►
From YouTube: 2022-09-19 meeting
Description
Open Telemetry Meeting 1's Personal Meeting Room
A
A
A
A
A
A
A
B
B
Josh
you
mentioned
that
you
do,
you
know
php
right,
so
we
got
a
new
service
merged
called
quote
service
and
it's
a
super,
simple
app
that
just
gets
a
request
and
based
on
the
on
the
json
that
it
gets
like
with
the
count
of
items
it
returns.
A
shipping
quote:
shipping
quote
costs
last
week
we
got
it
merged
and
it
was
working
fine.
It
is
still
working
fine,
but
now,
whenever
we
start
up,
we
are
getting
like
a
bunch
of
weird
messages
regarding
something
that
was
deprecated.
B
D
I
mean
honestly
as
long
as
it
didn't
I
mean
as
long
as
it
doesn't
make
it
into
a
release.
Then
it's
you
know
a
big.
C
C
I
think
something's,
going
on
with,
like
the
load
generator
posts
like
the
0.4
release,
wanted
to
spend
some
time
on
the
hero
scenario,
kind
of
planning
any
build,
optimization
stuff
we
can
do
and
then
maybe
also,
if
we
have
some
time
discuss,
you
know
how
we
would
potentially
want
to
have
like
a
live
version
posted
somewhere
and
what
that
may
take,
or
what
form
that
may
take
any
other
topic.
C
It's
not
margin
issues,
let's
see
php
service.
C
Let
me
take
that,
I
don't
know
okay,
so
we
had
the
load
generator
issues
already
taken
care
of
austin.
Did
you
say
that.
E
Sorry
carter,
just
one
quick
thing:
I
was
noticing
that
in
the
in
a
kubernetes
cluster,
for
some
reason
after
I
deploy
with
helm
charts,
the
cart
does
not
seem
to
be
working.
I
checked
it
about.
C
Yeah,
I'm
not
sure
if
that's
a
known
issue,
I
think
ping
han
mentioned
yesterday.
He
was
having
no
issues
with
his
kubernetes
cluster,
so.
A
C
Okay
cool,
so
I
think
the
load
generator
issues
were
already
addressed
before
I
joined
the
call
juliana.
Do
you
want
to
give
a
quick
update
on
the
php
service?
I
think
he
posted.
B
Something
so
I
beat
gary
this
week
and
we
well.
Actually
he
worked.
I
just
helped
a
bit
on
the
on
the
rest
service
to
get
some
load
into
the
into
the
quote
service
and
he
pushed
yesterday.
I
think,
or
three
days
ago,
yeah
I
tested
today
and
everything
looks
fine,
so
I
just
approved
the
pr.
We
just
need
to
get
it
merged
and
we
have
php.
I
can
which
one.
D
C
D
All
right,
that's
ready,
so
I'm
gonna
merge
it.
C
Okay,
awesome,
though
that
was
really
good
work,
though
giuliano
and
really
kind
of
key
deliverable
that
helped
us
meet
all
of
our
or
most
of
our
requirements.
That's
a
that
was
a
good
one.
Okay,
I
think
austin
you
mentioned
306.
306,.
D
C
D
C
C
D
Can
reassess
this
after
1.0,
but
at
this
point
like
we're
it's
the
19th
of
september
and
we
need
to,
I
feel
like
we're
in
the
cut
scope
phase.
E
Yeah,
I
remember
the
I
created
305
an
issue.
I
don't
know
it's
one
month
ago
when
and
I
think
without
this
pr,
we
we
have
the
kind
of
the
problem
that
the
tests
are
not
working
and
the
user
experience
is
bad
and
if,
as
a
user,
you
try
to
run
those,
they
don't
run
so
kind
of
merging
this
one.
I
think,
wouldn't
make
it
worse.
E
B
B
D
A
C
For
the
heroes
scenario,
we
can
maybe
wait
until
pierre.
I
think
it
said
to
be
a
little
bit
late,
so
I
feel
like
we
can
wait
on
him
a
little
bit.
C
So,
let's
talk
about
build
optimizations
and
really
anything
we
can
do
to
kind
of
improve
the
performance
of
the
applications.
This
is
kind
of
just
a
generic,
almost
like
a
brainstorming
item.
