,
and
I
think
that
lay
wrote
down
some
of
these
things,
but
it's
not
you
know
completely
planned
out.
So
would
like
to
kind
of
get
your
feedback,
and
you
know
other
vendors
also
will
reach
out
to
and
kind
of
see
if
we
can
get
more
more
requirements
on.
If
you
will.
B
The
second
part
was
that
I
also
wanted
to
understand.
You
know
I
mean
I,
I
am
biased
towards
the
open,
telemetry
layer
being
available
first,
the
extension
being
available
first
before
the
aws
distro,
you
know
version
being
available,
so
I
mean
do
you
have
any
thoughts
on
that?
What
do
you
think?
Because
I
do
want
to
stick
to
the
idea
of
you
know
us
and
our
teams
contributing
to
the
project.
First.
C
Yeah
I
mean
I
guess
I
guess
that
would
be
the
ideal
scenario
if,
if
we're
not
there
yet
currently,
because
we
haven't
had
the
cycles
to
get
the
ci
in
a
place
where
we
can
release
from
the
open,
telemetry
repo,
I
feel
like
we
can
kind
of
hash
this
out
to
just
the
sig
kind
of
bootstrapping.
Still,
if
you
wanted
to
still
release
the
aws
repo
first
or
whatever
like
I,
I
don't
think
it.
C
B
Yeah
I
mean
again,
I
think
that
we
can
easily
add
the
ci
cd
and
we
do
have
a
plan
for
that.
I
mean
layers
outlined
some
of
the
options
which
we
want
to
kind
of
walk
through
today,
but
again
the
thinking
also
was
that
you
know
we'd
like
to
make
sure
that
the
the
open,
telemetry
version
is
available.
B
You
know
first,
and
that
means
that
you
know
the
ci
cd
dependency
dependencies
are
also
and
the
testing
dependencies
are
also
built
built
out
and
they
say
and
then
looking
at
the
aws
distro
layer
so
as
a
best
practice.
That's
what
we'd
like
to
see
now
the
question
is,
as
you
said,
you
know
what's
what's
possible
on
the
project
with
the
given
set
of
maintainers.
Do
we
have
an
action
plan?
You
know
we
can
certainly
help
build
out
the
ci
cd,
but
let's,
let's
at
least
have
an
agreement
on.
C
B
Yeah
yeah,
I
mean
at
least
I
would
like
to
see
the
issues
being
clearly
available,
the
design
proposal
and
what
we
are
building.
You
know
and
what's
work
in
flight
being
itemized
as
issues
and
and
just
you
know,
kind
of
being
very
transparent
about
what
we're
building.
B
So
I
mean
that
said
again:
should
we
as
a
next
step
and
rock?
Did
you
have
some
thoughts
there
I
mean
again.
I
think
that
we've
gone
we've
discussed,
you
know
what
should
be
in
the
public
layer
and
you
know
on
the
hotel
hotel
layer
which
is
which
I'd
like
to
see
is
vendor
agnostic,
but
at
the
same
time
you
know
what
do
we
provide
as
a
basic
configuration
to
make
it
usable.
D
I
mean
if
the
question
is
whether
we
can
release
a
dot
before
upstream.
I
have
the
same
bias
of
releasing
upstream
first,
but
I
don't
want
to
release
something:
that's
not
good
like
at
least
some
baseline
quality
with
integration
tests
and
stuff,
and
that's
going
to
take
a
long
time.
So
I
don't
know.
B
You
know,
because
summer
is
rolling
in
so
we
will
have
more
interns
on
the
project
not
only
from
aws
but
from
other
teams
also-
and
there
are
also
other
vendors
who
are
interested
in
you
know,
extending
the
lambda
lambda
support
on
open
telemetry,
so
we
could
get
them
involved,
but
we
do
need
to
have
a
clear
blueprint
available
before
that
right.
E
B
E
B
E
B
E
E
I
mean
what
I'm
trying
to
say
is
like
let's
keep
old
like
all
the
work
should
be
upstream.
If
you
ask
my
opinion,
if
we
need
like
yeah
cicd,
if
you
can't
set
up
ci
cd
on
upstream,
is
that
the
case
we
can
take
the
code,
you
know
somewhere
else
and
do
the
testing
there
every
periodically
like
every
hour
or
something.
