►
From YouTube: 2021-05-20 Governance Committee private meeting
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
B
Hey
just
as
an
fyi,
I
have
to
switch
to
another
meeting
ten
minutes
past
so.
C
Well,
we
haven't
closed
the
transaction
yet
so
the
new
gig
is
basically
just
the
old
gig,
at
least
for
another
month,
but
it
seems
good.
I
think
I
mean
I
don't
know
this
is
a
recorded
meeting
but
genuinely
they're
like
trying
to
do
it
correctly
and
retain
what
makes
lifestyle
good
and
stuff
like
that.
So
I
feel
very
like
thankful
for
that
piece.
D
Finally
enough,
then,
I
think
I'm
coming
by
your
all
hands
tomorrow
to
fireside
chat
with
austin
mcdeveral.
C
I
know
that's
awesome,
just
yeah,
I'm
really
excited
about
that
should
be
fun.
D
C
That
that's
what
they
tell
me
as
well,
but
I
can't
trust
them.
I'm
glad
that
you're
that
you've
heard
the
same
thing.
That's
good!
I'm
just
kidding!
You
know
it's
it's!
It's
really!
Nice
like.
I
think
it
takes
a
lot
of,
at
least
for
me
personally,
the
things
I
enjoy
the
most
and
makes
them
like
easier
to
do
so.
C
C
Yeah
well,
my
house,
I
mean
I'm
as
far
as
I
can
be
from
people
with
jackhammers
right
now.
My
foundation
has
a
crack
in
it
and
so
they're
making
a
lot
more
cracks
before
they
fix
the
one
that
was
there
and
my
house
is
physically
shaking,
but
apparently
zoom's
noise
cancellation
is
very
effective,
so
yeah.
D
I
don't,
I
think,
it's
some
kind
of
they're
like
drilling
into
the
foundation
yeah.
So
it's
like
drilling
concrete.
E
B
G
G
Well,
do
we
do
the
seismic
retrofit
or
not
we're
like
screw
it
we'll
do
it
so.
C
I
put
something
on
the
agenda.
It's
actually
related
to
the
second
item,
but
I
know
that
due
diligence
is
kind
of
wrapping
up,
but
do
we
really
need
to
do
an
annual
review
like
it
seems
like
that's
kind
of
duplicative
or
am
I?
Are
we
able
to
push
back
on
the
cncf?
We
got
that
email
yesterday
being
like
it's
time
for
openflow's
annual
review.
I'm
like
really
I
mean
come
on.
Can.
B
H
A
D
All
right
do
we.
B
D
With
regards
to
ted's
question
about
the
incubation
doc,
have
we
heard
from
the
cncf
as
to
what
the
next
steps
are
like
is
the
he
is
the
toc
going
to
vote
now,
or
does
that
or
did.
A
Yeah
I
mean,
I
think,
liz.
What
elena
had
said
was
that
if
we
can
get
all
our
changes
in
by
may
20th
or
thereabouts,
which
is
may
30th
yesterday,
then
they
will
actually
start
looking
at
all
the
additional
information
we
have
provided
and
then
by
next
week.
Hopefully
they
should
be
able,
to
conclude,
you
know
their
review
and
and
then
thereafter
line
up.
The
public
comments
period
opens
up
for
their
final.
A
E
G
B
After
that,
this
is
the
deal
I
think
I
see
yep
did
we
get
the
code
covered
badge
included
in
all
the
repositories.
A
I
my
engineers
is
working
on
it
right
now,
but
we've
filed,
you
know
prs,
but
I
think
it's
just
in
the
process
of
you
know
the
maintainers
adding
that
so
so,
at
least
you
know
I
mean
the
work
is
in
progress.
Hopefully
it'll
get
completed
in
the
next
couple
of.
A
I
Yeah,
so
we
we
discussed
it
on
the
tce
co.
I
think
carlos
was
gonna
start
looking
into
due
diligence
of
that,
but
yeah.
I
just
wanted
to
give
this
brief
update
on
the
receipt.
Is
there
any
concerns
for
people?
My
my
kind
of
one
question
that
I
I
wasn't
like
comfortable
asking
in
the
open
forum
is,
is
like
what
is
their
kind
of
motivation
for
donating.
E
Yeah
I
mean
I
can
speak
to
that
because
it's
from
splunk,
okay,
so
splunk
bought
flo
mill
that
last
year,
flo
mill
will
be
brought
into
what's
called
now
the
splunk
observability
cloud.
This
is
like
omniscient
signalfx.
It's
going
to
be
a
new
network
monitoring
pillar
of
that
on
splunk
observability
cloud.
We
want
all
of
our
instrumentation
to
be
just
native,
open
telemetry,
like
we've
many
customers
who
just
use
the
pure
oss
distribution
and
run
their
own
builds.
E
We
would
like
to
donate
this
to
open
telemetry
so
that
a
open
telemetry
can
capture
network
telemetry,
be
perhaps
long-term
open.
