►
From YouTube: 2022-08-02 CNCF TAG Observability Meeting
Description
Quick update from TAG chairs:
* Kubecon NA participation from the TAG
* TOC update (from meeting prior hour)
Call out for observability topics / speakers interested in presenting at TAG meetings
* Aug 16th: Liz Fong-Jones, "Evolving and hybridizing signal types - my journey from metrics/logs to traces to profiles"
Supporting APAC timezones for OTAG meetings
New Google Drive folder for TAG Observability has been created by CNCF.
(keyval) Introducing https://github.com/keyval-dev/odigos
B
B
A
D
D
A
C
I
put
a
link,
a
link
into
hotel
profiles,
but
there's
a
completely
blank
document
there.
The
only
thing
there
is
a
title
that
says,
like
type
vision
document
that
we
was
was
bandied
about
and
I'm
a
little
out
of
order
here,
but
for
what
it's
worth
headphones
that
it's
owned
by
like
projects
at
cncf.
So
it's
not
a
human.
It's
the
cncf.
A
D
D
A
A
C
C
A
A
So
that's
good
news,
the
other
part
that
we
are
starting
to
think
about,
and
it
would
be
nice
to
be
able
to
do.
This
is,
to
you
know,
just
a
lot
of
folks
plan
to
attend,
kubecon
and
typically
especially
the
the
us
one.
For
you
know
many
of
us
who
are
based
in
the
us
and
and
we'd
love
to
kind
of
at
least
have
a
boss
session
of
some
sort,
we're
trying
to
figure
out
if
there's
any
way
to
get
some
space
from
the
cncf.
A
Even
if
it
is
a
room
in
the
evening-
or
you
know
just
having
the
birds
of
a
feather,
but
if
not
you
know
maybe
just
having
a
dinner
together,
just
to
meet
up
and
say
hi
in
person,
so
those
are
coming
some
of
the
areas
you're.
Thinking
of
again,
you
everyone's
welcome
to
provide
you
know
their
ideas.
If
you're
coming
to
detroit,
then
definitely
would
be
awesome
to
catch
up
and
meet.
A
C
Yeah
sure
we
just
had
that
meeting
the
toc
is
it's
synchronized
with
our
meeting
first
and
third,
they
saw
our
schedule
and
the
tlc
decided
to
align
themselves
with
us.
That's
how
that
happened,
bad
joke,
so
I
pasted
in
the
slide.
We
try
to
do
one
slide.
Sometimes
there's
more,
but
today
we're
brief
bright
and
gone.
We
just
kind
of
covered,
what's
happened
in
the
last
month
roughly
in
these
communities,
the
talk
of
the
day
is
hotel
profiling
and
you
can
see
the
slide
right
there.
C
A
Yeah
and-
and
I
think
related
to
that-
there's
an
exciting
opportunity
to
also
work
with
the
tag
security
group,
because
I
mean,
as
many
of
you
know,
there
are
specific
sim
and
security
use
cases
in
observability
that
you
know
typically
are
advanced
use
cases,
but
would
be
good
to
kind
of
also
have
discussions
around
identifying
what
are
the
most
popular
use
cases
and
how
that
can
be
supported
in
different
projects.
Hotel
has
definitely
discussed
it.
A
Moving
on
to
the
next
topic
again,
just
a
call
for
you
know
speakers
and
topics
who
are
for
doing
a
speaker
series
in
the
observability
tag.
Again,
please,
you
know
share
this
with
your
networks
and
and
we
will
maintain
a
google
doc.
You
know
it's
as
simple
as
just
adding
recommendations
for
topics
and
and
speakers,
and
it's
really
interesting
to
kind
of
invite,
even
folks
who
are
not
attending
on
a
regular
basis
to
come
and
present.
A
A
The
next
item
that
I
had
was
just
you
know,
calling
out
the
the
ability
to
support
apac
time
zones
with
our
tag.
Meetings
and
some
of
the
other
tags
are
also
starting
to.
You
know
kind
of
support
that
today
we
support
europe
and
and
the
us,
obviously
with
this
time
zone
and
this
time
meeting
meeting
time,
but
would
love
to
kind
of
at
least
have
one
meeting
on
the
apac
time
zone,
which
means
about
4
p.m.