B
B
Also
at
the
same
time,
right,
that
is,
the
artifacts
which
are
getting
published
is
the
lambda
extension
itself
and
because
that
also
has
another
dependency
on
the
c,
using
the
your
cncf
aws
account,
which
is
tied,
which
is
also
leveraged
by
open
telemetry
right
because
again,
you
don't
want
to
bind
it
to
aws
credentials
which
are
available
only
to
aws
and
that's
that's
the
other
dependency
which
you
know
we're
kind
of
trying
to
figure
out,
and
it's
that
it's
a
timing
issue.
It's
not
a
god.
It's
not.
E
B
D
B
Yeah
I
agreed
that
right.
D
D
B
I
mean
the
the
question
again:
let
me
go
back
to
this,
so
what
do
we
need
to
do
in
order
to
actually
pull
instrument?
The
ci
cd
upstream,
I
mean
I
want
to
have
a
clear
understanding
of
the
work
involved
and
and
lay
perhaps
you
know
once
you
have
your
google
doc
available.
We
can
look
at
it.
B
B
A
Right
now,
because
collector
and
the
president
called
our
system,
so
we
can
run
the
cd
test
framework
on
this
tool
implementation-
and
we
do
this
in
csv
downstream
by
column
the
source
code
from
open
financial
number
right.
So
we
can
run
our
hourly
task
to
continue
our
endpoint
test,
but
the
bacon
is
a
aws
pattern,
like
called
watch
on
the
x3.
A
B
A
B
No,
no,
so
why
don't
we
consider
that
if
we
added
a
light
light
step
back
end,
for
example
right
in
this
case,
because
alex
is
familiar
with
it,
so
we
can
actually
add
that
implementation
in.
Similarly,
if
we
provided
and
default
configuration
say
for
x-ray,
just
for
an
example
which
can
then
be,
you
know,
omitted
for
anybody
else
to
use.
B
A
Yes,
that's
okay
for
users
perspective,
that's!
Okay,
and
after
that
we
need
to
configure
the
ssd
test
test
case
yeah
and
the
trigger
the
laptop
sends
the
traffic
button
and
then
pulls
data
from
backhand
and
then
verify
this
result.
A
So
that
takes
a
long
time
for
aws,
because
we
already
have
a
mature
test
framework
to
verify
the
result
from
x.
So
we
can
easily.
A
B
C
B
C
Yeah
I
mean,
I
think,
that
that
would
be
doable.
That
being
said,
I
I
don't
know
how
much
you
know
complete
testing
all
the
way
for
all
the
back
end
that
we
want
to
be
able
to
support
in
this
collector
layer
is
is
needed
for
the
upstream
repo.
You
know
one
of
the
actually
two
different
users
came
and
asked.
You
know,
for
example,
for
the
data
dog
exporter
to
be
in
this
collector
layer.
C
And
I
mean
I,
I
don't
know
if
we
want
to
have
an
easy
way
to
just
get
other
vendors
to
be
able
to
add
their
exporters
to
this
layer
or
even
just
include
all
of
the
contrib
exporters.
I
think
we
had
talked
about
that
last
last
week.
At
this
I.
B
F
Isn't
it
also
like
the
main
reason
these
were
all
pulled
out
to
begin
with
was
to
reduce
the
image
size
because
that's
like
a
critical
issue,
so
it
seems
like
long
term,
adding
everything
back
in
to
one
single
image,
probably
wouldn't
work
or
at
any
rate
like
I
mean
it's
more
work
to
build
this
ci
cd
pipeline,
but
it
seems
like
what
people
would
want
is
if
they're,
using
the
data
dog
one
they're,
just
taking
like
a
base,
you
know
like
whatever
our
baseline
upstream
lambda
collector,
is
and
then
just
adding
the
exact
exporter
that
they
want.
E
Into
their
existing
extensions,
rather
than
open,
telemetry
is
providing
a
lambda
extension
that
works
for
everyone,
because
the
size
restriction
and
all
that
will
be
a
difficulty,
and
then
we
will
never
be
able
to
like
you
know,
get
out
of
this.
Like
you
know,
we
can't
make
any
compromises,
because
we
want
to
support
everyone
right.
F
As
long
as
we
provide
like
it's
straightforward
enough
to
to
build
your
own
version
of
this
thing,
I
think
it
might
be
fair
to
say,
like
we
don't
need
to
be
in
the
business
of
providing
ci
cd
for
for
all
the
different
iterations
of
this,
like
open
telemetry
could
provide
just
the
baseline
otlp
one
plus
the
amount
of
x-ray
that
you
need
to
make
it
work,
because
you
do
need
some
amount
of
like
you
know,
x,
amazon,
headers
and
stuff,
and
then
you
know,
amazon
provides
whatever
versions
that
they
want
to
provide
downstream
and
other
other
people
can
build
their
own
pipeline,
especially
if
it's
just
straightforward
enough
to
kind
of
clone
the
open
telemetry
like
if
it's
just
like
you
can
take
this
pipeline,
move
it
somewhere
else,
hook
it
up
to
your
own
credit
card
and
then
you're
done.
