►
From YouTube: 2022-09-12 meeting
Description
Open Telemetry Meeting 1's Personal Meeting Room
A
A
A
C
D
C
C
Yeah,
it's
interesting:
okay,
yeah
I
think
you
got
it
pulled
up,
so
I
just
had
a
couple
items
kind
of
top
of
mind.
I
wanted
to
talk
about.
So
first
we
have
the
PHP
service,
which
has
an
open
PR,
so
Juliano.
If
you
just
get
a
status
from
that
that'd
be
great
and
we
have
a
couple
remaining
kind
of
Trace
gaps.
So
I
wanted
to
get
some
eyes
on
I
think
it's
mainly
in
the
erlang
slash
feature
flag
service
and
then
also
maybe
one
thing
on
Rust
as
well.
C
Pierre
it'd
be
great
to
hear
about
kind
of
the
default
feature,
flag,
status,
update
and
then
just
a
couple
more
items
so
optimization
and
build
time
improvements.
Of
course
we're
close
to
V1.
C
So
this
is
kind
of
we're
closing
the
door
on
new
features
in
my
mind,
a
bit
outside
of
these
remaining
items,
but
we'll
have
those
kind
of
in
subsequent
versions,
and
then
we
actually
did
some
kind
of
internal
demos
to
some
internal
Microsoft
teams
and
I,
don't
think
of
Michael's
on
the
call
today,
but
he's
going
to
give
some
feedback
from
that
team
on
that
and
then
I'm.
Seeing
someone
adding
some
logs
Improvement
as
well
any
any
other
topics
people
want
to
cover
today.
F
Sorry
I
was
trying
to
find
my
zoom
and
mute
button.
What
about
the
kubecon
prep
announcements
sessions
to
talk
about
this?
If
any.
C
E
Cool
so
I
went
through
the
comments
from
Brett
and
Michael
solved.
Most
of
them
there
is.
There
are
two
issues
actually
that
I
cannot
solve
the
first
one
is
the
the
entry
point
in
PHP
we
pass
the
port
and
for
some
reason,
I
cannot
pass
the
environment.
Variable
I
tried
a
couple
of
different
ways.
So
if
you
scroll
a
bit
down
further,
no,
where
is
it
but
I?
Maybe
I
can
share
my
screen.
E
Okay
well,
but
the
the
main
point
that
I
want
to
say
here:
it
works
if
I
pass
the
environment
variable.
It's
fine
Docker
expose
the
part
for
me,
but
then,
whenever
I
try
to
use
the
same
environment
variable
here
in
the
entry
point,
PHP
fails.
I
asked
Brett
to
see
if
I
could
get
like
try
to
do
in
a
different
way,
but
I
always
get
the
embedded
address.
E
So
I
put
here
some
different
ways
that
I
tried,
but
all
of
them
failed
so
I'm
not
sure
how
to
do
that
or
if
it
is
possible.
Maybe
there
is
another
way
of
starting
that
instead
of
PHP
that
dash
s
and
but
then
for
that,
I
need
someone
from
PHP
and
there
is
another
comment
from
Brett
that
I
don't
know
how
to
solve.
Where
is
it
here?
So
he
mentioned
something
about
install
the
protobuf
extension.
E
The
apps
is
running
and
working
at
the
moment.
So
I
don't
know
what
exactly
he
meant
here.
If
he
wants
me
to
replace
the
grpc
that
we
have
or
just
add
the
protobuf
extension
or
yeah.
So
whenever
he
comes
back,
then
I
think
we're
good
to
go.
F
So
your
issue
is
you're
using
the
wrong
type
of
shell
expansion
for
your
entry
point
command,
get
rid
of
the
brackets,
don't
pass
it
like
an
array
just
pass
it
with
spaces
between
each
one.
Instead,
so
look
at
the
currency,
Services
Docker
file,
and
you
want
to
use
that
format.
Perfect,
sorry
and
then
you'll
be
Ableton
variable
through
I.
Think
so,
because
we
hit
the
same
thing
with
the
Currency
Service
and
that's
how
awesome
awesome.
Thank.
