►
From YouTube: 2023-03-13 meeting
Description
cncf-opentelemetry@cncf.io's Personal Meeting Room
A
A
I
I
was
worried.
I
I
tried
at
the
wrong
time
because
it
looks
like
the
the
US
they
how
you
say
like
they
switched
to
Daylight
Saving
Times.
Now
right
so
correct.
We
did
and
we
didn't
so
so
we
now
have
two
weeks
of
like
everything
is
different.
That's
annoying.
B
With
any
luck
this
year
and
next
year,
they
will
finally
between
Canada
and
the
US,
or
at
least
on
the
west
coast,
get
rid
of
daylight
savings.
A
B
Yeah
I'm
not
optimistic
for
like
us
wide,
but
certainly
like
California,
Oregon,
Washington
and
British
Columbia
and
Alaska.
They
have
their
own
private
thing.
Where
I
think
once
Oregon
passes
it,
then
they
will.
They
will
stop
doing.
B
Probably
but
I
think
that's
already
the
case,
yeah
I,
don't
know,
but
well
I,
don't
think
in
Canada,
but
I
think
in
the
United
States
there's
already
one
or
two
states
that
don't,
and
so
there
are
some
time
zones
already
that
there's
some
states
next
to
each
other.
That
are
one
hour
apart
for
part
of
the
year
and
the
same
time
zone.
Other
parts
of
the
year.
C
Do
you
know
what
the
West
Coast
folks
are
going
to
settle
on
whether
they're
going
to
settle
on
which
time
they're
going
to
settle
on
I.
C
B
Speaking
selfishly
I
prefer
the
summer
one
I
I
prefer
to
have
my
sunlight
in
the
winter
later.
In
the
day,
instead
of
in
the
morning,
we
will
see
what
happens.
Yeah.
D
A
And
Germany,
at
least
for
me
to
coordinate,
and
then
it
always
depends
like
who
put
this
appointment
into
the
calendar
right
so
yeah.
Suddenly,
sometimes
it
moves.
Sometimes
it
doesn't
so
it
really
depends
on
that.
B
A
Awesome,
let's
see
I'm
not
sure
if
Philip
is
going
to
join
since
he
had
added
two
things
to
the
agenda,
but
maybe
he
will
join
in
a
minute.
Let
me
get
started.
Anyways
awesome,
Patrice,
I.
Think
the
first
one
came
from
you
about
the
bootstrap
update.
C
C
A
I
mean
thank
you
for
making
this
possible
I
guess
this
was
a
bunch
of
work,
I
think
visually,
not
not
a
lot
changed
right,
I
mean
for
now,
but
maybe
at
some
time
it
would
be
called
or
I
will
maybe
also
spend
some
time
on
seeing
like
what's
new
in
doxy.
If
there's
anything
valuable
for
us
or
is
there
anything
you
can
call
out
specifically
where
you
say
like
hey?
E
A
A
D
Yeah
that
was
me,
so
this
was
a
thing
that
came
up.
We
had
a
like
a
governance
committee
like
in-person
Summit
with
the
TC
last
week,
and
you
know
one
of
the
things
that
was
coming
up
is
like:
how
do
we
get
information
from
end
users
better?
D
How
do
we
communicate
it
back
out
and
something
relevant
to
this
group
people
point
out
that
mainly
what
we
do
is
we
tell
people
to
come
to
slack
with
us
when
they
have
questions,
and
so
then
they
come
into
slack
and
people
are
actually
generally
pretty
good
at
answering
questions
there.
But
generally
that
means
you
know
typing
in
a
paragraph
into
slack,
so
that
person
gets
their
answer,
but
then
that
information
is
just
lost
because
slack
is
not
really
searchable
or
anything
of
that
nature,
and
so
we
were
floating
like
what?
D
D
You
know
some
chat
based
way
of
interacting
with
people,
but
when
we
do
find
ourselves
typing
out,
like
a
paragraph
answer
like
what
we
were
trying
to
spitball
about,
like
what
would
actually
be
a
good
Next
Step
to
ensure
that
that
information
does
get
search
indexed,
and
maybe
we
can
just
point
someone
to
a
link
later
and
so
I
wanted
to
bring
that
to
this
group.
My
thoughts
immediately
went
to
stack
overflow
I
mean
some
things
might
be
like.
D
C
Weeks
so
stack
Overflow
seems
like
a
reasonable
option
and
I
agree
about
the
disadvantages
of
slack.
There's
been
a
push
from
the
cnz
up
side
to
try
and
encourage
projects
to
go
to
at
least
GitHub
discussions
because
it
because
it
is
open,
indexable,
searchable
Etc.
D
Though
I
I
think
we,
we
were
also
discussing
killing
off
GitHub
discussions
because,
for
whatever
reason,
they're
not
working
well
and
honestly,
I,
don't
I,
do
not
find
GitHub
discussions
to
actually
be
I,
mean
I'm,
sure,
they're,
technically
search
and
indexable.
But
I
don't
know.
Google
is
generally
that
point
I.
B
C
So
it's
been
my
understanding
that
they
are,
they
they've
been.
They
should
get
hit.
