►
From YouTube: 2022-08-22 meeting
Description
Open Telemetry Meeting 1's Personal Meeting Room
C
As
you
can
see,
I
feel
like
we're
rounding
a
corner
here
and
coming
towards
the
finish
line
on
this
on
this
project,
at
least
on
the
first
release,
I
should
say.
A
Yeah
I
was
catching
up.
I
actually
spent
the
day
catching
up
the
the
whole
the
whole
thing
so
yeah
I
got.
I
had
like
180
emails
just
for
the
for
the
opentail.
A
D
Thank
you,
nash,
amiko,
hey
alexander
nice,
to
see
you
cool.
I
think
we
can
go
ahead
and
get
started.
I'm
still
filling
out
the
agenda.
So
I
guess
this
is
a
good
opportunity.
Are
there
any
items
top
of
mind,
you
know,
feel
free
to
add
them
or
call
them
out
now.
A
I
have
a
couple
of
questions
from
from
tests
that
I
was
doing
today,
but
no
not
actually
a
item,
so
maybe
I
could
just
shoot
them
whenever
we
we
have
some
time.
Don't
worry.
A
Good,
so
maybe
maybe
something
to
appear
that
changed
recently
the
low
gen
here
I
saw
that
we
do
not
have
a
login
to
the
that
generate
expense
to
the
ad
service
is.
C
Yeah,
hey
me
and
you
both
notice
that
and
recommendation
service
yeah.
I
think
we
need
to
talk
to
well.
The
new
front
end
definitely
changed
a
lot
of
things
and
I
think
what
it
is
is
that
it's
a
front-end
async
request
now
that
goes
off
of
the
ad
servers
and
recommendation
service.
So
that
is
one
of
the
things
that
I
think
we
need
to.
C
I
I
you
know,
ironically,
I
I
think
about
it,
but
it's
not
in
my
list
of
things.
So
thank
you
for
bringing
it
back
up.
So
I'm
going
to
add
it
to
my
list
of
things,
but
we
need
to.
C
A
Yeah
happy
that
I
wasn't
the
only
one-
and
there
is
another
thing
related
to
that.
I
also
noticed
that
the
low
gem
now
it's
a
orphan's
pen.
A
Orphan
spin,
so
like
we
have
like
login,
is
traced
by
itself
and
then
the
rest
of
the
traces
from
the
front
end
and
before
we
had
the
trace,
starting
on
the
login.
E
A
C
A
B
A
Well,
I
I
do
see
food
traces,
but
the
traces
are
not
starting
on
the
low
gen
anymore.
Well,
at
least
for
me,
I
I
just
started
again
and
I
have
like
the
low
gen
as
two
spans
and
then
the
rest
of
the
trace
in
the
front
end.
E
Did
I
mean,
I
think
there
was
a
generalized
feature
request
that
that
actually
be
how
it
works,
that
the
login
doesn't
or
that
the
front
end
creates
login
traces
as
through
links
rather
than
through
parent
child?
I
don't
know
if
that
actually
got
implemented,
but
that
might
explain
it
if
that's
okay,.
C
Julie,
I
know
I
see
your
problem,
I
see
it
as
well.
Okay,
all
the
get
product
calls
are
not
getting
traced
properly.
C
C
Just
all
the
http
gets
for
get
product
are
not
getting
continued
from
the
load
generator,
so
if
all
the
load
generate
does,
is
it
get
product,
it's
doing
its
thing
and
then
we're
doing
another
trace
on
the
front-end
side.
So,
for
whatever
reason,
that
trace
is
not
getting
continued.
E
Just
as
like
a
sidebar
or
to
continue
what
was
going
on
with
the
front
end
stuff
in
general,
like
the
locust
or
whatever,
is
just
directly
accessing
api
endpoints,
it's
not
actually
like
popping
a
browser
or
anything
right.
It's
not!
No
is
there
a
way
to
integrate
locus
with
like
puppeteer
or
anything.