I'd
love
to
hear
kind
of
from
the
group,
if
you
all
have
any
like
specific
ideas
or
also
kind
of
like
areas
of
investigation.
I
know
the
currency
service
is
probably
like
our
number
one
enemy
right
now,
but
curious
for
other
kind
of
input.
Here.
D
I
just
ran
off
of
maine
docker
composed
build.
No
cash
took
13
minutes
on
a
m1
max
or
whatever.
So
that's.
D
B
But
hasn't
pr
changed
that
in
his
last.
C
C
The
doctor
from
catch
or
dr
king
pose
catch.
That
was
pretty
helpful.
I
think
it's
like
a
five
minute.
You
know
non-build
for
users
now,
which
is
pretty
cool.
B
Well,
I
I
to
be
honest,
I
don't
see,
I
don't
see
a
way
of
doing
that
with
the
c
plus
plus
for
now
and
php,
because.
A
B
A
B
Building
everything
from
right:
it
has
the
build
alt,
jrpc
or
whatever,
and
you
fix
like
ages,
yeah
and
php.
We
are
downloading
some
protobuf
and
some
grpc
stuff
as
well.
Even
though
the
the
app
is
just
a
http
app,
but
I
think
we
use
jrpc
for
the
exporter.
B
Yeah,
so
it's
kind
of
it
takes
a
while
as
well
but
yeah.
If
we
compare
php
and
currency
service
doesn't
like
yeah,
php
is
just.
D
D
Because
the
only
other
option
I
could
think
of
would
be
to
build
intermediate
images
and
publish
those
intermediate
images
so
that
the
only
thing
that
needed
to
be
ch
whenever
you
rebuilt,
the
only
thing
that
would
actually
need
to
be
changed
would
be
the
like.
Actual
code.
A
D
From
a
prioritization
standpoint,
actually
I
guess
just
to
put
my
thumb
on
it
a
little
bit
from
a
prioritization
standpoint.
This
feels
like
something
that
can
wait
because,
yes,
it's
a
the
30
minute
build
is
very
annoying,
but
it's
much
it's
only
annoying
if
you're
developing
and
if
you're
using
it,
then
it's
pretty
fast
or
it's
at
least
acceptably
fast.
I
think.
D
D
If
you're
developing
more
than
one
app
yeah
yeah,
if
you're
just
doing
like,
we
probably
experience
pain
more
than
anyone
else,
because
we're
doing
full
builds
no
cash
constantly
due
to
changes,
so
I
would
be
willing
to
say
this
gets
pushed
post
one.
Oh.
D
In
my
mind,
the
like
fixing
bugs
buttoning
everything
up
and
getting
like
the
hero
experience
and
getting
the
telemetry
like
experience,
is
more
important
for
for
the
rest
of
september.
C
Yeah,
that's
fair
speaking
of
telemetry.
I
think
we
only
have
one
service.
I
guess
the
quote
service
giuliano
could
give
me
an
idea
of
the
tracing
completeness
of
that
I
was
going
to
say
only
the
future
flag
service
is
really
missing
any
components,
but
I
realized
I
haven't
thought
about
that.
A
B
And
we
extend,
we
extend
it
and
I'm
also
creating
manual
traces
manual,
expense
and
also
I've
added
events.
B
C
Yeah,
that's
fair,
okay,
cool
and
then
josh.
I
think
you
have
a
pr
open
for
the
feature
flag
service
right
now.
Do
you
have
any.
A
C
Cool
well
giuliano
austin
miko.
If
y'all
got
to
take
a
look
at
it-
and
let
me
know
is
that
yeah
I
was
just
looking
at
it.
It's
a
feature,
flag,
custom
attributes
and
spams
for
future
flag
services.
C
Yeah,
perfect!
Okay,
I
don't
think
pierre
is
still
here,
so
we
can
talk
a
bit
about
the
feature
flag
scenarios
anyways
or
the
hero
scenarios.
Excuse
me
so
essentially
for
hero
scenario.
It's
just
kind
of
a
guided
investigation.
I
think
some
other
vendor
experiences
for
their
demos.
Do
a
pretty
good
job
of
it.
I'm
not
sure
if
y'all
seen
examples
like
that
around,
but
so
it'd
be
great
to
have
at
least
one
kind
of
guided
hero
scenario.