Just
as
issues
do
it
may
be
that
they're
just
not
being
adopted
and
hence
not
showing
up
in
searches,
so
it's
kind
of
a
chicken
and
egg
problem,
but
if
everybody's
hap,
if
if
most
folks
on
this
call,
are
happy
with
stack
Overflow,
then
we
know
we
know
that
gets
a
lot
of
hits.
We
know
that
that's
a
go-to
place.
A
So
so
what
what
we
have
in
I
can,
let's
say
an
internal
perspective.
We
we
have
someone
here
who
is,
let's
say
responsible
for
our
let's
say
internal
question,
discussion
platform
whatever,
and
he
then
goes
into
let's
say
different
slack
channels
and
says:
like
hey,
there's
a
question:
can
someone
answer
that
so
I
think
what
would
help
as
someone
who
monitors
stack,
Overflow
and
then
maybe
pushes
those
questions
down.
A
Let's
say
someone
asks
about
like
hey:
how
can
I
instrument
a
Java
spring
boot
application
whatever,
but
let's
come
up
with
something
someone
needs
to
tank
the
question
and
brings
it
down
to
the
Java.
Stick
right
and
they
say
like
hey
here's,
the
answer
and
I
said
they
go
there
themselves
or
answer
it
right.
I
mean
that's
the
one
way,
but
the
other
way
is
also
like
someone
asks
the
same
question
on
Slack.
Who
is
then
taking
that
question
back
to
stack,
Overflow
and
asking
it
there
and
answering
it
there
so
I
think
stack.
A
F
If,
if
I
might
offer
a
thing,
I
mean
it's
yes
stack.
Overflow
is
awesome,
there's
clearly
a
shitload
of
people
there,
like
three
hours
ago,
somebody
just
asked
a
new
question
about
a
hotel,
so,
like
people
are
going
there,
whether
we
want
them
to
or
not
I
like
I
mean
I
feel
like
anybody
working
on
the
project
should
feel
empowered
to
just
go
in
and
answer
questions
like
this,
like
not
like
you
know,
say,
there's
like
a
problem
with
Java
or
something
like
the
javasig
shouldn't
have
to
necessarily
be
like.
F
Oh,
you
know,
we'll
go
and
help
these
people
it's
like,
but
like
that
doesn't
happen
out
of
nowhere.
I
do
I.
Do
wonder
if
there
is
a
bit
more
of
a
concerted
effort
to
be
like
hey,
like
Foster
a
little
bit
of
a
culture
within
your
slice
of
hotel
to
you
know,
be
like
hey
yeah,
let's
go
out
and
help
people
where
they
are
I,
don't
know
how
that
looks,
but.
D
Yeah
yeah
I,
wouldn't
that's
kind
of
like
the
source
of
this
query.
Is
you
know
if
we
we
can
all
agree?
That
slack
is
like
the
worst
place
to
leave
information,
even
if
that's
definitely
where
people
are
gonna
like
be
asking
us.
So
we
want
to
leave
that
door
open
it's
just
if
I'm
gonna
try
to
go,
encourage
the
community
to
like
adopt
some
other
habit.
I
wanted
to
to
pass
it
by
you
guys,
first,
because
one
it'll
be
yeah
trying
to
get
people
to
make
stack.
D
Overflow
accounts,
seven
I
think
that
was
a
really
good
point
that
actually
maybe
you'll
get
flagged.
If
you
answer
your
own
questions
there,
which
is
a
little
Annoying
you
know,
maybe
we
could
ask
people
to
ask
the
question
there
and
we'll
answer
it
there,
but
that's
annoying
for
them.
D
So
so
maybe
some
some
practical,
you
know
Road
blocks
I'll,
look
into
that,
but
the
other
part
is.
It
does
mean
as
a
community.
We
need
to
kind
of
like
also
develop
a
practice
of
triaging
stack
Overflow
right.
If
we're
gonna
start
driving
driving
people
there
and
I
do
worry
a
little
bit
that
it's
like
part
of
the
reason
why
we
want
to
not
even
part
like
the
main
reason
we
want
to
shut
down.
D
Github
discussions
is
because
we
kind
of
like
turned
them
on
in
a
bunch
of
places,
but
people
don't
on
the
project,
don't
really
like
look
there.
It
just
doesn't
feel
like
it
panned
and
turned
into
something
that
was
like
useful
for
our
community
and
I.
Don't
want
to
like
go
through
the
trouble
of
driving
everyone
to
stack
Overflow
and
just
have
the
same
thing
happen.
F
One
thing
that
I
really
love
that
the
ghost
sick
does
is,
at
the
end
of
every
call.
There's
there's
always
this
sort
of
like
hey
just
like
open
space
for
people
are
using
our
like.
What
have
you
heard,
maybe
that
that
could
be
something
that
other
cigs
are
encouraged
to
just
sort
of
ask,
as
just
sort
of
like
a
formalized
like
Hey,
we're
doing
something
that
people
use
and
what's
that
all
about,
and
maybe
a
part
of
that
could
be
like.
F
And
you
know,
oh
maybe
we
could
take
a
look
at
that
at
some
point.
I
feel
like
that
kind
of
I'm
looking
for
the
right.
That
kind
of
like
ritual
can
can
help
with
that.
