►
From YouTube: 2022-07-12 meeting
Description
cncf-opentelemetry@cncf.io's Personal Meeting Room
A
A
A
A
A
All
right,
we
have
a
demo
of
bind
lane
by
observer
iq
today.
I
think
that's
the
only
item
we
have
in
the
agenda
before
we
start
that.
I
think
it's
worth
saying
that
the
opamp
spec
is
now
declared
beta,
which
is
a
good
milestone,
yeah
we're
moving
forwards
toward
towards
more
stability.
So
that's
good!
A
Okay,
I
think
it's
two
minutes
past
we
can
go
ahead
mike.
Are
you
doing
the
demo
or
someone
else.
B
Yeah
I'll
do
a
short
demo
and
feel
free
to
ask
any
questions
as
we
go
through
it.
Okay,.
B
All
right
now,
if
everyone
can
see
my
screen,
okay,
yep,
oh
so
I'll-
give
a
little
bit
of
background
for
people
that
have
not
seen
a
version
of
my
plane
before
it's
a
product
that
we've
developed
and
was
primarily
you
know,
built
around
proprietary
agents
was
what
was
designed
to
manage.
Those
agents
manage
the
configuration
and
lifecycle,
and
so
this
is
by
plane,
op,
which
is
the
next
generation
of
that
product.
It's
it's
open
source,
so
it's
available,
and
you
have
to
give
a
link
in
the
dock
here.
B
If
anyone
wants
to
try
it
out,
but
it
just
went
into
beta
and
really
the
the
the
concept
behind
it
is.
It's
designed
for
agent
management
and
configuration
focused
on
open
telemetry,
at
least
initially,
and
that's
what
the
beta
release
is
doing
and
then
as
we're
adding
functionality
to
allow
things
like
observability
pipeline
management,
so
making
it
simple
to
do
filtering
along
with
configuration
either
reducing
data
or
sending
it
to
different
destinations.
B
So
I'll
just
go
through
a
quick
overview
of
it
and
if
anyone
has
any
questions
along
the
way
please
jump
in
and
if
anyone
andy
or
others
have
have
comments
on
the
way
it's
working
behind
the
scenes
specifically
with
opban
also
feel
free
to
jump
in.
B
So
this
is,
you
know
there
are
a
few
components
here
I'll
show,
but
this
is
a
product,
that's
designed
to
be
deployed
self-hosted
by
a
end
user.
So
I
have
it
hosted
in
our
environment
right
now,
but
our
sas
version
of
it
will
be
coming
later.
B
So
a
few
things
to
note
before
I
go
through
some
configuration
one,
it's
designed
to
have
a
lot
of
agents
all
at
once,
so
so
you
know
potentially
thousands
of
agents
connected
all
at
once
and
then
manage
the
configuration
for
those
and
so
on
this
first
page
you
see
these
are
all
of
the
open,
telemetry
agents
that
we
have
connected
along
with
their
configurations
and
some
basic
status
information
on
each
of
those.
B
So
I
have
a
few
hundred
of
them
connected
in
this
demo,
environment
of
varying
types.
So
some
windows
mac
os
some
linux
they're,
going
to
be
fairly
similar,
but
and
then
we
have
a
number
of
different
configuration
types,
and
each
of
these
configurations
refers
to
a
specific,
open
telemetry
configuration
that
we
can
apply
to
one
or
many
agents.
B
A
configuration
tab
shows
you,
each
of
those
that
we've
defined
and
I'll
go
through
and
create
one
as
well,
so
just
define
for
different
scenarios.
So
you
might
have
some
that
are
specifically
for
if
you
have
a
fleet
of
linux
hosts,
you
want
to
gather
os
metrics
and
logs,
maybe
a
few
other
specific
custom
items
from
those.
You
can
do
that
with
this.
If
it's
something
like
you
know,
maybe
you
have
a
production,
custom,
application
environment.
You
would
do
that
as
well
and
then
components
are
reusable
pieces.
B
B
B
So
you
know,
I
run
this
command
and
it
will.