C
Experience
for
our
v1
release,
I
think,
and
that
really
kind
of
drive
home
the
value
customers
are
deriving
from
it.
So
I
know
we
have
a
couple
of
feature
flag
scenarios
already
in
place,
but
I'd
be
curious,
kind
of
get
to
get
y'all's
perspective
on
like
how
comprehensive
you're
looking
for
this.
If
you
already
have
a
scenario,
you
know
kind
of
top
of
mind
or
really
just
you
know
any
input.
I
guess
so.
D
D
D
I
mean
the
general
thing,
though,
is
you
want
to
be
able
to
pivot
between
the
one
thing
that
would
be
nice
to
see,
and
I
don't
know
if
we
can
actually
easily
do
this
would
be
something
that
lets
us
lets.
You
combine
traces
and
metrics
right
so
that
if
you
know
something,
if
there
was
a
problem
that
showed
up
in,
you
know
like
resource
metrics
right
like
kubernetes
or
docker,
and
then
let
you
kind
of
pivot
into
a
trace
or
a
group
of
traces
to
see
that
in
the
application
level,
that
would
be
ideal.
D
I
don't
think
we
have
anything
quite
like
that
right
now.
The
only
thing
I
could
think
would
actually
fake.
That
would
be
something
I
guess
the
easiest
to
be
like
I
don't
know
disk
usage
or
something
or
just
like,
creating
a
creating
some
sort
of
like
really
tight
memory
allocation
thing
somewhere
to
cause
oom
kills.
C
I'm
trying
to
think
what
our
current
feature
flags
are
mainly
app-based.
I
think
maybe
we
slow
down
on
one
particular
transaction
or
product
and
then
maybe,
if
you
send
to
canada,
I
think
it
gets
canceled
right
now,
so
the
slowdown
seems
like
it
could
be
potentially
helpful,
but
it
is
tough
on
that,
like
resource
and
infrastructure
kind
of
correlation.
D
C
C
We're
essentially
trying
to
like
create
a
correlated
infrastructure
and
app
problem,
and
so
with
the
feature
flags.
We
have
really
good
app
failures,
but
we're
trying
to
think
of
how
we
can
connect
it
to
like
a
kubernetes
like
compute
or
cpu
or
memory
issue,
or
something
like
that
to
have
a
kind
of
more
comprehensive
scenario.
E
Yeah,
I
can
comment
that
I
I
think
we
have
one
which
is
working
and
and
it's
great
it's
it's
it's
a
start
and
yeah,
but
no,
no
like
a
clear
ideas,
but
I
think
we
could
have
like
many
of
those
and
some
can
be
simple
ones
but
but
yeah
I'm.
I
will
also
be
thinking
about
those
and
yeah.
Hopefully
we
can
come
up
with.
C
I'd
say
we're
pretty
poor,
but
we're
also,
we
don't
do
anything
with
metrics
today,
so
it's
kind
of
the
experience
sucks,
but
it
intentionally
sucks.
I
guess
okay,
yeah.
A
C
D
I'm
just
wondering
in
terms
of
prioritization
if
the.
D
E
D
B
I
actually
beat
someone
in
one
of
the
pr's
I
think
about.
Regarding
that,
like
hey,
could
we
get
some
default
dashboards
into
like
graphene
as
a
code
or
something
like
that?
So
whenever
we
spin
up,
we
have
the
dashboards
and
stuff
like
that,
because
at
the
moment
neither
prometheus
or
rafana
has
anything.
So
you
kind.
D
D
D
C
Telemetry
metrics
outside
of
instrumentation
libraries,
and
even
then,
I'm
not
even
sure
if
the
metric
provider
or
whatever
it's
been
configured
for
a
lot
of
these
I'd,
be
curious.
What
we're
getting
from
that,
but
yeah,
it
definitely
seems
like
working
to
be
done
there.
I
am
a
bit
stuck
on
that.
D
What
would
be
the
easiest
thing
to
okay?
I
I
got
myself
twisted
mixed
up.
I
think
the
I
think,
here's
here's,
I
think
the
way
to
do
this.
We
need
to
figure
out
the
narrat,
the
actual
hero
scenario.
Here
right.
We
need
to
figure
out
the
problem
we
want
to
simulate.
Then
we
need
to
figure
out
what
telemetry
we
need
to
implement.
D
C
D
That's
like
the
start
of
it.
