►
From YouTube: 2022-09-08 meeting
Description
cncf-opentelemetry meeting-2's Personal Meeting Room
B
A
C
I
always
cringe
a
little
when
people
talk
about
bus
factor,
because
when
I
worked
at
disney
one
of
my
project
managers
project
lead
for
the
entire
project,
stepped
out
of
the
building
one
day
and
was
immediately
hit
by
a
bus,
and
he
spent
two
months
out
of
work.
And
fortunately
he
you
know
was
fine
in
the
end.
But
but
it
was
always
like.
A
C
The
funny
thing
that
he
quoted
to
us
later,
once
he
was
back
to
work,
was
that
the
police
report
included
a
delicious
line.
That
said,
something
like
the
victim's
head
landed
25
feet
from
the
point
of
impact.
C
B
Yeah,
I
there
wasn't
much
time
between
sharing
it
and
today.
So
that's
totally
fine.
B
B
Think,
through
what
sort
of
use
cases
or
capabilities
like
an
open,
telemetry
native
sampling
design
ought
to
include.
I
thought
at
prudence
to
like
understand
what
capabilities
people
might
be
leveraging
in
other
products
today.
So
that's
kind
of
like
these
are
my
notes
along
those
lines.
So
it's
I
guess
I
would
say,
maybe
more
informational
than
anything
so
yeah.
B
I
guess
my
suggestion
would
be
to
like
try
to
sort
of
internalize
some
of
this
information
and
like
ask
clarifying
questions
and
like
and
and
then
I
do
have
some
ideas
on
like
what
a
sort
of
common
denominator
may
look
like
between
these.
These
systems
that
I
surveyed,
but
what
I
was
realizing
in
the
last
ten
minutes,
was
that
it
would
be
difficult
to
present
those
sort
of
persuasively
or
like,
like
you
need
a
certain
background.
B
I
think
in
like
what
what
is
used
today
to
like
contextualize
sort
of
what
I
was
going
to
suggest.
Potentially,
so
that's
what
this
is.
C
Yeah,
I
think
one
maybe
caution
if
I
would
on
the
survey
yeah,
is
that
all
of
these
are
moving
targets.
Sure
everybody's
you
know,
updating
all
this
stuff
and
and
the
the
capabilities
and
restrictions
will
change,
and
I
I
like
I
liked
your
ideal
state.
C
You
know
at
the
bottom
of
that
document.
That's
the
part
I'd
like
to
maybe
elaborate
on
sure,
flesh
out
and
see
you
know
like
and
and
to
some
extent
that's.
I
I
sort
of
mentioned
that
in
my
comments
in
the
other
document
like
I,
I
don't
want
us
to
forget
two
major
factors:
one
is
people
are
trying
to
understand
their
stuff
and
sampling
is
kind
of
necessary
once
you
reach
a
certain
scale,
so
we
have
to
figure
out
how
to
help
people
like-
and
I
guess
the
other.
C
The
other
thing
I
want
to
remember
is
that
most
people
aren't
even
able
never
mind
interested
in
thinking
about
this
stuff
too
deeply
mathematically.
So
I
so
I
think
we
should
bias
toward
finding
ways
to
make
it
hard
for
people
to
make
dumb
mistakes
that
cause
them
to.
You
know
sure
to
misinterpret
their
own
data.
C
Not
you
know
so,
like
that's,
not
an
argument
that
don't
do
things
with
mathematical
rigor.
It's
an
argument
that,
let's
make
sure
that
we
don't
forget
in
the
pursuit
of
mathematical
rigor
sure
that
there
are
normal
humans
who
don't
know
even
how
to
think
about
this
stuff.
C
B
Yeah
I'm
interested
also
in
like
hearing
yours
or
others.
Folks,
like
you,
mentioned
elaboration
on
the
ideal
state,
and
it's
like
you
know,
those
few
bullet
points
inspired
any
like
thoughts
and
reactions
for
many
of
you,
I'd
be
really
interested
in
them.
