►
From YouTube: 2023-02-21 meeting
Description
cncf-opentelemetry@cncf.io's Personal Meeting Room
A
B
A
C
B
Sounds
like
it
I
guess,
we'll
see.
Who
else
is
available.
D
E
E
C
B
Cool
just
a
little
bit
about
me,
I'm
Evan
I'm
at
a
dynatrace
I'm
on
the
open,
Telemetry
team
at
dietrace
and
have
been
contributing
in
to
op
amp
in
kind
of
minor
way
so
far,
but
hoping
to
ramp
that
up
right
now,
I'm
working
on
helping
getting
an
extension
into
the
collectors
that
we
can
get
the
the
supervisor
model
kind
of
up
and
going
shortly
here,
I'm
expecting
to
also
start
a
production,
ready
supervisor
implementation
in
a
collector
trip.
A
D
E
Sounds
like
we're
giving
intros
so
I
will
give
an
intro
I'm
Andy
Keller
from
observe
IQ.
We
do
a
lot
of
work
on
it.
We
just
want
to
show
you
on
the
collector
and
also
my
team
is
responsible
for
biplane,
which
is
open
source
implementation
of
op-amp
agent
management
platform.
C
C
So
in
it's
specification
or
well,
this
document
I
left
was
a
comment
there
and
I'm
trying
to
figure
out
what
exactly
the
the
thinking
there
is
and
maybe
having
since
you're
implementing
already
you,
you
have
a
stronger
opinion
and
or
Andy
obviously,
and
that
is
I
see
or
the
way
I
understand.
It
is
run
as
apparently
being
optional
and
My
worry
is
thinking
back
of
the
situation
that
we
had
in
the
docker
file.
C
Where,
if
you
don't
Define
anything,
then
it's
run
as
root,
which
is
really
really
a
big,
an
issue
trying
to
understand
a
what
is
the
default
value
right?
C
Does
that
even
make
sense
if
it's
optional
and
yeah?
Maybe
there
is
an
opportunity
to
actually
require
that
that
would
be
my
preference
yeah
just
generally
interested
in
how
you
deal
with
that
from
an
implementation
perspective.
So
far,
so.
B
B
So
when
you're
looking
at
at
least
in
I,
think
most
unix-like
operating
systems,
I
can't
speak
for
Windows
and
my
primary
expertise
is
in
Linux.
But
from
how
I
understand
it
works
is
usually
you
have
to
substitute
the
user
as
root.
B
There's
a
there's,
a
system
call
that
you
make
to
do
that
and
we
should
probably
not
try
to
encourage
running
the
supervisor
as
root
in
my
opinion.
So
normally
what
we
would
have
is
we
would
have
the
user
use
a
mechanism
to
you
know
like
pseudo
or
something
to
actually
do
this
authentication
and
make
sure
that
it's
done
per
their
their
security
requirements,
so
that
was
kind
of
my
primary
concern.
B
Although
I
think
that
you
make
a
good
point
in
that,
if
people
are
going
to
be
running
the
supervisor
as
root,
we
should
do
something
to
push
them
to
at
least
Run
The
Collector.
As
a
non-rew
user,
or
as
do
what
we
can
to
encourage
users
to
minimize
the
permissions
for
each
exactly.
C
E
That's
a
good
question:
I
I
don't
deal
with
my
return.
To
be
honest,
it's
when
she
seems
surprising
but
I'm
on
the
other
side
of
the
op-amp
exchange,
so
I
I
know
that
we've
historically
I
believe
had
to
run
as
root
in
some
situations,
but
of
trying
to
avoid
it
and
in
most
cases
let
me
I'm
curious
what
our
defaults
are
right
now
give
me
a
second
I'll
see
if
I
can
get
an
answer
from
our
our
team.
Okay,.
C
So
I
I
think
I
I
understand
where
you're
coming
from
and
I
I'm
trying
to
figure
out,
given
that
it's
optional,
like
by
default,
the
supervisor
would
run
as
the
user,
like
I
I'm,
still
struggling
with
these
semantics
So.