D
C
C
What
could
happen
is
that
it
may
help.
If
folks,
who
are
on
slack,
could
kind
of
direct
people
to
stack
Overflow
and
say.
Could
you
ask
your
question
there
yeah,
and
maybe
it
would
help
if
we
would
have
a
custodian
of
Hotel
slack
overflow
for
maybe
I,
don't
know
a
quarter
and
then
we
could
that
person
could
rotate.
But
at
least
there
would
be
one
name,
and
we
could
put
that
on.
You.
E
D
It
would
be
nice
if
we
had
some
kind
of
backstop,
yes
of
someone
who
at
least
was
like,
did
a
triage
on
some
level
to
notice,
if
like
something
wasn't
getting
answered
or
was
like
turning
into
a
troll
Fest
to
at
least
go
back
to
slack
and
like
point
the
correct
community
at
the
at
the
issue
and
say,
could
you
look
at
this?
Okay.
A
I
I
mean
I
I.
Think
a
good
starting
point
would
really
be
like
I
mean
use,
something
like
the
maintainer's
call
and
tell
them
like
what
Philip
said
like
hey
build
a
ritual,
let
around
that
yeah
and
another
thing
I'm.
Just
thinking
about
was
the
let's
say
going
from
slack
to
stock
overflow
I
think
at
the
end,
if
someone
asks
a
question
and
if
let's
say
someone
gives
an
answer,
I
think
telling
the
people
like.
A
Oh
that's
a
really
good
question:
can
you
put
it
over
to
to
stack
Overflow
and
then
ask
it
yourself
or
for
everybody
else,
I
mean
it's
also
kind
of
appreciation
to
to
the
people
asking
questions
I
mean
we
should
not
do
it
all
the
time
right
I
mean
then
it
becomes.
Then
then
it
becomes
like
oh
yeah.
They
do
this
for
every
question,
but
but
if,
if
we
have
the
feeling,
let's
say,
there's
I,
don't
know
one
good
question
per
week
and
we
say
like
hey
this
question,
we
pick
it.
A
We
reach
out
to
the
person
and
say
like
hey:
can
you
put
the
same
question
over
to
a
stack
Overflow
or
something
like
that?
I
mean
I
said
at
the
end.
It's
what
you
said.
It's
about
momentum
right,
it's
about
giving
the
feeling
to
people
like
oh
I,
should
go
to
stack
Overflow
and
then
find
my
answers.
There
eventually
yeah
I.
D
Think
the
the
main
thing
I'm
looking
for
is
like,
let's
I,
don't
want
to
create
extra
work
for
people,
but
it
is
just
like
in
that
moment,
like
a
thing
we
are
actively
doing
right
now
is
like
typing
out
a
paragraph
per
question
in
slack
like
that
is
a
a
service
we,
you
know,
community
members
provide
for
each
other,
and
it's
just
like
it's
helpful,
but
it
doesn't.
D
It
means
we're
we're
having
to
answer
the
same
question
over
and
over
again,
probably
in
different
places,
because
no
one's,
no
one
except
the
person
asking
it,
finds
the
answer
when
we,
when
we
put
it
in
slack.
So
so
it's
like
what
what
is
like.
Could
we
somehow
get
some
muscle
memory
for
the
community,
so
it's
like
when
you
find
yourself
doing
it
here
or
like
at
work
or
just
anywhere?
D
If
you
find
yourself
like
pounding
out
like
a
three
to
four
sentence,
answer
just
open,
Telemetry
question:
where
you
know
it's
either
like
this
should
be
in
the
docs
like
this
is
like
literally
docs
that
are
missing,
which
is
my
next
question
for
this
group:
how?
How
should
we,
what
muscle
memory
should
we
get
the
community
to
form
for
like
when
that's
the
case,
but
then
there's
probably
like
a
lot
of
times
where
it
really
is
more
just
like
we
don't
want
the
docs
to
be
like
literally
every
question.
D
D
I
would
like
to
know
how
what
is
the
commsig
think
about
is.
The
best
thing
people
can
do
is
if
they
notice
there's
something
missing
from
the
docs
like
update
it.
But
do
we
want
to
encourage
like
an
even
lower
bar
for
like
if
people
notice
something
was
missing
or
they
noticed
they
like
pounded
out
an
answer
like
if
it
increased
feedback
on
the
docs?
D
F
I
I
think
that'd
be
tremendously
helpful,
even
something
that's
just
like
directionally.
Accurate
is
enough
for
one
of
us
to
sort
of
turn
it
into
content
and
yeah.
You
can
always
iterate
from
there
and
I.
Think
I.
F
I've,
what
I've
found
the
problem
to
be
so
far
is
there's
just
these
holes
in
the
docks
where
it's
not
that
like.
Oh,
this
is
documented
wrong.
F
It's
like
there's,
literally
not
even
documentation
for
this,
and
so
sometimes
they're
like
I,
don't
even
know
where
to
begin
with
how
like
this
should
be
documented
I'm,
just
gonna
like
post
this,
for
you
right
now
to
help
you
out
and
move
on
with
my
day,
but
when
there's
like
actually
something
there
like
I,
think
a
good
example
of
this
today
is
we
have
this
like
really
old,
pretty
out
of
date,
metrics
conceptual
page,
and
somebody
just
did
a
drive-by
issue
of
like
there's
instruments
missing
on
this
thing
and
they
followed
up
with
the
PR.