D
Well,
I
I
don't,
I
think,
that's
a
good
transition,
so
we're
kind
of
in
the
last
month
before
v1
release,
so
I
think
we
need
to
start
narrowing
down
what
we
actually
want
to
be
present
in
that,
whether
that's
a
browser
you
know
generated,
trace
or
you
know
the
missing
php
service
or
we
optimize
our
build
time
or
something
like
that.
So
we
only
have
you
know
a
relatively
limited
selection,
so
integrating
selenium
or
playwright
or
puppeteer
or
any
other
frameworks
might
be
kind
of
a
big
much.
A
Sorry
character,
just
one
thing:
I
actually
don't
don't
think
that
this
is
an
issue
or
that
is
like
blocking
our
release.
So
we
can
release
with
this
broken
link
because
the
I
think
it
was
a
discussion
that
we
had
in
the
beginning
that
we
are
brought
up,
that
the
trace
shouldn't
start
on
the
login.
So
that's
definitely
fine
and
we
can.
We
could
even
remove
the
trace
from
the
we
could
remove
open
tail
from
the
low
gem
and
then
we
do
not
get
this
pin
at
all.
A
C
C
B
C
E
B
D
Well,
let's
talk
about
where
we're
kind
of
at
for
various
kind
of
remaining
pieces,
so
I
think
for
front-end
instrumentation
we
have
the
basic
instrumentation
set
up,
but
it's
missing
any
sort
of
baggage.
Today
someone
feel
free
to
correct
me
and
I'm
not
sure
if
we
want
to
add
any
other
enhancements
that
aren't
based
around
like
selenium
playwright,
any
of
those
to
our
front-end
instrumentation
as
well.
If
you
go
to
our
tracing
docs,
those
are
metrics
docs.
I
think
the
main
item
we're
missing
from
kind
of
our
v1
requirements.
D
E
C
Yeah,
that
was
the
idea
we
came
up
with
with
spam.
Links
was
to
generate
a
trace
on
the
login
side
and
then
have
the
front
end
capture
that
and
create
a
link
instead.
E
So,
let's,
let's
re-scope
this
front
end
this
low-gen
thing,
so
we're
not
going
to
remove
the
load
gen
and
if
we're
going
to
do
the
work
to
break
the
like
changing
the
propagation
from
parent
child
to
spam
link
is
to
most
things
the
same
as
removing
the
low
gen
from
the
trace,
because
it'll
be
two
separate
trace.
Ids.
E
So,
let's
like
that,
that
solves
that
problem
that
solves
two
problems
with
one
stone.
So
let's
say
we
add
baggage
in
we
find
one
place
to
extract
the
baggage.
Maybe
a
couple
places
to
extract
baggage.
I
don't
know
and
then
we
say
turn
login
to
front
end
into
a
link
rather
than
a
parent
child,
and
then
I
think
that
covers
us.
D
I
think
it's
receiving
it,
but
if
we're
not
passing
anything
down,
then
nothing's
being
picked
up,
oh
well,
then
I
mean,
I
guess
we.
E
Currency
doesn't
get
the
user
id
or
session
id
or
whatever.
Normally
I
don't
think.
D
E
D
E
E
C
E
Let
me
let
me
point
out:
I
10
000
do
not
want
us
to.
We
should
not
be
endorsing
people
using
the
hotel
baggage
as
a
side
channel
for
like
business
logic.
Stuff,
like
I
know
that
everyone
abuses
it
that
way,
but
we
like
we
should
only
pat
something
that's
passed
down
through
baggage,
should
only
be
used
for
observability
related
purposes
right,
so
the
currency
search
that
service
doesn't
have
a
way
to
get
the
session
id.
E
B
D
E
C
Track
class,
so
there
is
some
relation
right
there
with
the
span
link
thing,
because
when
we
do
that
spanning
thing
we
will
know,
are
you
coming
from
the
low
generator?
Yes
or
not?
And
if
you
are
you
get
the
synthetic.