What
at?
What?
In
what
user,
ID
or
what
user
does
the
supervisor
by
default
run?.
B
I,
don't
think,
there's
a
I
mean
I
think
that
that
would
probably
be
whatever
it's
been
set
to
run
by
the
I
guess
the
system
operator
I
mean
if
we
had
a
if
there
was
a
Docker
image
that
was
published
I.
Would
you
know
I
think
it's
pretty
standard
practice
that
you,
you
have
a
non-root
user
for
that
now,
of
course,
people
can
deploy
the
supervisor
as
a
system
demon,
so
they
can
have
that
running
his
route.
They
could
build
in
their
own
Docker
file.
B
Have
that
running
its
root,
so
I
think
that
there
are
cases
where
that
you
know
could
happen.
I
don't
know,
I,
don't
think
that.
Well,
there's
no
supervisor
implementation.
That
I'm
aware
of
out
in
the
wild
right
now.
I
know
that
there's
that
example,
one
in
the
op
amp
go
repo,
but
I.
Don't
think
that
there
are
any
I,
don't
think
anybody's
running
one
anywhere
I.
Think
one
to
your
point
about
running
is
Route
I.
B
Think
one
thing
we
could
potentially
do
is
check
if
the
supervisor
is
running
it,
root
and
kind
of
flag,
the
user
on
that.
Alternatively,
we
could
always
just
require
it
and
users.
Will
you
know
if
you
specify
we
could
rolled
that
that
warning
or
otherwise?
Just
so
that,
yes,
you
know,
this
is
if
you
do
run
as
root.
That
is
me
saying:
I
accept
the
the
consequences.
C
I
think
if
it's,
if
it's
required,
if
it's
a
required
field,
I
have
no
problem
at
all.
If,
if
someone
decides
to
put
root
in
there
or
it
could
be
precise,
the
uid,
the
actual
one,
but
the
problem
really
is
with
this-
you
know
it's
optional,
so
it's
kind
of,
as
you
said,
right
yeah.
C
If
it's
running
in
a
containerized
environment,
then
one
would
hope
that
you
know
someone
sets
the
user
in
in
the
docker
file,
which
you
know
I
believe
last
time,
I
checked
the
the
Registries
is
still
not
very
widely
use
that
pattern
it
should
be,
but
it
is
so
I
have
this
worry
that
if
we
don't
make
it
explicit
and
what
you
know,
this
security
related
things
concerning
I'd
rather
be
explicit
than
hope
for
the
best
and
plus
one
to
everything
else.
C
You
said
right,
flagging
that
or
validation
or
startup,
saying,
like
oh,
it
seems
you're
running
through
a
screwed,
fine
right,
but
having
That,
explicit
I
think
forces
you
as
an
operator
to
think
about
it.
What
do
I
want?
Someone
says
root.
Fine!
You
know.
If
you've
signed
your
waiver,
you've
sent
your
you
know:
I'm
fine,
with
that
foreign.
B
So
I
think
that
makes
sense.
I
think
the
one
question
I
would
have,
then,
is:
if
we're
going
to
include
this
as
a
a
required
supervisor
feature.
What
is
the
mechanism
that
we
give
to
users
for
doing
user
substitution?
Should
we
pre-package
wrappers
around
a
couple
of
utilities
that
do
this
I
know,
there's
pseudo
and
due
as
I'm,
not
sure
what
that
looks
like
on
Windows
or
maybe
on
some
of
the
other
Unix
systems
that
we
might
support.
I
would
have
to
double
check
what
the
the
support
Matrix
looks
like
there.
B
I
know
that
the
collector
definitely
supports
the
three
major
operating
systems.
I,
don't
know
what
support
guarantees
are
given
to
anything
else,
but
I
think
the
question
is
how
we
make
that
extensible
to
the
user,
but
also
not
not
overbearing
and
not
putting
too
much
into
the
supervisor
to
accomplish
that
right.