F
A
Okay,
yeah,
so
just
just
from
my
understanding,
this
would
mean.
Let's
say
someone
asked
a
question
on
open,
Telemetry,
slack
Channel
somewhere
from
the
community
answers,
and
all
we
would
get
is
an
issue
just
to
be
to
the
extreme
just
appointed.
To
that
conversation
and
saying,
like
hey,
this
question
was
asked
it's
missing
in
the
dark
series.
What
we
talked
about,
yeah
I.
A
D
Yeah
yeah,
obviously
the
ideal.
If
someone
makes
a
PR
that,
just
like
you
know,
fills
in
the
docs
but
in
the
I
think
for
me,
the
experiment
is,
if
we
just
like,
if
we
try
lowering
the
bar
to
providing
us
feedback,
another
Point
people
have
some
people
brought
up
is
like
not.
D
Everyone
actually
has
a
GitHub
account
who
might
have
feedback
I
feel
like
it's,
maybe
a
little
bit
less
true
for
like
this
kind
of
feedback,
it's
more
like
I
tried
to
stand
my
organization
up
on
open
to
limit
on
like
a
manager
of
some
kind
or
I
work
at
a
bank,
but
at
any
rate
not
just
like
for
our
slack.
D
If
we
do
want
this
kind
of
thing,
if
it
seems
like
it's
working,
I
would
start
encouraging
the
the
other
group
of
people
we
don't
get
active
engagement
from
right
now
are
like
sales,
engineers
and
customers
success.
People
within
a
lot
of
the
vendors
and
one
big
reason
why
we
don't
get
that
besides
they're
like
not
in
our
community
so
like.
D
Why
would
they
is
people
who
work
in
those
departments
tend
to
be
very
jaded
about
like
spending
effort
providing
feedback,
because
in
most
organizations,
their
experience
with,
like
bothering
to
like,
provide
feedback
on
things
that
aren't
working
or
things
like
the
customers?
They're
working
with
you
know
are
struggling
with,
is
like
it
just
it
just
goes
into
a
garbage,
can
somewhere
and
never
really
gets
actually
upon.
D
Exactly
this
is
like
I,
don't
think
I've
ever
worked
at
a
company
where
this
feedback
mechanism
wasn't
like
more
broken
than
it
should
be.
It's
just
kind
of
like
a
reality,
but
but
it's
it's.
The
other
group
that
I
want
us
I
feel
like
that's
the
that
is
like
a
huge
source
of
information
about
how
we
can
improve
open
Telemetry,
and
so,
if
it's
just
like,
if
if
we
can
really
lower
the
bar
for
feedback
and
and
I'll
at
the
same
time,
show
it's
like.
D
If
you
do
this
like
like,
we
will
actually
like
pick
these
issues
up
and
like
like
fix
things
to
the
degree
that
we
are
able
to.
We
might
be
able
to
encourage
those
organizations
to
give
us
more
more
interaction,
Patrice.
C
E
C
We're
open
and
and
and
people
are
responding,
so
maybe
that
message
just
needs
to
get
out,
I
and
and
yeah.
So.
D
Yeah
I
meant
more
less
like
end
users
and
people
who
are
part
of
the
community
and
more.
These
people
who
work
in
sales,
engineering
and
customer
success
within
within
vendors
are
are
there
are
people
who
are
like
actually
Fielding
answers
and
trying
to
help
people
with
open,
Telemetry
stuff
and
like
talking
to
people
who
are
either
really
enjoying
it
or
like
struggling
with
it?
Who
are
like
they
have
like
a
lot
of
active
engagement
but
are
like
also
like
super
not
like
in
our
community.
D
If
we
could
somehow
show
to
those
people
that,
like
their
life,
their
lives
got
better
if
they
started
working
on
the
project
and
providing
us
feedback
around
like
better
docs
and
things
like
if
we
could
show
that,
like
they
got
a
good
response
from
doing
that
like
like
you
know
over
time.
The
number
of
questions
they
have
to
answer
goes
down
because
they're
helping
us
improve
the
docs
anyways,
that's
maybe
we
don't
need
to
lower
the
bar.
D
Maybe
that's
the
wrong
way
of
putting
it,
but
it's
just
a
goal
of
mine
to
figure
out
how
to
actively
engage
those
people
and
show
it
I.
A
Mean
I
I
can
yeah
I
I
can
speak
to
that
I
mean
I
have
been
before
I
joined
product
here
I've
been
a
pre-sales
engineer
and
did
exactly
that
right,
I
mean
that's.
How
I
came
to
open,
telemetry
I.
Think
there's
that's
totally
true.
First
of
all,
I
mean
that's
how
pre-sales
Works
rather
I
mean
they
have
their
engagement
and
then
suddenly
they
have
another
one
and
they
just
yeah
move
on
and
and
that
question
is
not
important
anymore.