This
is
just
a
a
script:
installer,
that's
wrapped
around
open,
telemetry
and
that
will
install
the
hotel,
collector
it'll
then
set
it
as
a
service
and
that'll
run
locally
and
as
soon
as
it
does
that
it
also
sets
a
few
configuration
values.
So
it'll
come
up
in
here,
it'll
automatically
connect
to
buying
planop.
So
this
is
my
m
kelly
demo
system.
B
I
just
ran
that
on
it
connected
to
buying
plano
p-
and
this
was
this
was
actually
from
a
previous
install
by
plano
p-
knew
that
it.
It
was
looking
for
a
specific
configuration
and
sort
of
applied
that
config
to
the
system
so
anytime
they
run
out
they.
These
collectors
will
start
up,
they'll
know
to
collect,
to
connect
to
bind
op
and
then
pull
down
the
latest
configuration
that
they
have
set.
B
So
the
next
piece
you
know
is
is
understanding
configurations
and
by
the
way,
all
of
this
is
using
the
op-amp
protocol
to
configure
to
connect
those
agents-
and
this
is
you
know,
a
wizard
driven
configuration
builder.
We
can
also
use
raw
configuration
just
drop
in
open,
telemetry
config,
but
in
this
case
we'll
use
some,
let's
create
our
own.
B
And
what
this
does
is
it's
taking
known
receivers
and
also
log
plugins,
so
this
is
using
our
plugin
framework
and
it
allows
you
to
configure
those
really
easily.
So
in
this
case,
if
I
want
to
create
a
configuration-
let's
say
I'm
doing
this
for
that
host
scenario
that
I
talked
about.
Maybe
I
want
some
host
metrics.
B
B
You
can
do
multiples
here.
So
if
I
also
have
a
let's
say
I
have
a
you
know,
I
could
just
grab
a
raw
file.
If
I
wanted
to
there's
some
csv
logs
or
files,
I
can
do
any
of
those
use
any
of
those
options
or
if
I
have
something
that's
known
like
maybe
I'm
running
my
sql,
I
can
do
either
metrics
or
logs,
because
both
of
those
are
supported
here
and
I'll.
Take
all
those
required
parameters
fill
in
defaults
as
known
defaults
and
then
allow
you
to
modify
them
as
needed.
B
So
in
this
case
I
can
say
maybe
I
just
want
logs
for
my
sql
I'll,
save
that
and
now
I'm
gathering
logs
for
my
sql
metrics
from
the
host
so
next
and
then
I
need
a
destination.
B
We
have
a
few
that
are
supported
right
now,
we're
adding
a
lot
more
of
these
and
really
all
of
these.
If
you
see
a
destination
or
a
source,
these
are
supported
with
just
a
yaml
file
on
the
back
end
that
defines
what's
the
receiver
that
you're
using
specifically
the
exporter
that
you're
using
and
then
how
do
you
configure
it
and
then
you
know
you
add
an
icon
and
you
can
load
it
into
my
plane.
B
So
in
this
case
I'm
going
to
use
one
because
I
know
this
already
exists,
just
choose
one
that
already
existed
and
save
it
and
that's
it.
Now
I
have,
I
have
a
configuration,
it's
a
valid,
open,
telemetry,
config
I'll,
show
you
how
you
can
pull
that
down
from
the
back
end
as
well,
but
I
can
also
apply
it
to
any
of
the
agents
that
are
hosted.
So
in
this
case,
I'm
going
to
find
that
system
that
I
just
created.
B
And
I'm
going
to
select
that
and
apply
this
config
file,
so
you
see-
maybe
you
saw
maybe
you
didn't,
but
it
flickers
for
just
a
moment
where
it
goes
from
connected
to
configuring
to
configured
and
what
that
did.
Is
it
pushed
down
the
configuration
file
now
this
demo
system
is
running
with
this,
with
this
configuration
and
gathering
host
host
and
host
data,
some
mysql
logs
and
sending
those
to
google
cloud
destination.
B
So
that's
like
in
the
simplest
version
of
this
just
one
host
one
configuration
but
most
of
the
customers
that
we
have
they're
using.
This
are
dealing
with
much
larger
environments
and
across
a
bigger
scale
in
those
cases
we're
looking
at
something
like.