That's.
D
C
It's
still
a
pretty
good
amount
of
time,
but
we
do
have
to
be
really
limited.
I
know
that
we
literally
have
four
weeks
to
finish
whatever
we
want
to
finish
so
and
that
in
that
sense,
build
optimizations,
probably
already
good,
just
kind
of
deeper
for
now
but
yeah
it
would
be
nice
to
get
pierre's
perspective.
So
I
know
you.
D
Let's
figure
that
out:
let's,
let's
make
that
like
top
of
mind
and
then
we
can
discuss
it
async
and
make
some
issues
for
that
like
today,.
C
Yeah,
let's
just
try
and
have
like
our
narrative
by
the
end
of
the
week,
and
I
think
if
we
have
like
at
least
a
good
sense
of
that,
then
we'll
we'll
be
in
a
better
place.
I
agree
and
then
do
we
want
to
understand
any
question.
B
On
that,
are
we
focusing
on
metrics
or
not
adding
metrics
to
to
the
services
that
we
have
think
we
need
to.
A
B
D
A
D
Let's,
let's,
let's
get
a
north
star
and
then
work
towards
that.
If
there's
opportunistic
things
we
can
do
like,
if
I
don't
know
python
updates
with
a
standard
lib
metrics
thing
right,
I
think
the
other
like
a
related
point
here
is:
do
we
previously
we've
said
we
would
prefer.
We
don't
want
to
use
like
prometheus
instrumentation.
We
want
to
use
pure
hotel,
but
we
could
change
that
and
we
could.
D
We
could
go
back
on
the
decision.
We
could
add
in
prometheus
instrumentation
to
a
lot
of
these,
and
that
would
give
us
like
metrics.
D
C
Yeah,
it's
it
doesn't
the
only
ones
that
support
it
or
java.net
c
plus
python
and
go,
I
think,
yeah
those
five,
maybe
javascript.
C
A
B
A
A
D
D
D
D
D
Okay,
does
someone
want
to
take
that
for,
like
this
week,
just
doing
a
review
and
like
like
you,
don't
have
to
start
to
implement
it
just
like
put
together
an
issue
that
has
like.
C
Yeah
we'll
just
track
it
in
the
repo,
but
we'll
focus
on
that
narrative
and
kind
of
deal
with
issues
as
that's
what
he
to
implement
something,
I
think
so
it
should
be
okay.
C
We
only
have
10
minutes
left,
but
we
can
briefly
discuss
live
hosting.
I
think
it'd
be
a
really
cool
feature
to
have,
and
it
would
be
interesting
to
see
if
we
could
integrate
some
sort
of
like
mobile
app
component
in
that
long
term.
So
it'd
be
great
to
have
you
know
the
complete
language
representation
and
also
just
having
a
live
component
and
maybe
having
some
cool
live
experiences
to
be
really
interesting
as
well.
But,
of
course,
with
that
comes
billing
and
hosting
and
a
lot
of
other
fun
things
so
I'll
be
curious.
C
A
D
Yeah,
thank
you
submit
a
request
to
see
if
we
can
get
just
like
a
kubernetes
cluster
or
something
on
cncf,
let's
build
a
cncf.
It
might
take
a
little
bit
of
back
and
forth
to
kind
of
figure
out
the
best
way
to
do
it.
The
other
option
is,
I
know
some
people
like
digitalocean.
D
I
could
go,
ask
them
hey.
Do
you
like?
Would
you
all
be
willing
to
like
donate
an
account
with
like
x
dollars
per
month
or
whatever
to
this,
and
then
we
can
just
that
would
probably
be
quicker.
D
D
Asia
and
if
any
of
those
people
would
like
to
navigate
their
internal
corporate
structures
in
order
to
figure
out
how
to
get
that
done,
then
they
are.
I
would
certainly
be
happy
to
host
this
on
multiple
clouds.
C
This
is
where
it
gets
tricky
because
yeah
whoever's
willing
to
host
you
know.
Obviously
you
should
get
some
sort
of
benefit
from
it,
but
for
those
who
don't
choose
to
host
there's
a
connotation
too,
I
would
say
from
an
azure
perspective
like
we're
working
on
an
azure
equivalent.
I
guess
the
easiest
way
would
be
just
like
an
aks
cluster
or
something
like
that,
but
that's
kind
of
just
a
small
part
of
our
business.