C
I
I
also
don't
know
about
windows
are
mostly
working
working
in
the
Linux
environment,
but
my
argument
would
be
as
long
as
that
field
exists.
We're
dealing
with
that
problem.
Anyways,
it's
just
a
matter
of
dealing
with
it
silently
or
loudly
right,
because
if
it's
there,
you
would
still
have
the
same
problem.
If
it's
optional
someone
sets
it,
then
you
need
to
deal
with
it
anyway.
So
it's
just
a
matter
of
what
what
her
default
behavior
is.
C
What
what
kind
of
force
forcing
functions
we
have,
but
I,
don't
even
know
how
that's
that's.
Why
I
asked
this
like
how
Frozen
is
that
whole
thing
right?
How
like
are
we
still
kind
of
as
we
Implement
and
going
through
use
cases
open
to
changes,
or
is
that
already
you
know
written
in
stone
like?
Where
are
we
with
that
right?
That's!
That
was
actually
like
the
underlying
question
as
well.
C
B
Yeah
I
think
I
mean
this
document
is
to
the
best
of
my
knowledge,
definitely
like
in
development.
I.
Think
that,
just
about
anything
you
if
you
can
come
up
with
a
good
reason
for
why
it
should
change
it'll,
be
changed.
C
E
I
think
typically
the
Opium
spec
that
was
in
a
separate
repo
and
then
we
would
use
you
know
normal
PR
against
it
and
review
process,
and
then
it
becomes
much
more
formal.
This
is
more
of
a
proposal
for
us
back
and
I
I
think
eventually
it
should
live
and
get
somewhere
I'm,
not
sure.
If
that's
going
to
live
in,
you
know
maybe
following
a
folder
of
a
quadrical
trip
where
the
supervisors
implemented
or
or
somewhere
else,
but
that
would
that
would
be
the
process.
E
I
think
at
this
point
it's
still
very
much
a
work
in
progress
and
a
discussion
in
the
Google,
Doc
and
I
think
everyone
out
it's
just
a
matter
of
bringing
two
people's
attention,
because
it's
been
a
few
probably
months
since
that
was
introduced.
Really,
you
know
discussed
widely
so
right.
E
Confirm
that
we
by
default
do
run
as
root
on
route.
Of
course,.
C
I
always
when
I
see
that
fear
that
it's
most
like
there
are
really
not
many
good
reasons
to
to
run
this
route
and
I
I,
simply
don't
I
mean
there's.
On
the
one
hand,
the
supervisor
right
there
is,
on
the
other
hand,
The
Collector,
but
I'm
really
struggling
to
see.
Why
would
you
want
to
run
this
room
I'm
trying
to
to
go
through
that
in
my
head,
but
it's
probably
easier
based
on
implementation,
but
maybe
this
is
something
I
think
Evan
you,
you
said
you
haven't
really
explored
that
part.
B
It's
not
in
depth,
I,
don't
know
of
any
reasons
that
the
supervisor
would
need
to
run
his
route
so
long
as
it
can
get
a
I
think.
Opening
ports
is
probably
the
most
permissions
that
are
needed.
It
will
open
ports.
It
will
download
collector
executables
at
some
point,
but
that
shouldn't
need
root
permissions
by
any
means.
B
Right
right
and
then
it
will
need
to
it
would
need
some
mechanism
for
changing
the
user
ID
of
The
Collector.
If
we
Implement
that
but
again,
I
think
that
using
an
existing
system
tool
is
nearly
100
the
way
to
go
there.
It's
just
a
matter
of
how
that
works,
but
I
don't
see
any
reason
that
the
the
supervisor
should
be
run
as
root
under
normal
circumstances.
The.
C
Only
so
plus
one
to
it,
the
only
thing
that
came
to
mind
was
that's
still
something
I'm
trying
to
figure
out
what
that
actually
means
and
go
is
this.