B
B
You
know
one
integration
test
for
an
odlp,
backend
and
and
then
just
provide
that
as
a
baseline
image.
Would
that
be
something
that
works?
I
mean
that's
simpler
for
sure.
A
A
That
it
will
bring
data
down
and
bring
in
x3.
We
need
to
add
this
to
sensitive
test
casing
potential
that
will
inflate
the
size
of
the
code
and
probably
in
the
future,
it's
really
loss
of
loss,
our
control,
but
if
the
baseline
is
only
support,
otlp,
like
the
core
record.
A
Yeah,
that's
probably
a
simple
solution,
but
that
also
brings
another
thing.
The
second
question
do
we
need
to
do?
We
need
to
really
publish
a
public
layer
for
this
simple
version,
because
it's
always
support
otlp
backend
and
it
doesn't
contains
any
useful
server,
propagator
and
either
generator.
F
F
F
F
I
don't
want
this
to
be
like
a
stick
to
force
people
to
do
it,
which
is
why
I
think
it
should
be
possible
for
them
to
just
make
their
own,
but
but
I
think
there
will
be
enough
carrots
that
over
time
you
know
people
will
just
start
to
ingest
this
same
way.
They
ingest
zipkin
data,
or
you
know
anything
else.
B
So
I
mean
one
of
the
again
ted
just
to
just
for
context.
The
reason
you
know
we
are
we're
super
interested
in
identifying.
What
that
you
know
image
looks
like
is
because
we
want
to.
We
have
been
adding.
You
know,
support
in
python
as
well,
as
you
know,
want
to
roll
out
the
collector
support
that
has
been
built
for
lambda,
with
in
a
dot
right
and
and
we're
adding
javascript
and
javascript
implementation
too.
So
we
would
like
to
keep
rolling
out.
B
You
know
different
languages
being
supported
every
month
and
we
were
hoping
to
release.
You
know
lambda.
Another
lambda
version
this
later
this
month
in
10
days,
right
27th
is
the
release
for
aws
distro.
This
month
I
mean
we
do
a
monthly
release
and
you
know
again,
we
were
hoping
that
we
could
actually
release
aw
a
dot,
lambda
extension
right
on
lambda
layer.
So
hence
you
know,
our
interest
was
like
we
want
to
make.
B
Sure
upstream
is
actually
has
the
information,
the
the
basic
layer
available
first
before
we
release,
because
we
don't
want
to
release
it
before
the
upstream
has
that
functionality
also
right.
So
that
was
kind
of
the
thinking
that
could
we
define
a
generic?
You
know
public
layer
which
is
otlp
friendly,
and
you
know
supports
that
as
a
baseline.
F
Yeah
I
mean
it
certainly
seems
worthwhile
to
me.
I
know
it's
like
a
bummer
to
delay
work.
You
know
because
this
would
be
additional
front.
Loading
work
and
personally,
like
I
personally,
wouldn't
have
a
problem
if
it
was
just
like
an
order
of
operations,
thing
that
you
know
the
downstream
one
got
released
first,
this
one
time
like
that's
not
to
me,
that's
not
a
big
deal
as
long
as
you
know
we're
we're
committed
to
to
moving
over
to
this
pipeline.
Oh.
B
Absolutely
I
mean
the
point:
is
that
it's
just
in
the
short
run?
It
is
a
timing
thing
in
the
long
run.
Of
course
I
mean
we
want
to
see
everything
on
the
project.
That's
the
and,
and
we
have
engineering
lined
up
for
it.
It's
just
that.
You
know
we're
doing
both
in
parallel
right.
F
B
F
Yeah
yeah,
I
think
that's
great,
and
then
you
know
to
a
certain
degree.
Aws
is
then
just
kind
of
like
a
model
citizen
right
like
if
we're
doing
it
slightly
backwards,
but
if,
in
the
end,
it's
like
what
aws
is
running
is
basically
what's
what's
available
upstream
and
what's
available
for
someone
else
to
to
clone
to
make
their
own
pipeline,
so
they
can
include
their
own
plugins
and
publish
their
own
layers.