C
D
Cents
is
that,
like
azure
aws,
any
any
cloud
specific
one
like
you
would
want
to
run
your
version
of
it
right.
You
wouldn't
want
to
run
upstream.
You
want
to
run
something
that,
like
replaced
various
parts
with
managed
services
to
yeah.
You
know.
D
C
D
We
take
the
maintainers
call
I
can
go,
I
can.
I
can
put
in
a
request
with
the
service
desk
and
start
to
explore
like
cncf
options,
but
just
looking,
you
know
kind
of
my
gut
feeling.
I
know
I've
missed
a
lot
of
meetings,
but
looking
at
sort
of
the
issues
looking
at
the
like
gaps
on
like
metric
service
or
like
trace,
service
and
trace
service
looks
pretty
good,
but.
D
D
I
I
would
probably
say
something
like
hosting
something
on
mobile
clients
stuff
like
that.
That
stuff
sounds
great,
for
you
know
a
cube
county.
U
kind
of
time
frame
if
you
think
about
you
know.
I
I
personally
consider
this
as
a
like.
We
have
two
major
releases
a
year
and
one
is
cube.
Cutting
a
one
is
keeping
a!
U-
and
I
would
tend
to
put
this
in
that
second
bucket
of
we
can
work
on
this
next
year.
A
A
D
D
Second,
would
be
a
company,
a
end
user
company
that
does
not
have
a
stake
in
hotel
from
like
a
business
standpoint
like
the
digital
ocean,
or
something
third
would
probably
be
like
in
that,
and
that's
sort
of
the
line
in
my
mind
where
we
go
from
like
pretty
easy
conversations
to
much
harder
ones.
The
third
would
be
a
hyper.
D
You
know
like
a
hyperscaler
or
a
you
know
is
company
that
has
a
financial
interest
in
hotel
like
aws
or
google
or
microsoft,
and
then
the
most
challenging
ones
would
be
a
vendor
that
has
a
very
explicit
financial
interest
in
open
telemetry
like
datadog
or
lightstep
right
like
ever,
like
people
would
rightfully
get
onto
me.
If
I
said
oh
I'll,
just
use,
you
know
we
have
a
dev
account.
Is
there
a
good
example
right
now,
the
telemetry
that
comes
from
open20.io,
like
the
actual
collectors
for
that
are
running
on
like
lightstep
resources?
D
But
you
know
I
don't
put
that
anywhere,
because
that
would
sort
of
be
a
little
disadvantage.
That
would
be
that
wouldn't
be
kosher
in
my
opinion,
because
no
one
else
had
the
opportunity
to
do
that,
and
why
should
lightstep
get
free
advertising
just
because
they
were
there
first
right.
Thank.
D
D
It
feels
like
you
know
if
datadog
wanted
to
fork
this
and
run
a
fork
of
it
that
had
like
hey
click
here
and
you
can
and
like
you
created
this,
you
know
you
create
if
datadog
forked,
the
repo
and
did
a
you
know,
built
script
that
installed
that
sent
it
to
wherever
and
then
plugged
in
the
datadog
agent
and
built
had
some
debt,
pre-built,
dashboards
and
stuff
cool
right.
That's
in
your
fork,
like
we
don't!
This
is
open
source.
We
we
don't
care.
D
We
can't
control
that
that
is
completely
100
okay
and
we
would
even
link
to
your
fork
right
like
we
do
today.
We
we
will
link
back
to
everyone
that
creates
a
fork
of
this.
I
mean
within
reason,
but
in
terms
of
actually
like
the
main
one.
That's
where
we
have
to
be
more
careful.
C
C
But
I
agree
it
could
be
like
a
v2,
maybe
issue,
but
at
the
same
time
and
think
in
terms
of
thinking
of
like
telemetry
maturity,
I
think
having
a
live
version
would
help
us
build
a
better
kind
of
more
robust
telemetry
story
as
well,
if
we're
actually
using
it
to
operate
essentially
our
own,
like
hosted
version.
That
would
be,
I
think,
helpful,
that's
kind
of
where
I
see
some
value
but
yeah.
I.
D
That's
fair,
I
don't
think
it'll
take
that
long.
I
just
think
there's
yeah
well
I'll
start
asking
people.
I
will.