C
One
thing
I
did
notice
is
you
really
didn't
define?
What
supports
estimation
means
to
you
document.
B
That
particular
point,
I
think
I
attempted
to
define
it
if
you
near
the
tour
there's
a
section,
titled
other
dimensions
and
it's
it's
in
there.
Oh.
A
C
B
Yeah
and
like
you'll
note
that
yuri
one
of
the
maintainers
on
the
jager
project,
like
already
made
a
big
correction,
comment
yeah.
So
I
didn't.
I
haven't
yet
updated
the
mainline
text,
but.
B
B
I
remember
a
previous
conversation
where
one
of
you,
I
think,
had
mentioned
that
most
of
your,
the
bulk
of
your
experience
had
not
been
on
like
actually
building
software
systems
with
like
putting
sampling
into
into
practice
so
to
speak,
and
I
wanted
to
like
I'm
very
interested
if
either
of
you
do
have
thoughts,
but
I
also
just
want
to
like
make
sure
I
understand
like
that.
I'm
not
like,
prompting
you
guys
and
like
making.
You
feel
bad
for
not
having
much
to
say.
A
A
B
B
Can't
one
thing
I
wanted
to
run
by
you:
since
you
mentioned
you
have
read
it
a
little
bit
was
there
was
a
notion
in
my
like
ideal
state
that
as
a
user,
it
would
be
very
nice
if
some
of
this
configuration
were
used
to
the
phrase
set
it
and
forget
it
and
you've
worked
with
like
customers
a
lot
more
than
I
have
in
this
domain,
and
I
was
wondering
if
that
was
like
a
sentiment.
You've
heard
it's
something
that
I've
it's
something
that
I've
experienced
where
it's
like.
C
So
perfect
example,
actually,
so
we
did
in
fact
have
a
sort
of
what
we
call
a
mini
listening
tour
today,
yet
this
week
to
talk
about
refinery
and
and
talk
to
all
of
our
the
the
technical
sales
people
about
what
they're
hearing,
what
they're
seeing
and
that
kind
of
thing,
and
in
particular
we
had
you
know
like
it,
was
kind
of
amusing
to
me
that
you
specifically
called
out
the
problem
with
the
total
throughput
sampler
and
how
it's
on
a
per
node
basis,
so
that
when
you
scale
things
horizontally,
the
rates
change
and
we
had
a
support-
request
come
through
this
week
from
somebody
who
didn't
know
that
and
they
changed
their
number
of
refinery
instances
and
their
traffic
went
up
and
they
were
upset
and-
and
it
was
reported,
you
know
as
a
bug
and-
and
I
was
kind
of
like
well-
no,
it's
not
a
feature
yet,
but
but
but
the
problem
of
shared
state
like
one
of
the
things
that
is
most
interesting
to
me
with
this
whole
thing
is
to
try
and
figure
out
how
to
make
this
stuff
scale
more
sensibly.
C
Otel
collector
instead
of
refinery
is
that
hotel,
collector
doesn't
scale
horizontally
beyond
like
it
well,
because
you
need
to
shard
the
data
going
to
it
in
advance,
so
the
the
I
mean
you
could
do
a
kind
of
a
two
layer,
collector
thing
and
so
we're
gonna
we're
gonna,
investigate
that
and
work
on
that,
some
more
but
right
now,
refinery
is
closer
to
fire
and
forget
than
collector
is,
and
so
yes,
I
think
I
think
the
answer
is
customers
would
prefer
not
to
have
to
think
about
this
too
hard
and
especially
not
to
have
to
think
about
it
every
week
or
so.
C
That's
that's
not
fun
for
them.
So
you
know
I
had
a
customer
call
this
week
same
thing.
Customer
is
like
okay,
we
need
refinery.