F
Then
then,
that's
like
perfect,
and
then
you
know
we're
keeping
the
image
size
down
and
we're
avoiding
this
like
expensive
concept
of
like
a
collector
repository
kind
of
thing.
You
know
to
me
that
that
sounds.
F
That
sounds
perfect,
especially
because
I
think
we're
going
to
have
end
users,
especially
big
companies,
who
are
going
to
want
like
actually
to
customize
that
thing
and
might
be
willing
to
run
their
own
ci
cd
in
order
to
customize
it
and
they
might
be
similar
to
aws
and
have
their
own
security
requirements
where
they'd
prefer
doing
that
kind
of
thing.
So
so
I
I
kind
of
like
that
approach.
B
Okay,
so
I
mean
again,
you
know
we
wanted
to
make
sure
that
it's
something
which
is
very
clear
and
transparent,
so
that
you
know
there's
a
clear
plan
on
the
project,
there's
a
whole
commitment
to
the
project
you
know
and
and
being
upstream
first,
we
have
a
clear
guideline
that
this
is
what
we're
building
in
parallel
to
what
we're
releasing.
F
It's
more
like
there
are
collector
plug-ins
that
are
like,
like
official
or
validated
in
some
way
you
know,
but
there's
no
need
to
running
like
a
kitchen
sink
collector
or
a
particular
distro
is
not
like
you.
You
run
the
distro.
That
makes
the
most
sense
for
you,
and
this
is
a
good
example
of
open
telemetry
starting
to
provide
tooling.
That
makes
that
that
reasonable
for
people
to
do-
and
this
will
also
help
yeah
prevent
this
issue
of
like
every
vendor
being
like.
B
Also,
you
know
establishing
a
good
practice
of
actually
making
sure
there's
a
clear
plan
of
you
know
what
we
are
building
on
the
project
and
and
continuing
to
work
on
that
and
having
a
clear
roadmap
on
the
project.
At
the
same
time
that
you
know
what?
B
Yes,
if
you
can
plan
it
out,
well
enough
to
have
it
on
the
project
and
and
then
kind
of
also
have
a
flavor
downstream.
F
Yeah
and
I
don't
know
how
totally
reusable
it
is,
but
just
in
general
I
think,
there's
a
concept
of
like
a
collector
build
pipeline
that
we
want
to
give
people,
and
I
don't
know
to
what
degree
there
is
already
something
like
that
in
like
with
the
collector
in
general.
But
you
know
this
is
one
where
it's
a
more
complicated
one,
because
end
results
like
lambda
layers.
F
Maybe
the
collector's
simple
enough
for
people
to
build
that
doesn't
really
need
a
pipeline
or
or
ci
cd
tools.
But
I
mean.
B
F
Yeah,
maybe
it's
just
a
blog
post
at
some
point,
someone
can
write,
which
is
like
hey
here's,
how
you
build
your
own,
open,
telemetry,
collector
people
offer
them
there's
good
ones.
We
recommend
you
use
those,
but
you
might
have
your
own
plugins.
You
have
your
own
reason
to
build
one.
So
here's
here's
how
you
do
it
it's!
It's
really
easy.
Yeah.
B
B
You
know
if
you
want
to
build
your
own,
as
well
as
an
sample
public
layer
available
with
otlp
enablement,
which
is
available
on
the
project
and,
and
you
know,
additional
blog
posts
and
anything
else
that
we
need
to
communicate
like
a
backlog.
I.
F
I
think
this
sounds
great,
and
this
I
mean
I
feel
like
this
resolves
a
lot
of
issues
both
for
like
how
this
was
gonna
work.
So
I
was
starting
to
watch
that
brew
and
then,
in
general
it
allows
us
to
start
publishing
more
things
around
the
narrative
of
like
collector
distros
and
things
like
that,
because
that's
just
an
area
where
I
just
watching
the
public.
F
You
know
getting
all
paranoid
and
around
distros
yeah.
F
B
Yep
exactly
so
I
mean:
does
that,
provide
you
enough
clarity
on
what
what
the
scope
of
what
we
are
building
for
the
public
layer,
hotel
public
layer
clearance.
A
Oh
so
yeah
for
absolutely
okay,
supporter,
limited
exporter,
yeah.
Let's
help
us
build
the
city
verify
the
otl
button
yeah
and
our
downstream
will
keep
our
original
plan.
We
will
yeah.
A
B
B
Okay,
so
we
will
use
use
that
as
our
our
our
home
repo.