A
A
I
think
that
the
challenge
they
are
maybe
facing
is
that
their
questions
might
be
sometimes
extremely
vendor
specific
and
some
of
the
six
might
react
a
little
bit
allergic
on
that
and
say,
like
oh
you're,
asking
a
question
for
app
Dynamics
light
snap
honeycomb
at
least
go
to
them
and
ask
that
question
there.
Maybe
it
is
an
open,
Telemetry
question
right.
Maybe
it
could
be
valuable
for
everybody,
but
let's
say
they
could
make
a
bad
experience
and
then
just
never
come
back
again.
A
So
that
that's,
let's
say
what
makes
it
maybe
difficult
so
making
it
without
coming
from
them
for
them
is
Maybe
yeah
the
the
bar
is
slow,
but
but
maybe
let's
say
give
them
the
feeling
like,
even
if
they
come
with
something.
That's
maybe
vendor
specific.
It's
still
a
valid
question,
I
think
if
we
can
solve
that
I
think
that
could
be
extremely
helpful.
I
have
a
few
specific
situations
where
this
happens,
so
I
can
even
see
if
I
can
find
them
again
and
share
them.
F
This
this
is
something
I've
struggled
a
little
bit.
I
mean
like
this
is
how
I
actually
came
to
start
contributing
to
this
page,
because
my
sales
Engineers
were
like
there's
no
docs
and
I'm
like
well.
What
does
that
mean
and
they're
like?
What
do
you
think
it
means,
but
then
I
I
got
one
of
them
to
like
he
did
like
a
spreadsheet
of
like
these
are
all
the
topics
that
I've
like
fought
with
before
and
like
had
to
get
a
customer
involved
with
and
like
as
far
as
I
can
tell.
F
F
Think,
like
perfect
feedback,
but
like
even
that
is
really
really
hard
to
get
just
because
of
exactly
what
you
said
like
they
have
this
engagement
and
then
they
move
on
to
the
next
one
and
like
the
problem
from
the
previous
engagement,
is
completely
different
from
what
they're
they're
dealing
with
in
the
next
one
right.
I,
don't
know
how
to
crack
that
that
that
nut,
but
I
really
want
to
because
they
they
are
dealing
with
the
most
pressing
problems.
That,
like
are
relevant,
I,
mean
there's
a
lot
of
pressing
problems.
F
People
have
but
like
if
you
are
willing
to
pay
money
to
generate
open
Telemetry
data
and
send
it
somewhere,
you're
very
motivated
to
solve
problems
and
like
if
you
can't
solve
those
problems
in
a
self-serve
way,
with
the
docs
I
sort
of
feel,
like
that's
kind
of
a
failure
and
like
I,
want
the
people
who
are
dealing
with
that
tension
today.
To
even
just
do
a
drive-by
like,
even
if
they
just
say,
hey
the
tracing
Concepts
page,
doesn't
explain
tracing
well
enough.
F
I
can't
really
tell
you
what
needs
to
be
done,
but
I'm
just
giving
you
that
feedback
right.
Now
that
is
enough
for
me
to
actually
go
in
and
start
like
talking
to
other
people
about
how
it
could
be
improved
and
so
like
from
my
perspective,
I,
don't
want
to
turn
away
someone
who
gives
sort
of
like
pseudo
unhelpful
feedback,
because
it
is
helpful
as
like
a
signal
that
something
is
wrong
and
I
can
dig
into
how
it
can
be
improved.
F
I
think
there
is
a
tendency
in
a
lot
of
Open
Source
communities
to
be
like
well
I
want
you
to
tell
me
specifically
what
you
found
wrong
with
it
before
I.
Take
any
action
on
this
and
I.
Don't
think
that'll
get
us
anywhere
awesome.
D
D
It's
open
source
and
I
just
want
to
make
sure
like.
If,
if
we
think
it's
helpful
to
encourage
people
to
be
like
look,
you
don't
have
to
like
spend
a
bunch
of
time
on
this
feedback.
It
can
even
be
just
kind
of
vague,
like
what
you
said.
Like
you
know,
I
was
trying
to
get
people
started
with
tracing
using
the
docs
and,
like
it
wasn't
working,
you
know
we
were
doing
X
or
Y
or
like
just
like
I
pasted
this
answer
into
somebody
somewhere.
D
It
should
probably
be
in
the
docks
if,
if
we
think
like
that's
like
just
helpful
to
have
that
signal,
you
know,
I
could
tell
people
to
that.
Like
GitHub.
Issues
on
the
website
are
like
a
place
to
to
put
that
for
the
time
being
and
see,
if
see,
if
it's
possible
to
then
do
like
an
organizing
drive
to
start
getting
like
these
companies
to
to
do
it.
A
Awesome
yeah,
I
I
think
we
should
explore
this
a
little
bit
more
again
and
again,
but
I
think
that
that's
a
good
outcome.
Philip
do
you
want
to
go
next
with
the?
What
was
the
two
topics
you
had
in
mind.
F
Yeah
yeah
they're,
actually
somewhat
related,
so
the
first,
as
as
you
saw
with
the
pull
request,
I'm
already
doing
a
push
for
sampling,
docs
I'm
planning
on
spending
more
a
little
more
time
on
the
conceptual
stuff.