A
You
know
pacific
time,
which
is
typically
good
for
asia
and
and
just
enabling
you
know
many
of
our
participants
and
contributors
from
asia
to
join
in.
So
again,
I
think
it's
nice
to
have
that
we
can
figure
out.
You
know
what
makes
sense
from
an
periodicity
point
of
view.
If
there's
enough
momentum,
you
know,
maybe
we
can
start
with
it
once
a
quarter
and
then
kind
of
go
into
once
a
month
but
curious
to
hear.
If
other
folks
are
you
know,
what
do
a
folks
think.
A
Plus
ones,
I
know
we're
all
on
the
same
time
zone
so
to
us
it
would
work
out
on
on
specific
time.
It
works
out,
because
you
know
it's
either
8
a.m
in
the
morning
or
4
p.m
in
the
afternoon,
but
for
the
east
coast
it
becomes
a
little
later,
but
anyway
it's
it's
nice
to
have
folks
joining
joining
in
from
you
know,
different
projects
also
being
worked
on
out
of
asia,
and
last
but
not
least,
we
have
the
new
google
drive
folder.
Finally,
and
a
link,
please
bookmark
it.
A
C
Take
it
away
so
you
reached
out
to
me
a
week
ago.
I
want
to
say
I
said:
hey
I
made
this
new
apache
too.
That's
important.
This
apache
2
open
source
thing
about
observability
I'd
like
who
you
know
how.
How
could
I
you
know,
could
I
where?
Where
should
I
go
to
talk
about
it,
and
I
said
well,
why
don't
you
come
here?
So
this
is
not
a
cncf
project.
This
is
an
apache2
project,
but
it's
in
our
domain.
C
Our
area
of
you
know
what's
relevant,
so
it
is
in
scope,
and
I
can't
wait
to
hear
about
it
by
the
way.
Your
github
tag
has
to
be
one
of
the
cooler
github
tags.
I
think
I've
ever
seen
it's
key
valve
so
and
if
there
are
any
slides
you're
using,
please
feel
free
to
drop
them
so
that
others
watching
later
can
can
follow
along.
B
Yeah
sure
sure
we'll
be
happy
to
really
curious
to
hear
what
you
think
about
this
project.
So,
as
matt
said
just
last
week,
I
released
this
open
source
project
with
completely
apache
license.
Basically,
what
we're
trying
to
do
with
this
project
is
to
basically
help
developers
enjoy
like
the
entire
promise
of
observability.
You
know
traces,
metrics
and
logs
and
ability
to
correlate
between
one
another,
but
without
having
to
basically
be
very
familiar
with
either
open,
telemetry
or
evpn.
B
You
don't
have
to
know
any
apis
or
sdk
or
know
how
to
configure
any
collectors,
and
I
think
I
will
show
you
the
demo
yeah.
It
will
be
really
clear.
So,
on
the
one
hand
here,
I
have
like
I'm
downloading
on
data,
but
we
support
multiple
vendors
and
the
next
version
that
you
hopefully
come
out
tomorrow
will
support
many
other
open
source
vendors
so
before
we,
so
I
have
here
an
empty
database
that
I
just
created
and
on
the
other
end,
I
have
here
a
kubernetes
cluster
with
this
micro
services.
B
Second-
and
you
can
see
now
that
body
gas
is
detecting
all
the
running
pods
in
our
cluster
and
basically
that's
the
programming
language
of
every
application
running-
and
this
is
important
because
alice
is
going
to
use
their
the
best
intention
tool
for,
according
to
the
programming
language,
for
every
every
pod.
So
for
I
compiled
languages,
let
go.
B
We
are
actually
using
an
open
source,
another
open
source
project
that
we
released
and
after
telemetry
drawing
automatic
instrumentation
based
on
evpf,
which
is
finally
approved
by
the
telemetry
technical
committee
to
be
part
of
the
project.
So
hopefully
soon
it
will
be
under
open
telemetry,
but
for
for
other
applications
like
java.net
or
python,
we
will
use
the
regular
automatic
instrumentation
for
open
telemetry,
and
we
have
here
two
mods
of
operations.
