►
From YouTube: 2023-02-02 meeting
Description
cncf-opentelemetry meeting-2's Personal Meeting Room
A
A
A
A
A
A
A
Please
put
it
into
the
dock.
There
is
two
things
I
want
to
discuss
and
it's
the
first
one
is
the
release
rotation.
So
right
now
we
essentially
there
are
volunteers
that
help
us
with
releases
and
I.
Just
wonder
if
we
should
maybe
rotate
the
release.
Responsibility
across
maintainers
I'm,
not
sure
if
this
is
a
practice
in
other
Hotel
projects,
but
that
way,
we'll
kind
of
ensure
that
release
will
go
always
out,
because
we
will
have
somebody
responsible
for
a
specific
release.
D
I
think
it's
a
good
idea:
I
forgot
what
the
collector
stick
does,
but
I
do
think
that
they
have
some
type
of
rotation
or
volunteer
who
takes
it
on
every
for
every
other
week
or
something
Anthony
Craig.
If
I'm
wrong
on
that,
but
I
think
that's
the
case.
C
That's
correct:
there's
a
schedule:
it's
in
the
release,
dock
in
The
Collector
core
repo.
Every
week,
one
of
the
maintainers
or
approvers
is
slated
to
manage
the
release.
A
C
A
B
C
I,
don't
think
that's
something
we've
encountered
in
The,
Collector
I.
Think
the
expectation
is
it's
part
of
the
responsibility
of
being
a
maintainer
or
approver,
and
if
you're
not
up
for
it,
you're
going
to
be
a
maintainer
or
approver
s.
Yes,
I
think
a
little
easier
in
The
Collector,
where
there
are,
you
know
eight
or
ten
people
at
the
rotation.
So
it's
you
know
something
that
you
have
to
do
every
couple
of
months.
I
think
here,
it'll
be
a
bit
more
frequent,
but
would
be
my
expectation
notice.
A
Yeah,
that's
fair
enough.
I
think
it
will
as
well
maybe
keep
the
the
list
of
approvals
and
maintainers
up
to
date.
Since
you
know
we
will
require
some
responsibilities
regularly.
A
The
other
action
item
is
again
related
to
the
other.
Agent
item
is
related
again
to
the
release
and
we
we
release
the
operator
in
the
operator
repo,
but
then,
as
part
of
the
release,
we
submit
to
pull
requests
to
the
to
the
it's
olm
to
the
operator
have
repositories.
There
are
two
repositories,
and
it's
mostly
I
would
say.
Maybe
redhead
related
I
mean
somebody
can
use
the
operator
Hub
without
openshift
and.
B
D
A
A
D
I
see
yeah
I
mean
this
seems
like
pretty
easily
automatable.
D
Is
that
a
word
I
don't
know,
I
think
it
might
be
worth
like
setting
up
a
a
workflow
that
runs
posts
as
part
of
the
release
workflow.
D
That
then
opens
this
PR.
How
easy
is
it?
Do
we
have
to
like
talk
to
anybody
to
get
it
merged,
or
are
they
pretty
diligent
about
giving
this
merged
quickly?.
A
I
think
they
are,
it
has
to
be
approved
by
one
of
the
owners
of
the
operator
and
the
owners
are
defined
in
the
in
one
of
the
meta
files
here,
but
the
owners
are
essentially
usually
the
maintainers
of
the
operator
for
a
subset
of
those
maintainers.
D
Okay,
yeah,
it
doesn't
seem
terrible
and
GitHub
does
have
the
ability
to
like
open
up
these
PRS
from
from
actions
now
I
think,
rather
than
just
like
doing
a.
A
D
Question
I
feel
you
should
probably
be
able
to
like
pipe
it
through,
but
that's
a
lot
of
yes,
a
lot
of
piping,
but
I
don't
see
why
not?
Let
me
check
their
API.
D
B
A
D
Yeah
that'd
be
great.
Thank
you.
It
looks
like
you
use.
They
have
a
convenience
action
called
GitHub
script,
which
is
basically
a
pre-loaded
JavaScript
environment
with
a
GitHub
token,
and
like
everything
that
you
need
to
run
scripts,
so
I
think
that
as
long
as
they
as
long
as
their
API
exposes
the
ability
to
set
the
owner,
it
should
be
possible.
A
A
B
Some
reason
I
can't
speak
once
I'm
sharing
I
make
a
screenshot
and
put
it
into
it.
The
document.