I
think
it
needs
a
bit
more
elaboration,
even
with
the
stuff
that
I
put
in
there
and
then
every
language
page
I
want
to
have
like
a
sampling
doc.
That
just
said,
like
the
main
one
with
what
you
can
do
today
with
sampling
and
head
sampling
with
your
sdks,
is
the
trace.
F
Id
ratio
based
and
just
a
like,
just
like
a
hey
I
need.
X
percent
of
my
traces
just
show
them
how
to
do
it
needs
language
I
feel
like
that's.
That's
good
enough.
F
I've
seen
that
come
up
speaking
of
sales
engagements
I've
seen
that
come
up
constantly
through
my
my
work,
where
a
lot
of
people
aren't
even
like
aware
of
the
fact
that
head
sampling
is
an
option
for
them
and
like
when
they're
trying
to
like
size
up
a
deal
that
they're
trying
to
do
like
they
don't
want
super
duper
Precision
up
front,
but
they
want
at
least
like
some
idea
about
like
what
volume
of
data
they're
producing
and
how
that
can
help
translate
into
their
bill
and
they're
like
I.
F
Don't
even
know
how
to
do
this
and
it
really
sucks
and
I
hate
it
and
like
it,
comes
in
like
last
minute
in
like
sales
engagements,
and
it's
like
that
shouldn't
be
the
case
right,
like
yeah.
What
you
roll
out
eventually
is
probably
going
to
be
a
lot
more
sophisticated
than
just
taking
some
percentage
of
your
traces,
but,
like
you
can
get
yourself
in
the
right
order
of
magnitude
and
I
want
to
solve
that
problem
through
docs,
or
at
least
try
to.
A
A
Like
hey
I,
don't
know,
look
at
the
docs
and
they're
still
like
yeah:
there
are
no
dogs
so
and
then
you
maybe
Point
into
I,
don't
know
the
go
Community
because
they
have
something
in
their
in
their
repost,
so
yeah
I
think
it's
it's
awesome
and
then
I
think
there
there's
many
more
topics.
I
mean
I
did
something
similar
with
resource
detection.
Right
I
mean
we
have
the
same
discussion
going
on
with
like
hey.
We
want
to
correlate
infrastructure
with
with
service
data.
A
F
I
I
plan
on
also
doing
stuff
with
the
tail
sampling
processor,
but
I
I.
It's
it's
also
like
every
time
I've
been
working
with
a
customer
I've
had
to
like
lead
them
away
from
the
tail
sampling
processor
in
The
Collector
because,
like
they,
they
start
thinking
that,
like
oh,
it's
just
a
processor
I'll
put
in
my
collector.
F
But
then
they
get
hit
like
with
a
freight
train
that,
like
this
jumps,
their
operational
complexity
up
several
orders
of
magnitude,
and
now
they
have
like
basically
their
own
distributed
system
that
they're
running
that
handles
tail
sampling
and,
like
they
start
getting
into
like
the
complexity
of
rules
and
depending
on
how
large
your
systems
are,
those
get
huge
and,
like
you
find
that
there's
like
you're
you're,
still
missing
certain
things
that
you
want
to
be
able
to
do
like
you
know,
there's
no
ability
to
dynamically
sample
based
off
of
stuff
and
then
you're
like
okay.
F
Well,
I
went
down
this
Rabbit
Hole
now
I
need
to
like
reach
out
to
my
vendor
for
the
vendor
specific
solution
and
I
just
wasted
a
bunch
of
time.
Doing
that
but,
like
I,
don't
think
it's
a
bad
thing
to
say
that
it
exists.
I,
I,
just
I,
don't
know
what
the
right
bar
is
for
like
guidance
for
use.
D
So
I
don't
know
if
this
is
helpful,
but
just
describing
sampling
to
people
is
hell
and
I've,
never
written
docs.
This
way,
but
I
have
given
like
toxin.
D
Like
you
know,
whiteboard
discussions
on
this
subject
with
BHS
from
lightstep
and
a
thing
he
has
done,
that
I've
seen
cause
the
penny
to
drop
with
people
is
to
to
go
at
the
sampling
problem
from
the
other
from
the
direction
of
features,
in
other
words,
there's
different
kinds
of
things
you
can
do
with
observability
data
right,
like
you
can
be
doing
like
root
cause
analysis
right,
you
can
be
doing
live
querying
of
you
know
your
data,
you
can
be
doing
average
latency
analysis
over
time
and
different
sampling
strategies
take
away
different
features,
and
so
the
way
to
think
about
what
sampling
can
I
do
can
best
be
thought
about
like
well.
D
What
am
I
doing
with
this
data
right,
like
am
I.
If,
if
I'm
trying
to
do
auditing,
then
yes,
you
need
to
save
a
hundred
percent
of
it
right.
If
auditing
is
like
a
required
feature,
for
example,
if
you're
only
trying
to
do
like
performance
analysis
of
like
performance
improvements-
and
you
don't
care
about
outliers
right,
then
you
can
like
do
really
heavy
ped
sampling
upfront
sampling
right
because
averages
are
all
you
need
you
don't
care
about
the
outliers.