Maybe
I
want
to
roll
this
out
over
a
fleet
of
production
systems,
and
so
I
have
a
in
this
case.
B
I
have
a
raw
config
file,
it's
not
applied
to
any
agents,
but
I
can
add
those
agents
and
just
do
something
like
I
have
these
systems,
so
I
have
over
100
of
these
they're
all
you
know
linux
systems,
none
of
them
have
a
configuration
currently
applied,
and
so
I
will
just
select
all
those
and
apply
and
they
take
you
know
a
couple
seconds
and
that
will
what
it
does
behind
the
scenes
is
it.
It
applies
a
label
to
each
of
those
hosts.
B
That
label
is
then
used
to
identify
what
the
configuration
version
is,
and
then
the
configuration
is
pushed
down
to
those
specific
agents
and
again
all
using
op-amp,
and
then,
as
this
has
changed,
you
know
I
can
do
something
like
I
could
make
a
a
change.
I
shouldn't
right
and
you
know
let's
say
that
I
I
do
this
it
will.
You
know,
may
pop
up
with
an
error.
I
know
it's
validating
that
change,
and
now
I
need
to
know
that
this
batch
size
needs
to
be
corrected.
B
You
know
the
the
high
level
of
how
you
would
apply
this
to
say
hundreds
of
systems
and
do
that
you
know
really
easily
and
really
quickly
and
then
manage
a
config
either
a
raw
config
like
this
or
from
that
you
know,
ui,
driven
version
where
you
know
in
this
case
I'm
never
I'm
never
going
to
be
using
that
like
a
raw
config,
I
can
actually
just
use
the
sources
and
destinations
here
well,
I'll
show
a
couple
of
other
things,
but
one
one
thing
to
note
is
in
this:
just
has
you
know,
receivers
and
exporters
we're
adding
in
processors
and
that's
going
to
be
in
a
release,
expect
at
the
end
of
the
month
and
that'll.
B
Allow
you
to
add
on
some
additional
filtering
on
any
of
these
sources,
so
you
can
say
really
easily
all
right.
Well,
I
also
want
to
filter
down
the
data,
that's
flowing
through
from
say
my
sql
or
raw
log,
and
build
out
your
configuration
all
from
this
ui,
and
one
place
that
is
particularly
useful
is
in
a
aggregator
model.
B
So
this
is,
you
know
it's
very
common
that
we
would
see
a
customer
that
has
maybe
a
hundred
or
a
thousand
nodes,
but
they're
not
sending
that
data
directly
to
the
destination
they're
going
to
go
through
an
aggregator
or
a
processing
node
and
in
those
cases
typically,
that's
you
know
somewhat
complicated
to
to
do
within
configuration
or
maybe
you're,
using
one
agent
to
one
agent
type,
assuming
it's
like
a
proprietary
agent.
B
If
you
use
one
agent
type
to
gather
the
data,
you
send
it
to
another
agent
type
that
that
is
designed
as
processor
or
aggregator
and
then
that
that
sends
it
along.
In
this
case,
we
can
use
open
solution
for
all
of
this,
and
so
I
would
just
create
a
new
configuration.
I'm
going
to
call
this.
My
aggregator
config
say
it's
linux.
B
In
this
case,
my
source
is
just
otlp.
So
now
this
is
going
to
support
my
metrics
logs
traces
coming
into
this
default
port
and
I'll
make
my
destination
the
place.
I
was
sending
data
to
this
this
google
destination
here
and
I've
now
created
an
aggregator
config.
So
then
you
can
add.
If
you
wanted
to
this,
is
this
can
act
as
either
just
a
central
collector
that
will
you
know,
gather
data
and
then
forward
it
along
to
destination.
B
So
if
you're
in
a
secure
environment,
you
don't
want
all
of
your
nodes
pushing
that
to
the
destination.
This
is
a
very
common
scenario,
but
we
also,
as
we
add
filtering,
then
you
can
do
processing
on
this
board
as
well.