E
Cool
yeah,
so
that
would
live
just
for
the
this
protability
and
at
the
moment
I.
E
Don't
I
do
not
have
any
traffic
on
the
on
the
PHP
service,
but
I
think
I
can
do
a
call
and
if
not
from
the
rest,
if
not
I'll,
just
being
Gary
and
ask
for
some
help
there
but
I
think
it's
pretty
easy
to
do
a
call
from
list
to
there
and
that's
the
the
the
currency
I,
the
the
PHP
service
was
already
getting
too
large
and
then
I
decided
yeah,
let's
get
it
merged
and
then
we
add
the
traffic.
Because
then
it's
easy
to
to
see
the
things.
C
Okay,
great,
that
seems
like
we're,
making
some
good
progress
just
a
little
more
things
to
buckle
down,
but
we
really
appreciate
you
putting
in
the
work
on
that
Giuliano.
That's
much
appreciated!
Okay!
So
let's
just
talk
quickly
about
the
remaining
Trace
Gap,
so
I'm
gonna
share
my
screen.
B
C
Okay,
so
just
looking
at
the
docs
real,
quick,
yeah
I
have
a
feeling.
Some
of
these
are
probably
just
out
of
date,
so
someone
could
take
a
look.
That
would
be
super
helpful,
but
it
looks
like
manual
span
creation
and
span
data
enrichment.
Here
you
probably
used
a
feature
flag
service.
The
most
recently
is
this.
So
missing.
F
It's
still
missing:
okay,
I'm,
not
sure
how
it's
done.
We
were
going
to
nerd
snipe
Fred
at
honeycomb
for
this,
and
then
he
was
on
vacation
and
then
last
week,
of
course,
honeycomb
had
a
not
very
good
operational
week.
I
will
leave
it
to
that
and
I'm
gonna
try
the
nursing
cred
this
week,
but
just
last
week
the
rsre
team
was
working
some
very
long
hours.
You
know
there
was
no
way
I
was
going
to
bother
them.
C
Yeah,
okay,
yeah!
That's
my
problem!
Okay!
So
we'll
get
some
attention
on
this.
That's
fine
and
then
it
looks
like
gash
been
a
product
catalog
as
well,
which
seems
a
little
random.
F
We
should
probably
fix
that
it's
just
creating
a
span
right.
Do
we
have
another
go
service,
checkout?
Does
this
yeah?
Can
we
just
say
not
applicable.
C
F
F
Yeah
somebody's
looking
to
do
this
Amino
span
creation
and
go.
We
have
a
good
example
for
them
in
the
checkout
servers
for
this
I
I
would
prefer
we
made
this
not
applicable,
and
then
perhaps
in
the
the
docs.
F
We
definitely
need
to
figure
out
how
to
document
how
we
achieved
all
these
things
Point
people
to
where
to
look
in
code.
We
should
just
kind
of
mention
for
go.
Go
here.
C
F
Gonna,
try
and
fret
again,
you
know:
I
know
how
many
more
favors
we
could
get
out
of
Tyler
I'm
I'm,
definitely
not
knowledgeable
in
Elixir.
To
do
this.
Yeah.
C
Yeah
I
can
paint
Tristan,
we
couldn't
see
if
we
can
go
I'm.
Sorry,
not
Tyler,
Christian,
yeah,
yeah,
we'll
see
what
happens
but
So.
Speaking
of
the
feature
flag,
I
think
you
implemented
a
default
feature
flag
status,
but
then
I
went
on
vacation.
I
didn't
get
a
chance
to
try
it.
So
you
can
maybe
like
yeah.
F
F
I
think
we
wanted
to
do
an
additional
stability
check
for
the
future
flag
service
and
that
if
postgres
doesn't
start
up
on
time,
there
could
be
an
issue
there
and
you
might
have
to
just
restart
the
feature
flag,
service
and
you're
fine.
F
But
you
know,
do
we
started
Maybe
doing
it
in
like
in
in
a
container,
but
you
don't
really
have
the
same
concept
of
Docker
compose,
but
something
that
just
forces
the
feature
flag
server.