You
know
packages
or
sub
packages
or
whatever
like
go,
doesn't
really
have
this.
You
know,
idea
of
a
plug-in
or
whatever
so
I
understand
that
you
know
you
have
a
new.
C
There
is
a
new
body
right,
like
you
want
to
replace
that
great
I
I
get
that
unclear
if
you
need
probably
maybe
but
other
than
that,
yeah
there's
still
anything
that
came
to
mind
where
potentially
you
might
be
need
to
be
rude.
B
C
B
Hopefully,
this
week,
I
should
have
a
like
a
skeleton
out
at
least
I'm
kind
of
wrapping
up
a
few
other
things
right
now.
But
that's
my
target
is
end
of
this
week.
Right
now,
and
then,
after
that,
it's
gonna
be
mostly
opening
up
issues
and
then
figuring
out
which
pieces
we
can
get
consensus
on
and
then
implementing
those
as
we
solidify
that.
A
E
Far
been
handling
all
reconfigures
as
sort
of
restarting
the
collector
within
the
process
and
referred
into
a
lot
of
issues
with
that
there
was
we
fixed
a
bunch
of
them,
but
they're
they
seem
to
crop
out
periodically.
So
some
some
things
were
I
remember
the
last
one
was
like
the
Prometheus
scraper
running
as
a
Singleton
and
didn't
properly
reset
when,
when
the
collector
instance
was
restarted,
so
I
think
having
a
supervisor
and
actually
restarting
the
you
know,
collector
process
is
a
new
process.
B
Yeah,
that's
the
collector's
not
quite
ready
for
that
sort
of
kind
of
component
level
restart.
Yet
from.
B
Was
some
work
underway
to
solve
that,
but
the
other,
the
other
key
piece
that's
going
to
be
needed
to
really
make
this
work
is
going
to
be
the
extension
I'm
currently
doing
some
work
in
the
collector
core
to
try
and
get
that
get
the
kind
of
foundation
for
that
ready
and
then
Sean
Porter
is
working
on
the
actual
extension
right.
So
hopefully
that'll
that'll
keep
moving
here,
although
that's
been
yeah,
it's
been,
it's
been
coming
along.
C
Have
you
heard
anything
because
I
pinged
him
on
on
slack,
but
maybe
he
was
PTO
or
I,
didn't
see
it,
but
I
I'm?
My
understanding
is
that
the
extension
is
kind
of
like
the
the
dependency
for
everything
else.
It
first
needs
that
extension
in
the
collector
in
order
then
to
to
have
the
the
non-extension
part
outside
of
like
supervised,
Etc,
workable
or
useful.
B
Right
yep,
so
the
well
what's
needed
is
there's
a
couple
pieces
that
need
to
change
in
The
Collector
for
the
extension
to
get
the
information
it
needs.
The
extension
then,
like
you
said,
will
communicate
with
the
supervisor
to
kind
of
tie.
Everything
together.
B
C
C
C
Yeah
and
then
obviously,
if
you
have
feedback
on
that,
one
I
love
claiming
I
am
an
op-amp
expert.
Yet,
but
I
I
definitely
want
to
invest,
get
there
so
yeah.
B
All
right
is
that
oh
we've
got
on
the
agenda
for
today.
E
There's
nothing
on
the
agenda
really
so
and
I
didn't
have
anything
Dad,
I'm,
I'm,
gonna,
add
a
note
just
so
that
when
when
Tegan
comes
back
or
whatever
anybody
else
looks
at
the
agenda,
they
see
that
we
discussed
this
and
and
hopefully
we
can
reactivate
this
discussion
and
get
some
clarity
and
maybe
formalize
the
spec
a
little
bit
more
as
we
implement
it.
B
E
D
C
Harder
to
go
to
Valencia
last
year,
I
had
to
go
by
Portugal
or
whatever
attempt
to
most
of
the
most
of
the
bigger
cities
in
the
US
from
Dublin
really
well
connected
yeah.