So
if
you
want
to
offload
the
processing
from
those
external
nodes,
move
it
to
this
aggregator,
that's
becomes
really
easy
to
do
here
and
I
can
drop
this
onto
a
you
know
any
of
these
nodes,
I'm
going
to
pick
this
rod
33
because
it
happens
to
be
handy.
B
I
push
this
configuration
there
now.
This
is
listening
for
otlp
and
it's
sending
data
that
it
gets.
And
then
I
can
go
back
to
that
config
that
I
just
created
right
and
I'm
going
to
say.
Well,
I
don't
really
want
this
to
go
to
directly
anymore.
I
want
to
go
through
that
aggregator
node,
so
I
can
do
that
by
aggregator
I'll
delete
this
one
and
it
just
updated
and
sent
it
down
so
just
redirected
the
traffic
from
going
directly
from
my
host
to
the
destination
to
host
to
aggregator
destination.
B
So
that's
just
a
you
know,
kind
of
quick
overview
of
what
we're
doing
with
this
one.
Last
piece
you
know
everything
here
is:
let
me
get
things
a
little
bigger
everything
that
you're
seeing
here
is
also
available
via
cli.
B
And
as
I
mentioned
before,
you
know
I
built
configuration
via
a
ui
wizard,
but
I
can
then
export
it
if
I
want
to,
in
different
formats,
just
say:
get
config
demo,
so
the
one
I
just
created
into
a
yaml
file
and
that's
your
raw
hotel
configuration
file,
and
so
you
could
you
can
export
that
and
use
it
manually.
So,
even
if
you
just
want.
B
For
the
raw,
that's
the
that's
the
one
you
would
apply
to
another
fine
plane
system.
Okay,
so
there
we
go
yeah
exactly.
C
B
Yeah,
so
you
know
this
even
if
you're
just
using
bind
plane
to
say
like
just
just
to
create
these,
you
can
also
use
it
and
then
export
these
configuration
files
so
yeah.
So
that's
I
just
want
to
give
a
quick
overview
and
show
people.
You
know
what
we're
up
to
and
obviously,
if
anyone's
interested
in
trying
it
out
we'd
love
to
hear
your
thoughts
on
it
and
I'm
happy
to
answer
any
questions.
B
It
sounds
like
josh
had
a
question
in
chat:
do
we
plan
to
support
the
processors
from
the
contrib
repo,
I'm
not
sure
the
exact
list
of
initial
processors
that
we're
going
to
support
but
they're
pretty
simple,
to
create
using
a
yaml
file
and
a
go
template?
B
So
I
imagine
it's
going
to
be
limited
by
the
processors
that
are
supported
by
the
agent
and
I'm
not
sure
everything.
That's
included
right
now
in
our
our
distribution,
which
is
the
only
agent
we're
currently
compatible
with,
because
it's
the
only
agent
that
is
speaking
up
amp
at
the
moment.
Okay.
So
it's
intended
to
be
pretty
broad,
though
what
we're,
including
in
this
as
processors
and
and
like
andy,
mentioned
the
the
ability
to
sort
of
make
a
processor
available
in
fine
planop
is
pretty
straightforward.
B
It's
just
laying
out
what
those
parameters
are
and
specifying
that
processor
or
group
of
processors,
and
then
it
becomes
available
to
to
add-on
awesome.
Thank
you
josh.
I
can.
I
can
post
a
link
in
the
there's,
a
blind
playing
slack
by
the
way
that's
linked
on
the
for
everybody
else.
I
know
you're
in
it
josh,
but
there's
a
blind
plane
slack
in
the
fine
plane,
op
repo
and
I'll
I'll
paste
the
link
to
the
pr
that
includes
processors.
B
That
will
give
you
an
example
of
what
a
processor
looks
like
and
we're
we're
hoping
to
have
that
feature
complete
this
week
and
then
probably
released
into
the
month
sweet.
D
I
did
have
a
quick
question
so
related
to
status,
reporting
right
when
you
go
on
to
agents
and
and
look
at
those-
I
I
believe
those
are
not
just
when
we
are
pushing
down
conflict,
but
actually
they'd
be
basically
be
able
to
change
every
time
the
status
of
the
agent
or
the
collector
in
this
case
changes.