B
You
can
choose
either
opt
out
which
will
basically
instrument
all
the
applications
and
automatically
instrument
any
new
application
that
is
being
deployed
into
this
kubernetes
cluster
or
we
can
choose
another
theme
mode
which
manually
select
which
applications
we
want
to
instrument.
So
I
choose
about,
and
basically
now
I
can
choose
wherever
where
I
want
to
send
my
observability
letter
which
vendor
I
would
like
to
use.
As
you
can
see,
we
currently
supported
four
vendors,
but
hopefully
tomorrow
there
will
be
like
a
forum
over
here,
mostly
open
source.
B
B
B
Again
without
having
to
it's
like
it's
obstructed
by
all
the
complexity
of
either
ebps,
which
is
really
complex
or
open
dynamic,
which
is
a
little
bit
nicer.
But
still
you
have
to
learn
on
those
sdks
and
apis
and
know
how
to
configure
these
collectors,
and
we
have
the
tracer,
the
metrics
and
for
other
vendors.
We
also
have
our
logs
and
yeah.
That's.
Basically,
it
really
curious
to
hear
what
you
think
about
it.
C
I'm
sorry
there
we
go.
Thank
you
so
much
for
doing
a
demo.
Might
you
walk
through
what
this
is
sort
of
below
the
ui?
You've
got
a
great
in
your
docs.
You've
got
a
reasonable
architecture
page,
so
perhaps
you
can
kind
of
talk
through
your
design
for
for
how
it's
implemented
for
those
that
might
be
looking
at
it.
For
the
very
first
time,
I've
also
dropped
in
a
link
to
the
cncf's
project
template.
C
B
Yeah
sounds
good,
look
at
it.
So
the
way
we
think
about
this
is
sort
of
like
a
control
plane
for
observability
data.
It
takes
care
of
integrated
to
either
instrumentation
or
collector-wise
in
a
high
level.
It
contains
four
different
components:
kubernetes
operators
and
instrumental
scheduler,
auto
scaler
and
the
user
interface
that
I
showed
you
basically
instrumental
is
the
component
that's
responsible
to
detect
every
pod's
programming
language
and
apply
the
the
right
and
instrumentation
either
open
the
lg
or
vpn
base.
B
But
of
course
you
can
do
it
in
yemen,
if
you
prefer-
and
the
scheduler
is
just
deposited-
ties
everything
together
and
the
schedule
between
waiting
for
the
collectors
to
come
up,
and
then
we
do
the
instrumentation
and
basically
coordinates
everything
together
and
yeah.
That's
about
it
having
some
more
documentation
again,
it's
a
really
new
project.
We
released
it
just
last
week,
so
you
can
see
all
the
different
data
that
we
know
how
to
how
to
send.
We
have
traces,
we
have
metrics.
We
actually
have
three
types
of
metrics.
B
We
have
like
metrics
that
are
based
on
the
traces,
which
is
something
that
most
of
the
benchers
do
automatically
and
we
have
like
runtime
metrics
like
a
garbage
collection
and
threads
and
heat
and
memory,
and
we
have
like
more
infrastructure,
metrics
like
host
environment,
cpu
memory,
disk
locations
and,
of
course
we
have
prices
and
loads
as
slots
are
pretty
near
and
open
telemetry.
We
are
not
able
to
like
correlate
them
efficiently
to
traces
and
metrics
in
every
programming
languages.
B
Basically,
it's
additional
fields
on
top
of
deployments
or
stateful
set
like
which
programming
languages
which
instrumentation
we
should
use,
which
collector
we
should
talk
to
and
collapse
group
is,
is
a
part
of
the
collector's
pipeline.
That's
responsible
to
achieve
some
common
goal:
either
collect
logs
or
collect
metrics
or
send
them
as
a
gateway
to
one
of
the
vendors.
D
Do
you
you
mentioned
that
it
uses
ebpf
as
well?
I
guess,
are
there
any
limitations?
I
guess
it
sounds
like
you're
kind
of
going
to
routes
to
get.
You
know,
for
example
like
traces.