E
C
E
C
E
I
mean
I
think
that
there's
a
like,
I
think,
the
the
way
you
could
talk
through
this
would
be
like
the
payment
service
knows
it's
synthetic
because
of
whatever
convoluted
reason,
but
from
an
observability
point
of
view,
maybe
our
slo
for
the
observer
for
the
currency
service,
maybe
our
slo
is
defined
to
skip
synthetic
traffic
to
the
payment
service.
Sorry
right,
like
that's
kind
of
the
story,
you're
telling
that
you're.
B
C
C
B
A
Quick
question
that
we
need,
so
the
baggage
in
all
other
services
will
just
be
a
propagation,
and
then
the
extraction
will
be
payment.
Is
that
right.
B
D
Do
we
have
a
specific
order?
We
want
this
work
to
follow
and
I'm
assuming
the
front
end
will
need
to
be
fleshed
out
first
potentially
to
spam
link
created
from
the
load
generator
and
then
payment
would
be
need
to
be
updated.
Or
can
we
do
payment
separately
just
depend
on.
D
C
B
C
I
think
the
hardest
part
will
be
the
front
end
where
we,
if
you
are
in
trace
context,
to
break
the
trace
and
use
links,
that's
going
to
be
the
the
crux
of
the
logic
is:
is
oscar
here,
the
iceland
coffin
oscar.
This
is
we're
going
to
lean
on
you
for
this
one,
I
think
sure
we
could
provide.
I
could
definitely
provide
some
guidance
on
it
if
you
need
any
assistance
here,
but
we
need
to
in
the
front
end
where
we
we
receive
all
the
requests
right
there.
B
C
That
we'll
have
the
load,
generator,
add
a
baggage
item
to
start
off
and
then
we'll
go
from
there,
not
a
baggage
item.
I'm
sorry,
the
load
generator
will
we'll
add
an
attribute
that
we
can
use
to
to.
C
So
to
span
links
would
be
created
on
the
front
end
and
what
you're
going
to
do
in
the
front
end
is
instead
of
continuing
the
trace,
you're
going
to
start
a
brand
new
trace
and
that
root
span
is
going
to
have
links
to
the
load.
Generators
trace,
okay,
okay,
it's
all
part
of
when
you
say
new
span
right
in
there.
In
that
call,
you're
going
to
say,
root
equals
root
equals
null
whatever
it
is,
and
spam
links
I'm
going
to
give
the
spam
links
details.
A
Pierre,
how
does
that
work
with
the
auto
instrumentation?
We
need
to
suppress
it
and
then
use
the
link.
Is
that
right.
B
A
D
Can
take
this
discussion
offline
too,
just
to
make
sure
we
get
to
the
rest
of
the
items
and
then
we
can.
We
can
talk
in
the
slack
channel
if
needed,.
B
D
Okay,
let's
talk
our
metric
story,
there's
a
whole
lot
of
construction
signs
present
and
then
we
have
you
know,
like
I
said
about
a
month,
maybe
a
month
and
a
week
to
go,
so
I
think
we
need
to
consider
you
know
what
what
support
we
want
to
add.
I
know
I
think
I
think
five
languages,
only
maybe
java.net
go
c
plus
plus
and
maybe
python
have
metric
support
ready.
So
we
do
have
a
relatively
more
limited
subset
of
features,
but
then
we
also
have
the
the
various
kind
of
requirements
around
these.
D
E
Like
whatever's
there,
like
java
java,
has
metrics
built
in
python.
D
E
E
If
josh's,
you
know,
if,
if
whatever
josh
is
going
to
do,
gets
merged,
then
cool,
then
we
update
the
thing
and
we
just
deal
with
it,
but
I
think
we
need
to
more
aggressively
like
close
on
these
issues.
If,
for
no
other
reason,
then
we
need
to
have
like
one.
E
You
know
there's
there
is
an
issue
I
believe
for
desired
metric
at
metrics
right
if
I
could
type
metrics.
E
E
D
It
might
might
be
a
good
demonstration,
some
hotel
capabilities.