D
So
this
is
just
like
big
feedback,
because
it's
hard
to
like
give
this
kind
of
it's
a
little
bit
difficult
to
to
give
this
approach
to
people,
because
it
what
features
you
have
access
to
depend
on
the
back
end
you're
using,
but
for
what
it's
worth.
I've
found
that
that
helps
make
sampling
make
sense
to
people,
because.
D
D
So
that's
that's,
like
my
hopefully
helpful.
Two
cents
is
like
just
starting
from
the
perspective
of
like
yeah.
If
you're
trying
to
do
this,
then
these
forms
of
sampling
are
available
to
you.
If
you're.
F
F
They
like
tell
you
that
sampling
is
a
thing
and,
like
you
know,
represent
representative
sets
of
your
data
is
often
good
enough,
but
like
that,
that's
not
elaborate
enough.
I,
don't
think
it's
very.
D
Hand
wavy
that's
the
problem,
it's
like
unless
you're
specific
about
the
classes
of
features
you're
talking
about
it's
hard
to
be
specific,
about
which
sampling
strategies
to
recommend
to
people
anyways.
That's
that's
the
thing
I've
noticed
from
trying
to
explain
this
to
people
if
you
go
from
that
direction
of
being
like
these
are
the
different
classes
of
features
or
use
cases.
D
If
you
want
to
control
costs
and
you're
doing
this,
you
can
use
these
following
sampling
approaches
if
you're
trying
to
do
this
other,
you
know
like
root,
cause
analysis
tool
right
like
where
you're
really
interested
in
like
outliers
and
all
this
other
stuff.
Then
you
have
to
use
this
other
tool.
You
know,
if
you're
doing
statistical
stuff,
you
need
to
have
statistical
lead.
You
know.
Probably
you
have
to
have
sampling.
D
F
F
At
the
very
least,
around
sampling
by
kubecon
I
mean
like
the
past
two
coupons
I
was
at
like
seems
like
a
top
question
that
people
have
certainly
I've
noticed
even
more
of
a
shift
in
my
own
work,
where,
like
people
coming
in
through
the
sales
process,
are
talking
about
sampling
and
cost
control
measures
up
front,
as
opposed
to
much
later
in
the
cycle,
like
you
know,
I
mean
you've
seen
the
size
of
the
discussion
groups
on
the
on
the
dedicated
days
that
we
do
and
there's
always
a
topic
about
sampling,
and
it's
got
at
least
half
the
people
who
showed
up
at
the
event
talking
about
it.
F
I
feel
like
it's
a
must-have
for
us
to
do
and
I'm
willing
to
do
a
lot
of
that
work,
but
are
there
other
topics
where
we're
like
yeah?
We?
We
should
have
this
documented
by
a
cubecon.
B
I
I
think
enough
will
be
to
have
like
probably
not
like
a
grand
announcement
but
enough
to
have
a
lot
a
lot
of
focus
on
it
and
Philip
I
I
generally
agree
with
the
sampling
conversations.
Like
speaking.
For
you
know,
person
works
at
Splunk,
it's
less.
We
do
100
sampling.
So
for
us
it's
a
lot
less
relevant,
but
in
any
of
the
open
source
events
I
go
to
I
completely
agree.
It's
a
major
major
topic.
E
B
But
yeah
Ted
I,
expect
logs
will
be,
will
be
big,
like
that.
This
might
be
one
where
people
outside
of
the
core
of
the
community
aren't
desperate
for
it.
Just
because
we've
been
plugging
away
at
working
on
it
and
I
think
we're
a
little
further
ahead
than
people
may
have
anticipated
yeah,
which
is
good.
That's
a
good
problem
to
have
right.
D
Yeah,
it's
it's
I
have
found
a
hundred
percent
of
my
conversations
as
people
are
like
we're.
Never
gonna
get
logs
I'm
like.
C
B
And
they're,
like
no
I,
didn't
actually
it's
like
the
anti-metrics,
where
I
think
for
metrics
in
like
2020
like
when
I,
even
when
it's
way
back
at
Google
like
saw
my
boss
is
like,
as
for
a
few
months,
it's
almost
done.
It's
super
close
and
then
yeah
in
2022.
It
finally
goes
GA
I.
Think
logs
is
the
opposite.
Where
most
people
are
thinking,
it's
a
long
way
off
and
like
no
it
it's.
It's
pretty
good
shape.
Actually.
B
C
F
That
so
that
so
I
guess
that
to
that
end,
I've
sort
of
two
questions-
I
I,
know
that,
like
language
level,
support
is,
is
differing,
as
is
the
case
with
everything.
F
E
B
Would
say
both
like
if
you'd
asked
me
a
month
or
two
ago,
I
would
have
said
The
Collector
part
because
that's
fairly
robust,
like
I
know
it's
not
marked
as
stable,
but
like
it
works
and
people
are
using
it.
The
we
got
an
update
at
the
maintainers
call
about
the
SDK
connections
and
all
that
this
morning,
it's
in
pretty
good
shape,
at
least
on
Java
and
I.
Think
a
lot
of
other
languages
are,
in
surprisingly
good
shape.
There.
B
So
I
think
we'll
want
to
talk
about
both
logging
log
processing,
from
like
tailing
files
on
The
Collector
and
with
the
direct
structured
logs
being
streamed
into
the
collector
via
OTL
page
from
our
sdks.