So
wait
on
postgres
to
be
ready
before
it
actually
starts
and
that's
kind
of
a
little
bit
of
the
issue.
There.
F
It
waits
for
ready,
it
waits
for
ready,
but
I
I,
guess,
there's
a
weird
state
where
postgres
ready
will
say
I'm
ready,
but
it's
not
quite
ready
yet
to
accept
a
connection
so
I'm
not
clear
on
that
I
I.
It's
happened
like
once
for
me,
but
not
again,
so
I'm
not
sure,
and
that's
because
I
do
a
lot
of
start.
No
cash
just
to
really
flush
things
out.
Every
time.
C
Yeah
and
then
for
the
default
flags.
Are
they
available
in
that
UI
admin
experience
or.
C
F
They're
they're
available,
if
you,
when
you
started,
we
have
the
link
there
to
to
get
into
the
localhost
Port,
whatever
we
picked
there
for
the
future
flight
service.
If
you
go
to
it,
you'll
see
two
feature:
Flags
listed.
C
E
F
F
Yeah
yeah-
that's
that's
probably
where
that
is
at,
and
we're
about
to
make
some
further
modifications
to
the
shipping
service.
So
we're
going
to
be
calling
both
the
PHP
thing
now
and
this
feature
flag
from
within
it
I
think
it's
the
last
remaining
big
code
gaps
that
we
have
are
on
shipping
and
getting
this
PHP
service.
Pr
finished.
C
C
Okay
feature
flag
is.
C
For
optimization
and
build
time
improvements,
this
is
honestly
just
a
giant
open
question,
so
I
would
love
to
hear
you
know
as
many
ideas
suggestions
or
that
type
of
thing
kind
of
as
we're
going
forward,
and
this
is
also
probably
going
to
be
an
ongoing
kind
of
question
as
well
too.
So
you
know
obviously
we're
not
looking
for
Solutions
necessarily
today,
but
we
do
have
quite
the
build
time
to
cut
down
a
little
bit.
A
C
Yeah
I
think
we
had
started
something
with
Gary
was
using
like
the
catch
from
with
the
docker
capabilities,
and
then
I
think
he
was
pointing
that
to
the
docker
Hub
like
latest
version,
and
then,
ideally,
you
wouldn't
have
to
rebuild
the
full
package.
Every
time.
Gary
was
that
that
correct
yeah.
E
E
I
yeah,
it
also
failed
to
start
on
my
machine
and
I
got
a
message
from
someone
that
was
trying
to
to
use
in
and
got
the
same
error
message.
So
three
users,
I,
don't
know.
Is
it
a
versions
issue
or
something
because
as
far
as
I
know,
I'm
using
the
latest
Docker
compose?
E
B
I,
don't
know
how
yeah
the
cash
from
I
believe
comes
from
buildkit,
which
I'm
not
sure
how
much
consistency
they
have
across
different
installations
and
what
like
add-ons
you
might
have
to
have
for
that
to
work
automatically
it.
You
know
classic
case
of
it
works
great
on
my
machine,
but
it's
really
hard
to
test
without
putting
it
in
other
people's
hands.
E
Yeah
because
I
think
I
think
if
we
can
document
like
hey,
please
install
this
this
and
this
dependencies
and
then,
if
that
saves
a
lot
of
time,
then
it's
great
for
us
but
as
it
was
crashing
to
start,
we
decided
to
just
comment
out
just
to
to
have
the
the
app
running.
B
Yeah
I
would
be
happy
to
debug
that
with
somebody
who
actually
had
the
issue
because
I
I
mean
I
didn't
have
the
problem
and
I
don't
think
that
I
have
anything
exceptional
installed
on
Docker
like
I'm,
not
let
this
Aficionado
that
has
40
extensions
for
my
Docker
installation,
I
have
the
default,
Docker
desktop
and
then
I
was
even
I.
Think
I
was
a
couple
of
versions
behind
when
I
was
testing
this,
so
it
shouldn't
have
been
that
he
had
to
be
latest
and
greatest
for
it
to
work.