I
think
it
says
that
you
do.
You
know
either
ebpf
or
hotel.
I
guess
you
know
sometimes
it's
hard
to
use
ebpf.
I
guess
you
know
you
have
to
be
on
the
right.
You
know
like
linux
version
and
stuff,
like
that.
I
guess.
Are
there
any
limitations
that
you
currently
kind
of
face
with?
You
know
things
that
rely
on
ebpf?
B
B
Yeah
so
good
question,
so
I
think
you
you
said
it
right.
Basically,
we
depend
on
being
on
the
linux
nodes.
I
think
that
evpn
for
windows
is
currently
in
development,
but
I'm
not
really
sure
about
what's
the
status
of
this
and
whenever
windows
will
support
that
our
automatic
instrumentation
will
be
supported
on
windows,
also
other
channel
version.
B
I
think
it
requires
kiana
version
four
point,
something
we
don't
remember
what,
but
anyway,
all
the
all
the
managed
kubernetes
offers
that
I
looked
on
the
popular
cloud
providers
or
the
hosted
or
the
locally
ranked
event
is
like
a
mini
cubic
kind.
All
of
them
are
currently
supported
in
the
recent
enough
channel,
and
so
I
guess
there
is
some
cases
which
we
won't
support,
especially
windows
and
like
really
old
kernels,
but
we
are
not
using
a
very
new
recent
features
of
ebpf,
which
basically,
let
us
use
a
pretty
old
candles.
C
I
have
a
I
have
a
question,
but
I
want
to
let
anybody
else
ask
questions
first,
okay,
so
so
this
is.
This
is
neat.
It's
clearly,
you
know,
you're
configuring,
open,
telemetry
components
directly,
that
that's
helpful
and
and
introducing
you
know,
operators
and
such
is
there
any
record
or
artifact
in
terms
of
what
configuration
has
been
generated
and
then
sent
to
a
server.
You
know
such
that
that
could
be
worked
into
a
git,
ops,
workflow
and
then
sort
of
similarly,
but
in
a
slightly
different
dimension.
C
C
B
So
yeah,
I
think,
that's
like
this
is
like
the
one
of
the
main
reasons
why
I
made
it
open
source
and
one
of
the
main
reasons
why
I
came
to
this
meeting,
to
show
it
to
you
that
we
want
to
be
as
open
as
possible
about
whatever
we
do
to
your
cluster
and
specifically
ebpf,
which
requires
elevated
permissions
and
again.
We
hope
that
we'll
have
many
users
that
are
using
it
and
there
will
be
a
sort
of
like
easier
way
to
use
open
parameters
in
the
vpf.
B
I
think
about
it
like
more
obstructed
the
way,
if
you
know
the
open
geometry
operator
project.
So
it's
a
little
bit
like
that,
but
like
not
obstruction
above
it
and
using
technologies
that
are
not
specifically
open,
telemetry
and
regarding
to
like
regulated
clusters,
and
it's
our
own
valid
points
and
stuff.
We
will
need
to
to
take
into
considerations,
make
sure
to
add
it
to
our
roadmap
and
look
into
it.
C
I
guess
cool
thank
you
for
that
and
a
follow
on
question.
If
there's
already
the
open,
telemetry
collector
configured
on
some
or
other
namespaces,
or
you
know,
parts
of
the
cluster
of
a
whole
cluster
systemically
or
other
open,
telemetry
components
that
are
already
there
and
then
a
user
walks
up
with
with
this
and
then
uses
it.
Are
you
doing
a
merge
of
configuration
or
is
it
unknown
you?
You
know
what
I
mean
again,
I'm
thinking
either
like
operationally.
B
C
Thank
you.
I
only
have
one
last
comment
and
one
suggestion
that
I
give
well
one
one
suggestion
and
one
one
final
question
that
I
give
to
all
all
folks
that
present
to
the
tag
from
from
wherever
they
come
from
in
terms
of
project,
whether
it's
cncf
or
otherwise
but
I'll
start
with
the
suggestion.
If
you
haven't
already,
I
strongly
recommend
it
and
other
folks
on
the
call
here,
steve
or
alolita
to
point
you
specifically
where
I
I
don't.