Let's
say:
there's
an
out
of
memory
concern
because
the
data
that's
being
sent
in
so
this
status
will
change
thereafter.
Is
that
accurate.
B
Yeah
right
now
we're
not
so
so,
there's
a
newer
message
in
in
op
amp
for
health.
That
would
probably
more
be
more
appropriate
for
the
out
of
memory
situation
and
that's
not
being
reported
at
the
moment.
We
do
plan
on
including
that
in
our
next
release.
That
was
a
feature
that
tigran
emerged
last
week.
B
I
believe
we
do
currently
report
configuration
errors,
so
if
we
send
down
a
bad
config
that
will
put
the
agent
in
an
air
state
and
that
air
state
will
then
be
reported,
one
thing
that's
that's
kind
of
interesting
about
that
is
that
the
collector's
not
designed
to
be
run
with
a
certainly
a
bad
config,
but
also
with
no
config,
so
we
actually
have
kind
of
a
an
empty
config
that
all
it
has
is
a
single
host
receiver
that
collects
metrics
every
hour.
B
So
we
tried
to
make
it
as
lightweight
as
possible,
but
that
that
configuration
is
applied
when
to
put
the
agent
kind
of
in
a
neutral
state
when,
when
that
error
could
be
fixed,
got
it
thanks.
B
Yeah
and
so
that's
what
I
was
showing
her
a
little
bit
earlier,
which
is
you
know
if
you
do
end
up
if
this
is
in
an
error?
State
you'll
see
that
show
up
as
a
an
issue
here
and
the
intent
is,
while
we're
still
determining
the
best
way
to
like
escalate
status
issues
or
performance
issues
with
the
host.
Those
would
also
be
here
presented
here.
A
I
think
this
is
great,
so,
first
of
all,
we
should
link
from
the
opamp
spec
to
to
bind
clay
as
a
great
usage
example.
I
I
I
think
I
missed
the
part
where
you
added
the
aggregator
as
a
destination.
A
Can
you
clarify
how
that
works
there
in
the
ui
for
one
of
the
agents,
you
you
define
that
the
aggregator
is
the
destination?
What
exactly
does
that
mean.
B
Yeah
so
there
are,
and-
and
that
was
one
step
here
that
was
you
know,
you
can
create
an
aggregator
within
a
configuration,
and
let
me
just
show
you
what
that'll
be.
B
B
You
know
some
platforms
use
otlp
by
default
and
we
can
just
say
this
is
the
the
destination
you
choose
a
protocol,
the
host
name
that
that's
going
to
be
targeting
and
then
the
port,
and
so
in
that
case
just
be
an
ip
address
for
an
aggregator
node.
And
then
you
can
use
this
repeatedly
in
different
configuration
files
and
you
missed
it,
partly
because
he
had
created
it
before
the
demo.
There
was
already
a
destination
at
the
very
beginning
of
the
demo.
B
In
theory,
the
architecture
would
support
reusing
sources
and
processors
as
well,
but
it
kind
of
complicates
the
user
experience.
We
think,
and
so
right
now
destinations
are
the
reusable
component
and
we've
also
been
doing
quite
a
bit
of
thinking
about
facilitating
the
deployment
of
an
aggregator
where,
when
you
install
an
aggregator,
it
automatically
becomes
a
destination
for
for
other
for
for
better
pipeline
management.
B
So,
but
there
was
a
step
that
he
took
before
that
you
didn't
see
originally
yeah.
A
Yeah
so
so
the
destination
is
the
components
here
are
partial
configurations
right.
Is
it
fair
to
say
so.
A
A
B
A
source
can
contain
any
number
of
receivers
and
processors
and
then
like.
If
you
want
to
do
like
get
source
types
and
then
maybe
look
at
a
source
type
and
do
dash
our
yml,
and
we
can
actually
see
what
that
looks
like
is
it
it's
typically
a
hyphen.
We
should
probably
add
an
alias
there.
B
I
don't
know
what
a
good
example
is,
but
why
don't
you
just
grab
like
nginx
or
yeah
rattus
is
fine
and
then
do
dash,
oh
yaml.
So
as
we're
looking
at
this,
so
we
have
kind
of
partial
pipelines.