A
No,
for
I
just
want
to
mention
that
if
you
want
to
publish
layer
to
cover
every
commercial
region-
probably
it's
a
big,
it's
a
big
work,
because
there
are
more
and
more
new
religion
region
build
so
right
now
we
have
a
contestant
region,
yeah,
so
be
careful.
Yeah.
B
You
know
there
isn't,
we
can
add
documentation
for
that
and
then,
if
you
know,
let's
go
based
on
what
people
need
right.
So
if
there
is
a
request
for
it,
then
we
can.
A
A
For
our
side,
we
have
to
deploy
the
regions
by
ways,
so
we
cannot
deploy
everything
at
one
time.
That's
a
that's,
also
a
security
concern,
but
for
lifestep,
when
you
deploy
the
collector,
so
you
can
make
the
decision
yeah.
B
F
Yeah,
it
seems
like
a
practical
limitation,
I
mean
so
this
basically
means
end.
Users
would
get
the
layer
and
would
have
to
upload
it
themselves
following
some
instructions.
Yes,.
A
F
F
F
B
G
D
E
Yeah,
that's:
is
there
a
way
to
programmatically
like
query
regions,
I'm
new
to
aws
all
right.
A
Let
me
show
a
sample
what
what
the
release
notes
would
be.
The
book
like.
F
B
For
you
know
covering
all
regions
everywhere,
you.
A
B
A
F
I
see
so
it's
like
a
security
issue.
B
No,
no,
it
depends
on
the
security
guideline.
Any
customer
has
right.
That
is
what
what
is
their
preference?
Are
they
actually
looking
at
multiple
regions
with
one
account
or
or
if
they
are
divvying
it
up
there?
They
have
a
single
account
per
region,
yeah.
A
A
A
B
A
C
C
B
For
being
clear
about
like
the
link
that
alex
just
shared,
so
we
should
just
refer
to
those
links.
If
somebody
has
a
special
request
again,
let
let's
let
them
make
it
on
the
repo.
You
know,
and
we
are
all
there
to
help.
B
Yeah
yeah
because
I
mean
there's
the
public
posted
documentation,
so
we
should
refer
link
that
out,
I
mean
again,
the
key
is
having
good
documentation
because
that's
what
confuses
most
people
on
lab
for
sure,
yeah,
all
right,
cool
nice.
So
thanks
thanks
lee
for
bringing
this
up
there
and
you're
in
sync
with
that
right
with
the
implementation
or
the
deployment
details,
all
right,
cool
all
right,
oh
yana!
E
Oh
no,
I
think
you
answered
my
question
I
was
about.
I
came
here
to
ask
like
hey:
should
we
think
a
little
bit
about
the
prometheus?
You
know
I
opened
that
conversation
previously
in
the
previous
meeting.
How
are
we
going
to?
You
know
export
prometheus
metrics,
but
given
the
default
is
going
to
be
using
otlp.
A
E
I
think
that
question
just
it
doesn't
exist
anymore
and
for
the
distro,
we'll
figure
out.
My
only
question
is
like:
how
are
we
doing
health
checks?
Do
we
do
any
health
checks
or
anything,
because
it's
also
pool
based-
and
I
was
wondering
like
how
do
we
do
this
from
the
extensions.
E
E
B
A
Think
lambda
is,
should
have
mechanism
automatically
rebuild
the
longer
sample
if
it
found
an
essential
question
so.
D
E
The
health
check
would
be
available,
but
some
of
the
components
might
be
failing,
so
they
want
to
also
surface
those
type
of
like
you
know,
issues,
I'm
not
sure
we
want
to
go
that
deep,
but
you
know
even
what
what
is
health
is
kind
of
like
a
questionable
thing,
but
at
least
if
we
can
surface
the
the
the
existing
health
check,
you
know
I
started
all
the
components
and
you
know
collectors
running.
That
would
be
a
good
first
step.
B
Yeah,
but
it's
a
very
good
point:
you
bring
up,
because
I
think
that
we
need
to
have
that
introspection.
Otherwise
and
at
least
an
understanding
of
what
you
know.
Users
can
use
good
how
long.
B
Open,
yes,
awesome
cool
thanks
so
make
that
thanks,
okay,
good,
at
least
those
were
my
topics.
That's
all
I
had.
H
D
D
A
So
for
that
our
developer
is
almost
finished,
the
lambda
instrumentation
and
he
is
going
to
submit
pr
to
country,
donate
a
country
vehicle
yeah.
As
we
said,
he
will
work
on
the
sample
application.