B
B
Yes,
it
might
be
a
bit
ugly,
so
it's
that
Pavel
and
I.
We
have
a
presentation
on
first
on
Sunday
and
we
wanted
to
speak
about
the
open,
Telemetry
operator
and
all
the
signals
and
yeah.
So
since
there
is
the
target
allocator,
which
makes
life
a
bit
easier
with
scraping
metrics
we
or
I
glued
this
thing
here
together
and
I
was
curious
to
hear
your
opinion
about
it.
B
If
I
forgot
there's
something
which
is
basically,
we
have
the
the
scrape
configuration
which
gets
distributed
by
the
Target
allocator
or
it
gets
created
by
the
Potter
service
Monitor
and
then
distributed
by
the
Target
allocator.
B
Is
there
something
on
this
slide,
which
is
super
confusing,
which
is
wrong?
What
you
would
like
to
see
there.
D
I
think
one
thing
to
fix
would
be
the
connection
from
part
two
and
three
to
collector
one
and
two,
because
it's
scrape
based
rather
than
push-based
so
I
think
the
arrows
should
probably
go
the
other
way.
Oh.
B
D
See
yeah
I
think
in
that
case,
yeah
a
label
on
the
arrows.
D
Good
otherwise,
I
think
this
all
looks
right
to
me.
C
So
I
just
wanted
to
blow
this
diagram
from
the
original
design
dock
on
this,
it
kind
of
shows
the
same
sort
of
flow.
Does
this
help
make
sense
of
this
I
think
what
you've
got
is
in
alignment
with
what
I've
shown
here.
B
So
it's
someone
I
just
say
what
I
see
and
then
you
correct
me.
So
someone
is
creating
a
CR.
The
CR
gets
then
collected
by
the
operator.
So
from
the
controller
part,
this
one
is
then.
C
I,
don't
we
were
calling
the
calling
the
target
allocator
load
balancer
at
the
point
at
which
this
was
created,
but
yeah
substitute
check
it
out
here
where
you
see
load
balancer
here.
B
Okay,
so
then
we
create
the
collector
instances
and
The
Collector
instances
get
them
the
infos
from
the
load
balancer
so
from
the
target
locator
so
which
is
basically
then
the
same.
B
C
B
A
D
I
think
we're
finding
a
lot
of
people
are
using
the
CR
functionality,
the
one
where
it
receives
configuration
using
the
Prometheus
service
monitors
and
pod
monitors.
Definitely
something
that
we
hear
a
lot
about.
I
think
that
most
people
who
are
migrating
to
hotel
from
Prometheus
are
going
to
use
that
the
way
that
most
cluster
operators
that
I've
talked
to
and
that
I've
seen
configured
Prometheus
is
using
their
coup.
Prometheus
stack
chart
and
it
uses
those
custom
resources
out
of
the
box.
D
A
C
Not
really
I
think
from
the
beginning,
we've
talked
about
being
able
to
reuse
the
Prometheus
operator
CRS,
just
because
they
are
already
purpose
built
and
in
use
and
widely
adopted.
I
think
for
open
exploitment
reading
native
instrumentation,
the
the
whole
Target
Discovery
and
allocation
thing
is
not
really
necessary
because
it's
push-based
rather
than
poll
based,
and
so
you
just
put
a
load
balancer
in
between
your
targets
and
your
your
sources
and
your
destinations.
D
Yeah
and
also
the
I
think
the
instrumentation
resource
covers
most
of
the
like
Hotel
metrics
SDK
use
cases
right
where
it
can
just
inject
the
environment
variables.
The
SDK
is
expecting
to
send
to
the
in
cluster
collector.
D
D
Like
the
Autumn,
instrumentation,
libraries,
okay,
I,
don't
think
so.
I
think
the
python
one
can
Auto
configure
the
exporter,
but
I
don't
think
it
like
default
exports,
anything
other
than
maybe
up
like
aliveness
or
something
but.
B
C
C
I
think
we
wouldn't
have
built
this
entire
system
if
we
thought
it
was
likely
to
happen
quickly.
I
expect
people
will
be
using
Prometheus
instrumentation
libraries
for
a
long
long
time.
D
Yeah
I
agree:
I
think
that
there
are
more
people
who
are
starting
to
do
like
Hotel
metrics
first,
but
that's
you
know
it's
going
to
take
time
for
wider
adoption.
A
A
If
no
I
guess,
we
can
wrap
up
and
get
half
an
hour
back
cool.