E
Yeah,
I'm
just
trying
to
think
like.
E
E
E
B
Yeah,
what
about
the
metrics,
the
collector
metrics,
do
we
get
matrix.
F
E
F
E
Yeah
yeah
so,
let's
say
yeah,
add
docker
metrics
in
for
the
docker
one
kubernetes
for
the
kubernetes
one.
C
I
I
wonder
if
we
could
do
the
docker
one
through
extra
config
files
and
we
just
have
a
file
in
there
called
open,
telemetry
dash,
docker
config
and
just
like
how
we
have
the
the
exit
config
for
vendors
yeah.
We
do
an
extra
config
there,
which
includes
your
docker
receiver
stuff,
but
it's
only
part
of
the
docker
compose.
C
F
E
E
Well,
that
maybe
isn't
the
way
into
the
world
either
way.
I
think
we're
getting
a
little
too
detailed
right
now.
We
we
do
need
to
get.
I
think
if
we
can
get
like
hey.
This
is
running
on
docker,
then
get
docker
stats
if
it's
running
kubernetes
kubernetes
stats,
metrics,
throw
them
through
the
collector
boom
done
like
that,
satisfies
this.
For
now,
yeah.
D
Sounds
good,
okay.
This
is
an
item
kind
of
coming
from
pierre,
but
we
have
build
times
and
optimization.
That's
one
thing
we
potentially
want
to
take
a
look
at
as
well.
I
know
gary
eventually
opened
up
a
caching
improvement
for
docker,
which
will
help
at
least
on
the
docker
build
some,
but
now
we
need
to
look
kind
of
more
holistically
at
image,
size
and
build
time
as
well,
and
I
believe,
michael
opened
an
issue
to
track
image
size.
C
Now
that
we've
got
the
zero
three
images
I'll
do
zero.
Three
three
images.
C
There's
a
okay:
it's
still
there,
I've
seen
them
there's
a
pr
waiting
and
it
was
like
there
was
a
spelling
mistake
in
the
word.
Push
apart
that
pr
which
may
or
may
not
be
related
to
that
one
yeah
that
one
right
there.
C
C
E
E
Oops!
That's
why
you
don't
don't
do
any
of
this
from
well.
G
At
the
risk
of
saying
that
we
might
need
3.5,
I
did
notice
that
the
dot
n
file
always
defaults
to
the
docker
version
being
point
one
eight
or
something
like
that,
and
so
the
cache
doesn't
work
or
doesn't
pull
the
updated
version
by
default.
Unless
that
end
file
uses
the
most
recent
version
as
well.
So
if
you
want
to
use
the
cache
the
way
that
the
cache
is
intended
to
be
used,
meaning
that
it
like
pulls
the
most
recent
image
that
dot
n
file
is
really
critical.
E
B
Yeah,
I
didn't
figure
it
out
either
and
neither
did
wph
who
took
over
at
the
at
my
end
of
the
day
on
friday,
yeah.
B
G
No,
no,
no.
This
is
exclusively
for
the
purpose
of
when
you
build
locally.
If
you
want
it
to
use
the
cache,
then
it
has
to
point
to
the
latest
version
latest
is
better
than
using
the
exact
dot
version.
I
don't
think
that
that
actually
impacts
the
release
version,
because
if
I
remember
correctly,
the
github
file
uses
not
the
dot
end,
but
things
that
it
interpolates
from
like
the
release
tag.
E
E
There
was
a
change
to
the
runner
version
on
github
side.
That
was
the
only
thing
I
saw
that
was
different
between
the
two,
but
that
seems
like
such
a
mate,
like
that
seems
like
a
big
enough
change,
that
they
would
have
specified
it
somewhere,
and
I
didn't
see
anything
in
kind
of
release,
notes
and
trying
to
like
diff
stuff
in
the
github
actions,
repo
or
the
actions
runner
repo.
Like
I
don't
know,
I'm
going
to
cut
034
now
and
we
will
and
it
should
work
yeah.