C
Can
you
help
me
to
get
the
sort
of
easiest
possible
answer
to
get
me
started
and
then
we'll
go
from
there
and
and
it's
the
same
sort
of
thing
it's
like
they
have
many
other
problems
to
worry
about,
and
they
want
actionable
insights,
and
so
it's
going
to
be
a
big
factor
in
what
I
think
about
just
just
for
transparency.
C
I
we
are
spinning
off
a
team
at
honeycomb,
that's
going
to
be
focused
on
basically
collector
and
refinery,
and
I'm
not
going
to
be
the
lead
on
that
team.
So
so
it's
my
roadmap
for
the
next
six
months
is
figuring
out.
You
know
figuring
out
how
what
end
of
life
of
refinery
looks
like.
C
So
we
can
move
people
over
to
collector,
but
also
putting
enough
features
in
refinery
as
almost
a
test
bed
to
to
to
have
it
scale
better
and
smarter
and
stuff
like
that,
because
as
it
is
right
now,
it
doesn't
scale
well,
so
you
have
to
you,
have
to
sort
of
manually
tune
it
every
time
you
scale
it.
So
we
tell
people
just
set
up
your
refinery
cluster,
don't
auto
scale.
It
leave
it
alone
because
it
changes
too
much
if
you
do
that.
Otherwise,.
B
Okay,
yeah,
I
mean
I'm
glad
to
have
validation
for
for
some
of
what
I
what
I
wrote
here.
It
sounds
like
other
customers
have
said
to
you.
Yeah.
B
All
right:
well,
we
don't
have
to
just
sit
here
and
yeah.
Like
I
said,
I'm.
I
totally
understand
that
I've
shared
this
very
quickly
before
this
meeting.
B
Yeah,
so
I
mean
in
terms
of
commenting
yeah.
This
is
like
a
gist,
so
I
guess
you
could
do
what
like
yuri
did
and
like
quote
things
below
it.
It's
like
you
can
either
have
it
nicely
formatted
in
here
or
you
could
have
it
like,
not
very
nice
in
google
docs.
I
guess
I
guess
it
could
still
be
nice
in
google
docs
if
you
don't
try
to
write
it
in
markdown,
but
my
thinking
was
like
I
bet.
B
Eventually,
some
of
this
content
will
be
adapted
in
some
form
into
the
otep
right
now.
It's
just
in
these
like
separate
little
notes
or
memos,
but
well
yeah,
one
of
the
struggles
and
like
one
of
the
things
I
would
love
input
from
this
group
eventually
is
like,
as
we
get
sort
of
a
more
comprehensive
mental
model
of
like
what
people
are
using
today
and
what
capabilities
they
need
and
stuff.
Like
part
of
the
challenge.
B
With
this
whole
sampler
configuration
line
of
investigation
was
that
yeah,
it's
like
as
peter
was
saying
there
are
like
many
directions.
One
can
go
and,
like
I
spent
a
lot
of
time,
perhaps
spent
a
lot
of
time
like
going
broad
in
like
research
on
on.
B
You
know
what
might
be
relevant
here
and
eventually
like
like.
I
don't
know
how
much
yet
should
should
go
in
like
an
hotep,
proper
versus
like
sites
other
things
or
whatever,
and
maybe
it
doesn't
matter
that
much.
But
but
that's
kind
of
why.
B
The
vehicle
for
like
me,
sharing
research,
is
just
these
like
little
short
memos
anyways.
If
no
one
has
other
comments
right
now
or
anything
they'd
like
to
talk
about
I'm
content
to
adjourn.
C
I'm
not
sure
if
it's
there
yet,
but
maybe
having
some
repository,
maybe
a
private
fork
of
the
otep
repository
in
which
we
can
start
writing
an
otep
that
is
intended
for
pushing
upstream
but
doesn't
show
up
as
a
pull
request
yet
might
be
an
easier
way
to
share
and
comment
because
then
it's
you
know,
you've
got
an
md
file
that
we
could
use
the
standard
github
tooling
to
to
do
yeah.