B
B
C
Great
and
I
think
Michael
Maxwell
at
least
started
issues
to
track
like
the
service
like
data
size
or
storage
size,
and
we
can
also
potentially
do
something
like
that
with
the
performance
information
as
well
and
then
just
kind
of
slowly
improve
on
it
from
there.
C
Let's
see,
okay,
so
yeah
I,
don't
think
Michael
Maxwell's
on
this
call.
So
we
had
two
questions
around
ad
service
traces
not
showing
up,
and
then
logging
Improvement
so
who's
that
whoever.
C
A
F
Can
explain
I
can
explain
why
so
previously,
when
we
had
to
go
based
front
end
and
you
hit
an
end
point,
the
go
base
front
end
as
a
back
end
would
do
all
the
API
calls
to
all
the
other
microservices,
including
the
ad
service
recommendation
check
out
and
so
on
so
forth,
and
then
render
a
go
template
back
up
to
the
end
of
the
ship
today,
with
the
next
JS
based
front
end
the
browser
downloads,
a
bunch
of
JavaScript
and
then
that
makes
further
API
calls,
including
calls
to
the
ad
service
card
service
to
checkout
service
and
so
on
and
so
forth.
F
F
Either
we
need
to
make
sure
it's
exposed
as
an
API
route
and
we
just
hit
that
API
Row
from
the
load
generator
or
we
expose
it
and
hit
that
API
right
from
the
load
generator.
So
so
like
an
ad
service
route
or
something
like
that.
A
F
Do
see
in
in
in
Gateway
here,
for
we
we
one,
we
should
create
an
issue
for
it
and
I
see
something
that
looks
like
API
routes
here.
I
got
a
remember
how
to
work.
My
next
JS
stuff
here
again,
but.
F
G
F
A
Could
next
logs
Improvement?
This
has
nothing
to
do
with
version
one,
but
I
saw
like
we
have
a
lot
of
inconsistencies
around
logging
of
each
application
like
I,
wanted
to
open.
If
it's
okay,
if
you
can
actually
do
like
a
consistent
logging
across
our
application,
even
among
the
two
go
applications
we
have
in
the
checkout
and
the
product,
catalog
one
is
doing
Json.
Another
is
doing
a
text
formatting
like
it's,
not
consistent
across
the
board.
So
like
do
we
think
it's?
Should
we
target
that,
or
even
had
some
improvements
around
this.
F
Yes,
we
should
I
am
indifferent
on
what
we
decide
and
what
the
logging
format
should
be.
F
A
F
I
I
work
at
honeycomb.
We
accept
Json
format
as
well,
I
think
if
it
comes
to
doing
for
myself,
we
should
be
thinking
about
targeting
otlp
logs
for
this
and
then
having
no
PLP
providers,
export
console
and
two
I
collector
backend,
or
something
like
that
I'm
not
sure
so.
I'm
I
want
to
be
cautious
on
how
much
effort
we
put
into
using
console
like
generic
console
routers.
A
F
Tology
logs
now
I
thought
Java
added
support
recently
for
it,
and
what's
the
other
language
decided
to
support
for
it,
it's
coming.
It's
it's
got
work
to
do,
but
if
we
were
having
this
conversation
at
six
months,
I
think
we'd
definitely
be
chatting
about
all
photography
log
support
being
greater
than
what
it
is
today.
B
F
Yeah
javapython.net
are
so
far
the
only
languages
that
have
some
form
of
support,
probably
Java
the
furthest
along.
C
Okay,
great
so
do
we
have
any
other
specific
items.
We
want
to
get
some
attention
on
today
or
we
could
potentially
yeah.
D
The
last
one
last
one
is
still
there
I
just
added
the
fixing
tests
configuration
and
there
yeah
I
have
been
testing
this.
It's
great
those
carries
in
the
call
and
yeah
we're
not
I've
been
testing
these
cases,
both
on
the
on
my
Mac
and
the
Linux
messy
and
I
I
get
this
different
problem.
This
I
don't
know
this
error
response
from
demon
Network,
something
not
found
and
Juliano
got
I
think
something
else
when
reading
the,
but
but
anyway
it's
there's
it
yeah.
C
E
It's
good
that
Oscar
is
in
the
engine
call.