E
Yes,
I
think
for
build.
It
should
just
for
local
for
release
builds.
I
don't
necessarily
care
about
caching,
because
whatever
it's
not
my
time
but
for
local
yeah,
if
we
can
pull,
do
oh.
G
I
remember
why
latest
doesn't
work,
it's
because
we
are
tagging
them
not
by
service
changing
the
actual
name
of
the
docker
image.
It's
all
tagged
as
open
telemetry.
G
E
Dash
whatever
what.
A
G
A
E
G
D
G
C
I
mean
when
you
release
helm
charts.
You
got
to
change
the
the
chart
yaml
file,
each
time,
anyways
right,
it's!
Yes,
it's
it's
a
thing.
E
Yeah,
okay,.
E
G
Yep
and
I'll
change
that
to
3
4
in
that
pr
that
I
have
open
just
so
it's
consistent
in
the
time
that
it
got
that
release.
E
E
A
F
F
D
Okay,
I
think
we
only
have
a
limited
amount
of
time
left.
So
pr,
unless
you
have
anything
specific
on
optimization,
should
we
maybe
quickly
talk
about
improving
the
vendor
work
experience.
I
know
an
issue
with
the
helm
charts
today:
they're,
not
configurable,
for,
like
other
kubernetes
implementations
easily,
or
at
least
it's
not
easy
to
like
alter
the
collector
of
that
piece
of
the
config.
So
I
think
I'd
like
to
make
at
least
an
enhancement
there
for
v1
to
make
kubernetes
more
customizable.
E
Signed
I
do
like
the
cur.
I
think
the
current,
at
least
from
a
docker
compose
perspective,
like
the
extra
file
is
good.
F
C
D
B
E
And
something
else
we
should
do.
I
guess
I
have
this
to
a
documentation
up,
but
we
should
make
sure
that
we
document
forks.
E
I
know
I
know
we
have
them
linked
in
the
readme
but
like
for
when
we
go,
we
do
like
doc
socks.
As
an
aside,
I'm
start.
I
am
trying
to
rustle
up
the
I'm
trying
to
push
to
get
this
on
main
stage
at
cubecon,
a.
C
Oh,
thank
you.
I
understand
we
don't
have
a
dave
to
show
this
off
and
maybe
carter
can
get
some
microsoft
dollars
to
help
sponsor
this.
For
the
tuesday.
B
E
Yeah,
you
should
talk
about
if
you,
if
anyone's
interested
in
this,
just
make
a
thread
in
hotel
dash
coms.
F
B
The
the.
D
F
Yeah,
no,
I
locally
changed
it
and
tried
installing
it.
That's
where
I
was
seeing
an
issue
which
is
shared
in
the.
F
B
D
Okay,
well
to
quickly
look
back
at
the
agenda.
I
think
the
the
only
real
missing
or
a
huge
missing
piece
we
have
right
now
is
that
missing
php
admin
service,
so
I'll
follow
up
with
the
dev
who
volunteered
to
do
that.
This
week,
he's
been
a
bit
m.I.a
and
then,
if
not
I'll
talk
to
some
of
the
sick
members
as
well
and
see
if
we
can
get
maybe
another.
E
D
E
B
For
any
of
the
helm,
tart
stuff,
if
you
open
the
issues
in
either
a
repository
but
ping
but
ping
us,
we
can
help
out
like
the
helm.
Chart
community
can
help
out
with
like
completing
those
issues
and
then,
whenever
the
three
3.0.3.4
images
are
released
today,
just
ping
me
and
I'll
get
the.
F
Thanks
tyler
I'll
create
some
issues.
I
know
the
problem
there
with
the
future
flag,
but
I
know
one
part
of
the
problem.
I
don't
know
the
whole
part
of
the
problem,
so
at
least
one
part.
Okay,
we
can
see.
E
B
We
made
wph
code
owner
for
the
helm,
charts
as
well,
so
that
we're
tied
in
on
both
communities
so
yeah.