C
I
don't
actually
know
offhand,
but
where
would
the
best
place
within
the
hotel
set
of
community
groups
and
meetups
and
and
whatnot
sigs
would
be
best
suited
to
see
this
presentation
as
well,
because
you
know
there
might
be
similar
efforts
or
or
it'd
be
cool
to
have
their
feedback,
and
that
was
the
suggestion.
Then.
C
B
A
Yeah,
that's
that's
awesome
because
I
mean
being
on
github
eden.
I
think
it's
easy
to
contribute
and
track
the
issues
on
the
project
itself,
but
from
an
open,
telemetry
perspective.
I
would
recommend
that
again,
if
you
haven't
already
please,
you
know,
given
that
you're
working
with
the
go
sdk
definitely
join
into
the
go
sig
and
sig
meetings
which
are
weekly
and
present.
You
know
an
overview
of
the
project
as
well
as
the
typically
the
maintainers
meeting.
B
A
A
C
Else
on
the
agenda,
if
folks
want
to
talk
about
that
landscape,
graph
project
or
other
things
related
to
it
around
supply
chain
or
some
of
the
opportunities
to
work
with
tag,
security
or
any
of
the
other
stuff
we've
mentioned,
please
do
reach
out
on
slack
or
or
directly
but
yeah.
That's
all
I've
got.
Thank
you.
Yeah.
C
D
D
We
are
working
on
an
official
otep,
so
basically
an
official
proposal
sort
of
like
clarifying
the
high
level
vision
and
then
sharing
that
with
everyone
to
make
sure
that
both
the
community
is
aware
of
the
efforts
that
we're
doing
and
agree
that
it
is
a
you
know:
worthwhile
endeavor,
before
we
kind
of
nail
down
more
of
the
specifics,
so
that's
kind
of
what
we're
focusing
on
right.
Now
we
got
a
second
tc
member
to
sort
of
oversee
things.
Josh
sewer
with
swears.
C
Jogged
my
memory,
there
was
one
thing
in
the
toc
related
to
this
when
talking
about
you're,
just
saying
hey,
there's
this
thing
going
on,
I
do
strongly
feel
that
we
should
we
should
once
we
have
that
otep,
that's
just
the
the
initial
kind
of
vision
or
as
early
as
possible.
We
should
be
engaging
with
communities
that
are
open
source
communities,
but
perhaps
not
in
the
cncf
proper
or
perhaps
during
a
different
linux
foundation.
Landscape.
C
You
know,
like
you,
know,
edge
hardware
manufacturers
that
are
running
socs
to
run
kubernetes,
you
know,
or
or
other
other
sorts
of
mobile
mobile
clients
or
mobile
scenarios
around
all
manner
of
stuff.
So
you
know,
I
think
that
that's
something
that
the
tag
can
facilitate
so
again,
there's
only
a
handful
of
people
here,
but
there
are
hundreds,
hundreds
of
people
on
our
slack
channel
and
and
people
that
download
the
youtube
videos
and
such
so.
C
If
you
know
someone
in
your
professional
network
or
or
of
a
community
or
group,
that
is
also
interested
in
this
topic
around
a
standardized
format
for
profiling
as
a
and
ancillary
topics
or
whatever
their
feedback
would
be
great.
For
example,
I've
also
reached
out
to
tag
security
about
this,
in
particular,
there's
a
link
in
the
in
the
slide
from
the
toc,
but
you
know
continuous
profiling
from
from
systemic
settings
like
cluster
wide
or
or
even
namespace
wide.
That
forms
a
very
interesting
signal
for
threat
assessment.
C
C
I
would
imagine
to
to
get
feedback
from
them
on
what
their
use
cases
are
and
what's
important
to
them,
because
it's
probably
different
than
what's
important
to
the
folks
that
are
already
engaged
in
the
conversation,
and
I
think
that's
a
good
thing
right.
We
can
get
all
of
these
different
scenarios
together
from
that
from
the
outset,
I
think
we
have
a
much
better
longer
term
chance
of
really
getting
broad
consensus
and
support
for
hotel
and-
and
this.