Maybe
maybe
he
can
tell
us
because
I
think
I
I
just
got
a
error
on
the
checkout
test
and
he
mentioned
something
about
the
currency,
not
returning
the
the
right
number
of
currencies.
E
G
E
G
Is
that
an
error
you
got
from
the
integration
test
or
for
the
front-end
tests
front
end?
Okay,
let
me
see
the.
E
Integration
went
through
in
all
of
them,
so
17
tests
there
test
and
then
in
the
front
and
test
I
got
one
of
three
failed.
G
Okay
is:
is
that
a
I
don't
know,
there's
I'm
dealing
with
this.
We're
really
we're
press
condition.
I,
don't
know
what's
going
on
between
one
of
kind
of
next
year's
rendering
the
page,
and
sometimes
she
doesn't
have
the
enough
information
on
time
to
have
all
of
the
rendered
data
yeah.
But
I
would
like
to
see
if
you
have
like
any
exchanges
that
you
want
to
that,
you
could
share
with
me
or
some
information,
maybe
I.
G
E
I
haven't
test
after
you
merged
Maine
into
the
fixed
test,
but
as
you
have
the,
as
you
haven't,
changed
anything
so
I
guess
that
was
the
same.
But
I
can
try
it
out
again
yeah,
but
in
mikko's
case
he's
getting
two
different
errors
on
two
different
hosts.
G
So
yeah
I
think
that's.
That
may
be
because
it
probably
requires
the
application
request
to
to
remove
all
of
the
services
and
it's
been
spin
it
up
from
scratch
again.
G
What
happens
is
that
when,
when
you
switch
from
different
versions
of
the
docker
compose,
sometimes,
if
you
don't
do
Docker
compost
down,
you
will
get
some
resources
up
that
doesn't
match
with
the
current
Docker
compose
version
that
you
have
and
if
you
try
to
run
it,
there
could
be
some
network
interfaces
or
resources
that
are
not
available
yet
yet
or
because
of
the
new
version
of
the
docker
compose.
G
G
Sure,
let
me
know
you
have
an
issues.
I
think
this
is
an
important
paper.
I
I
still
need
to
figure
out
that
we're
really
weird
race
condition,
I
I
expected
to
finish,
then
the
previous
change
that
I
did
but
I'm
not
sure.
Probably
it's
still
there,
even
though
we
have
that
small,
weird
failing,
sometimes
fails,
I
think
it
it
makes
it
will
be
good
to
have
at
least
some
of
the
fixes
that
I'm
adding
on
this
PR.
G
For
instance,
it
fixes
the
way
that
we're
running
the
integration
tests,
because
if
you
go
right
now
to
the
main
branch
and
try
to
run
them,
it
will
fail.
So
I,
don't
know
I,
don't
know
what
to
do.
There.
G
E
By
the
way
here
it
worked,
the
the
entry
point
changes
that
you
suggested
I
already
pushed.
Thank
you.
C
So
I
think
our
last
item
right
now
is
documentation.
So
if
you're
not
sure,
if
you
had
some
specific
kind
of
thoughts
already
on
how
you
wanted
to
push
the
documentation
or
if
you
just
wanted
to
raise
it
in
general,
so.
F
C
F
Have
two
approaches
here
for
dogs:
one
is
a
language-centric
approach.
The
other
one
is
a
service-centric
approach
and
my
ideas
here
is:
if
it's
language
specific,
we
write
one
DOC
for
go:
here's
how
you
do
traces,
metrics,
logs,
whatever,
as
we
go
out
through
and
with
links
to
specific
examples
of
how
we
do
it
to
the
services.
F
F
But
then
the
other
approach
is
to
do
this
service
specific.
So
for
each
service
we
write
a
readme
which
describes
how
we
did
all
the
things
within
that
we
duplicate
some
Services.
A
little
bit
I
think
go
is
probably
the
only
one
left
that
duplicates
python.
Do
we
still
duplicate.
C
F
C
C
E
B
B
F
C
Yeah
I
think
Austin.
Sorry
to
interrupt
me.