Here
we
have
separate
headings
for
logs
and
and
metrics,
and
then
they
can
specify
receivers
and
processors,
and
the
receivers
and
processors
are
actually
that
pipe
and
emil
is
is
saying,
is,
is
making
that
all
a
string.
B
So
it's
technically
a
string
that
includes
a
a
go
template
that
can
reference
the
parameters
that
are
specified.
B
So
it's
quite
easy
to
create
one
of
these
source
types.
There's
parameters
specified
right
in
the
same
yaml
file
as
the
source
type
and
then,
when
a
source
type
exists
in
the
system
it'll
be
available.
When
you
click
add
source
and
then
you
can
create
any
number
of
sources
that
use
the
source
type,
so
destination
type
works
the
same
way.
It
can
include
any
number
of
processors
and
destinations
or
sorry
exporters,
and
then
processors
are
generally
just
processors.
B
There's
a
lot
of
detail
about
and
I'm
sure
you're
familiar
with
the
tigran,
but
not
everybody
would
be,
but
the
detail
of
how
to
author
that
config
with
multiple
pipelines.
So
some
of
these
can
get
quite
complicated
when
you're,
enabling
both
logs
and
metrics
with
multiple
sources
and
and
a
destination
that
can
support
them.
So.
A
Okay
and
and
the
ui
is
rendered
based
on
the
the
source
type
definition
yamo.
B
A
A
B
A
B
Remove
them
just
for
fun
mike,
if
you
want,
you
could
type
that
to
bind
plain
apply,
dash,
f
dash,
that's
my
favorite
demo
example.
I
showed
it
to
josh
last
week
in
the
chat,
but
if
you,
if
you
just
pipe
and
then
bind
plane,
apply
dash,
f
and
then
a
dash
after
that,
that'll
take
it
from
standard
in
and
then
you
will
effectively
reapply
the
redis
source
type.
B
But
from
there
you
can
kind
of
get
the
idea
of
how
you
might
grab
the
reddest
one
make
a
few
changes
and
then
reapply
it.
So
the
way
we
we
actually
embed
the
yamls
into
the
binary,
I'm
using
go
in
bed,
and
so
all
the
source,
types
and
destination
types
of
ship
are
embedded
and
then
they
can
be
added
any
additional
ones.
A
Nice
nice
is
this
documented
on
your
website
somewhere.
B
I
think
I
don't
think
this.
That's
a
good
question.
B
I
think
well,
at
least
from
this
high
level,
the
the
adding
and
removing,
but
maybe
not
the
specific.
I
don't
know
that
we
have
documentation
for
how
to
create
them.
Yet
hopefully
it
will
be
coming
if
it's
not
there
yet
and
just
as
an
example,
so
that's
still
loaded,
but
we
can
remove
it
or
add
it.
I
actually
don't
know
if
I
remove.
E
It
I'm
sorry
you
have
to
refresh,
I
think,
would
I
would
reload
the
changes
now.
B
There
you
go
so
I
just
I
just
deleted
redis
from
here
and
export
it,
and
then
I
can
just
do
the
same
thing
to
to
add
it
back
in.
B
Cool,
so
we
think
it's
possible
that
someone
may
configure,
you
know,
bind
plane
for
their
environment
to
only
have
sources
that
are
relevant
to
them
and
and
destinations
that
are
relevant
to
them
and
and
and
then
let
the
users
within
their
organization
use
it
in
sort
of
a
constrained
fashion,
and
maybe
it
has
a
couple
additional
source
types
that
are
specific
to
applications
that
are
running
in
their
environment
that
have
their
own
log
parsing.
And
it's
it's
fully
intended
to
serve.
B
A
B
Awesome,
I
appreciate
you
sharing
some
time
for
us
to
go
through
with
you
and,
of
course,
happy
to
have
hear
any
feedback.
If
it
comes
in.
You
know,
after
the
fact,
just
let
us
know
and
our
select
channels
open.
A
Great,
thank
you.
Please
file
a
pr
to
add
to
other
link
to
combined
claim
from
from
opump.
I
think
it
will
do
it
fully
deserves
that.