The
sample
education
will
yeah
use,
collector
layer.
B
C
A
C
Connections
for
tlp
via
environment
variables
that
will
be
fixed
in
the
next
in
the
next
release.
B
C
A
B
Yeah
we
wanted
to,
we
wanted
to.
You,
know,
release
the
next
update
with
with
the
latest
python.
C
I
don't
have
any
comments
on
that
one.
Yet.
B
A
A
D
F
F
So
I
think
it
is
going
to
be
a
little
bit.
There
is
a
proposal
now
tegan
just
made
one.
Let
me
see
if
I
can
get
it
for
you
real
quick.
He
just
published
this
yesterday.
F
Okay
yeah,
so
so
this
is
part
of
what's
needed
is
just
you
know,
some
definition
of
what
stable,
actually
the
means
we
produce,
because
we
can't
really
say
the
instrumentation
stable
until
we
figure
that
out.
The
other
thing
that
we
need
help
with
is
just
making
sure
that
our
our
conventions
are
correct.
F
Like
are
we
missing
stuff?
We
do
get
some
feedback
that
people
get
more
data
out
of
other
systems,
and
I
can't
currently
tell
without
you
know,
doing
work
to
what
degree
that's
our
conventions
aren't
sufficient,
or
is
it
just
that
the
instrumentation
we've
already
written
is
kind
of
crap
and
like
doesn't
actually
like
conform
to
the
conventions
or
like
include
all
the
optional
stuff,
so
for
them
to
get
stable?
I
think
we
have
to
first
research.
Our
conventions
make
sure
we're
happy
with
them,
and
jano
is
doing
some
good
work
around
just.
E
F
Yeah,
but
that's
like
you
know,
when
we're
standardizing
stuff,
that's
the
kind
of
work
we
gotta
do
and
so
just
like
getting
that
done
and
then
actually
doing
a
sweep
through
our
instrumentation
and
getting
it
up
to
par
and
then
the
third
part
of
that
is
configuration.
We
don't
actually
have
any
guidelines
about
configuration
for
instrumentation
and
my
experience
from
open
tracing
is
that
turns
into
like
a
hell
for
the
end
users,
because
what
happened
there
is
we
had
no
guidelines,
and
so
people
just
make
issues
and
pull
requests
with
individual
repo.
F
A
E
F
Just
what
even
what
options
people
should
provide
would
be
helpful
and
if
we
could
go
all
the
way
to
like
here's
like
a
yaml
file
for
open
telemetry
that
you
can
consume
to
like
configure
all
this
stuff.
That'd
be
great,
but
but
at
least
standardizing
what
should
be
con
baseline
configurable
and
how
that
should
be
presented
really
help
a
lot.
D
F
F
Telemetry
has
a
lot
of
stuff
in
parallel
right
now,
and
I
think
the
maintainers
have
a
real
are
really
afraid
of
the
instrumentation
and
contrib
because
they
see
their
bandwidth
is
taken
up
entirely
by
by
just
getting
all
the
baseline
stuff
over
the
finish
line
and
that's
actually
been
creating
bad
experiences
for
people
who
like
try
to
contribute
stuff
over
there
or
get
things
fixed
just
because
the
response
time
is
really
poor
and-
and
we
don't
have
a
great
answer
to
like
how
we
actually
want
to
manage
that
ecosystem.
F
We've
got
some
ideas,
nikita's
working
on
some
proposals,
so
there's
a
hotel,
instrumentation
channel
where
we're
discussing
this
stuff.
But
you
know
it
is
that's
long
term
instrumentation
is
like
the
biggest.
The
biggest
bug
bear
in
the
whole
thing
like
this
is
actually
why
open
tracing
started
to
begin
with.
Was
everyone
not
wanting
to
maintain
all
their
instrumentation?
And
then
now
we've
made
this
organization
we're
like
wow.
We
don't
want
to
maintain
it
ourselves
either.
As
an
organization,
so
so
that's
just
long
term
going
to
be
really
hard,
it's
the
hardest
part
actually
yeah.
F
Yeah
so
again,
yeah
configuring.
That
stuff
is
going
to
be
interesting
right,
especially
in
an
environment
like
lambda,
where
you're
trying
to
bootstrap
all
of
this
stuff.
For
them.
That's
where
something
like
a
configuration
file
or
configuration
language
starts
to
become
helpful
right,
because
even
accessing
that
stuff
to
configure
it
through
code
might
might
be
difficult
in
some
of
these
environments.