Okay,.
D
Yeah
and
right
now
the
the
logs
page
I
mean
I,
don't
know
if
end
users
even
dig
down
to
the
point
where
they're.
Looking
at
these
signal
pages,
I.
E
D
Yeah
yeah
I
think,
if
like
if
this
page
was
just
flushed
out
a
bit
to
explain
like
the
different
pieces
logging
pieces
available
in
open
telemetry
that
might
help
like
just
point
people.
It's
like
The
Collector.
You
do
a
bunch
of
log
processing.
Here's
the
collector
log,
docs
stuff.
You
know
sdks
Can
consume
logs
from
existing.
You
know
log
sources
today
in
all
these
languages
like
go,
look
there
that
that
would
probably
be
helpful.
A
F
Out
yeah
yeah
like
this,
like
the
part
of
the
intended
motion
of
like
these,
are
directly
like
these
are
attached
to
your
traces.
Basically,
oh
I
mean
through
strong
correlation,
but
like
that's
not
very
clear
on
this
single
paragraph,
yeah.
D
Yeah
yeah,
if
we
just
listed
out
the
different
things
that
open
Telemetry
has
you
know
that
would
probably
just
help
a
lot.
E
A
B
As
for
like
what
we
have
at
the
conference,
it's
the
same
as
usual,
so
we
have
like
two
or
four
hour
project
meeting
and
various
other
forms
for
for
announcing
things
and,
of
course,
there's
the
observability
day
that
Austin's
organizing
I
forget
so
there's
lots
of
forums
for
this,
but
I
think,
like
blog
posts
that
are
in
there
is
a
is
a
good,
an
area
where
we're
often
we
could
often
do
better
and
it'd
be
a
good
thing
to
start
teeing
up
right
now.
Oh.
E
A
But
because
if
we
start,
let's
say
anticipating
it
today,
it
still
takes
a
week
or
two.
Then
we
have
it
so
I
would
say:
let's
say
one
or
two
weeks
before
kubecon
it
should
be
ready.
Do
we
have
a
good
overview
of
of
what's
happening
on
kubecono
around
open
telemetry
like
I.
A
I
I'm
doing
wrong
together
with
travel
and
a
few
other
people
around
the
hotel
operator.
I
think
there's
a
few
other
ones.
So
should
we
yeah
I
think
should
we
have
I
think
that
was
the
question
I
took?
We
have
like
a
blog
post
around
kubecon
and
it's
like
hey
there's
an
observability
days
and
yes.
B
B
There
was
one
for
unplugged,
but
though
it
was
mostly
sort
of
mechanical
like
yeah,
yeah,
organizing
vendors
and
stuff
yeah.
E
A
D
Yeah,
are
we
Morgan,
do
you
know?
Are
we
having
a
project
meeting
again
we
are,
or
is
that
so
we
are
having
one
okay,
do
you
have
to
be
registered
for.
B
I
got
the
confirmation
forwarded
to
Austin
when
I
got
it,
but
let
me
just
pull
it
up.
Project
meeting
is
on
Tuesday
April
18th
at
one
o'clock
to
three
o'clock
in
the
conference
center.
D
A
Awesome
yeah,
let
me
maybe
fill
up.
Maybe
we
can
also
check
with
Austin
if
he
already
maybe
have
had
something
in
mind,
also
for
observability
day
or
not,
and
then
we
can
write
something
up
or
I,
don't
know.
D
E
D
B
C
D
E
B
F
We
can
answer
that.
Let
me
that
that's
more.
F
A
Down
yeah
by
the
Run
query,
and
then
oh,
no
okay:
where
is
it
there
H
here,
okay
and
then.
B
D
A
By
the
way,
there's
something
here
but
yeah
exactly
and
the
tail
sampling
one
is
high
up
and
what
not
so.
C
D
D
I'm,
just
realizing
I'm
gonna
miss
our
our
open
Telemetry
session
on
Friday,
looks
like
like
the
like
I
guess.
It's
like
whatever
our
project,
one
that
we
all
get
signed
up
for
the
maintainer
track.
Oh.
B
B
I
felt
bad
in
Detroit,
like
I,
didn't
fly
out,
I
kind
of
wish
I
had,
but
like
the
poor,
dynatrace
guys
had
a
talk
that
was
accepted.
It
was
like
Friday
at
4
pm
and,
like
I,
think
you
and
I
were
walking
around
because,
like
we
were
in
the
lounge
together
at
the
airport,
yeah
and
like
I,
remember
walking
up
to
it
like
white
was
wrapping
up
and
like
basically
no
one
had
attended,
because
it's
4
P.M
on
a
Friday
yeah,
brutal.
B
Well,
my
hotel
ends
on
April
21st,
which
is
a
Friday
yep
yeah.
Okay,
I
will
get
that
sorted,
I,
wonder
if
we
just
want
to
cancel
it
I
doubt.
B
D
B
F
Cool
well,
we
are
at
time
actually
seven
minutes
over
time,
but
I
think
this
is
a
good
set
of
like
I.
Have
some
pretty
concrete
next
steps,
I
can
take,
dig
it
I'm
gonna,
head
out.