Go
previously
mentioned
that
the
readme's
and
the
actual
Services
should
just
be
focused
around
like
actual
like
development,
and
then
the
documentation
should
be
kind
of
the
like
user
entry
point
so
to
speak.
Okay,
so
that
aligns
is
what
I
think
too.
Okay.
F
So
that
that
that
helps
out
things-
and
it's
just
a
matter
now,
just
unfortunately,
you
know
nobody
like
to
write
docs,
but
we
gotta
write
them,
though.
C
Yeah,
it
makes
sense.
So
what
is
service
based
stocks?
It
sounds
like
that's
what
we
want
to
do
and
then
I
think
Juliana
mentioned
like
honeycomb,
and
you
guys
used
to
have
a
pretty
good
structure
for
that.
Do
you
think
you
have
like
a
good
kind
of.
E
F
F
No
I'm
going
to
use
the
same
structure
we
did.
We
did
for
honeycomb.
We
did
the
same
thing.
We
did
service
based
stocks
as
well.
We
you
know
we
put
them
all
inside
the
the
individual
service
readme's,
but
I
I'm,
gonna
and
those
docs
are
no
longer
valid
versus
what
we
have
here
today.
I
mean
use
the
same
approach,
so
the
first
section
is:
how
do
you
initialize
the
provider?
How
do
you
just
initialize
tracing
some
of
these
languages?
It's
ridiculously
simple!
F
It's
done
in
the
docker
file
itself,
which
is
kind
of
cool
like
like
python.
It's
just
adding
a
Docker
file.
Well,
maybe
not
sorry
Java
Society,
the
Java
agent
there
so
cover
that
part
and
then
I'm
going
to
get
into
the
manual
span.
Creation
manual
span
attributes
if
there's
any
events
or
links
thereafter
and
baggage,
probably
at
the
bottom,
and
then
we'll
do
another
heading
to
for
metrics
and
a
heading
three
for
logs
or
heading
two
for
locks
and
we'll
get
there.
C
Yeah,
that
makes
sense.
Could
you,
when
you
do
this,
just
maybe
submit
one?
Isn't
it
like
one
service
doc
as
an
example,
and
then
we
can
kind
of
iterate
on
that
and
the
pr
and
then
widen
it
out
to
the
other
services
and
I
can
do
some
of
these
documentation
stuff
too.
So
it
shouldn't
all
be
on
you
as
well.
C
F
B
C
Sense,
okay,
I
think
we're
around
10
minutes,
so
we
still
have
plenty
of
time.
Josh,
I
think
this
is
your
first
meeting
welcome.
We
appreciate
your
help.
Is
there
anything
you're,
particularly
interested
in.
B
So
I
am
I
was
an
Elixir
developer
in
a
past
life,
so
helping
out
with
the
feature
flag
service
is
probably
something
I
could
do,
but
I'm
also
somewhat
new
to
the
project.
So
probably
we'll
need
a
little
bit
of
guidance,
but
yeah
hi
happy
to
be
here.
Okay,
that.
F
Oh
boy,
Josh,
let's
connect
it
in
slack,
but
we'll
do
super
duper.
High
level
is
we
need
to
get
some
manual
span
creation
and
attributes
done
as
part
of
the
feature
flag
service
so
somewhere
in
the
service
code
itself.
Just
need
to
kick
off
a
span
had
attributes
and
I
want
to
also
had
attributes
to
stand
in
context
to
one
of
the
auto
instrument
and
spans.
So
I'll
cover
more
details
of
what
I'm
thinking
of
where
it
seems
to
be
done
and
I'll.
Let
you
how
about
it
sounds
good.
B
C
Amazing,
but
we're
happy
happy
to
have
that
on
the
team
regardless
so
welcome,
but
yeah
I
think
I
think
we
call
it
here,
guys
I,
really
appreciate.
You
know
everyone
taking
the
time
and
joining
as
well.
F
I'm
eager
to
find
out
about
the
feedback
from
Microsoft,
so
if
you
have
a
way
to
decimate
that
and
we
don't
need
to
do
a
sync
for
it-
I'd
love
to
hear
it.
If
not,
hopefully,
Michael
can
make
the
next
meeting
yeah.