Telemetry
people
want
to
do
more
of
ebpf.
This
could
be
the
launching
ground
for
that
or
we
could
rediscuss
it
doesn't
really
matter
and
see
for
ourselves.
We
can
then
just
continue
to
tell
people
go,
use
the
bog
standard,
open,
telemetry
or
use
our
distribution
they're
effectively.
The
same.
I
E
A
Yeah,
we're
definitely
interested
in
adding
you
know
more
instrumentation
and
and
additional
capabilities
around
it.
So
yeah.
E
I
think,
like
long
term,
I
think
like
it's,
it's
really
neat
to
have
some
ebpf
collection
stuff
inside
the
project,
because
we
could
expand
it
beyond
network
telemetry
in
the
short
term.
I
think
we're
all
because
of
due
diligence
very,
very
sensitive
to
scope
creep,
and
hence
why
they've
positioned
it.
The
submission
is
just
another
receiver
for
the
collector
that
happens
to
use
ebpf,
but
it
just
captures
network
stuff
and
uses
the
exact
same
data
types
that
open
telemetry
has
today.
A
Yeah
and
and
the
other
issue
also
to
consider
morgan
is
again
what
kind
of
impact
it
will
have
on
metrics.
You
know
going
stable
because
there
will
be
some
additional.
You
know
integration
testing
as
well
as.
E
Extendability
they'll
definitely
be
like
integration
testing.
What
we've
told
them
is
like
go
use
the
existing
data
types.
If
the
metric
stacks
the
spec
snaps
to
something
else,
you
better
snap
to
it
too,.
I
And
is
it
gonna
be
sort
of
like
a
lot
of
friction
with
the
prometheus
ecosystem
of
node
exporters
because
they
sort
of
do
the
same
thing
right.
E
It's
a
good
question,
so
I
I
haven't
dug
into
too
much
into
their
individual.
Like
the
capabilities
of
this,
I
think
it
probably
does
more
than
you
can
get
just
from
prometheus,
but
that's
a
good
thing
for,
like
you
and
greens,.
A
A
A
You
know
also
participating
in
this
due
diligence.
What's
the
process,
do
you
know,
morgan
or
others
for
participating
in
that
due
diligence
for.
E
Flo
mill,
yeah,
probably
best
asked
yuri
I'm
not
on
the
tc
side.
I
We
don't
have
like
a
formal
process,
really,
I
think
it's
it's
like
the
only
time
we
did
it
so
far
is
with
the
login
donation,
the
tigran
yeah.
C
About
whether
it's
decoupled
from
the
spec
and
the
collector-
and
it
sounds
like
it
is-
which
is
great
news,
so
it's
not
going
to
create
like
drag,
but
then
it's
like
okay
cool.
So
why
is
it?
Why
does
it
need
to
be
donated
at
all
like?
Why
can't
splunk,
just
put
it
out
on
github.com
splunk,
slash,
email,
collector,
whatever.
C
Decoupled,
that's
great
yeah
seriously
great
and
I
hope
it
succeeds
and
it
should
because
it's
awesome
ebpf
is
great
but
like.
Why
are
we
aside
from
a
branding
standpoint?
Why
are
we
putting
in
hotel?
I
think
the
answer.
D
E
So
that
one
so
a
we
should
all
chat
together.
So
I
chatted
with
ishan
a
few
times
at
pixie
and
he
said
hey.
They
wanted
to
pursue
their
own
oss
project
like,
but
let
me
clarify,
they're
mostly
focused
on
open
sourcing
at
the
time,
their
back
end,
rather
than
the
instrumentation,
like
the
big
focus,
was
on
on
open
sourcing
pixies.
E
So
people
can
use
end
to
end
pixy,
not
just
the
ebpf
part,
which
is
really
cool
and
that
their
second
priority
after
that
would
then
be
some
kind
of
convergence
with
open
telemetry
with
their
instrumentation.
His
suggestion
to
us
was
that
ishan
was
not
super
wedded
to
the
way
that
they
do
their
instrumentation
necessarily
and
his
he
said
basically
for
us
like
go
make
your
submission
to
open
telemetry.
E
If
you
want
to
pursue
that
and
we'll
figure
out
how
to
how
to
make
this
work
with
pixie
and
make
this
all
one
later,
yeah
we're
meeting
again
with
him
later.
I
think
in
a
week
or
two
to
sort
of
take
the
next
steps
on
that.
But
but
we
had
given
him
like
a
heads
up
in
february
that
we
were
going
to
do
this
and
he
was
like
yeah
go
ahead.
Cool.
A
Yeah
I
mean
I'd
like
to
add
to
that,
so
I
mean
we've
been
evaluating
both
pixie
and
and
flo
mill
and
and
and
again
ben
answering
your
questions.
Specifically,
I
mean
there
is
a
strategic
value
in
open,
telemetry
supporting
ebpf.
You
know
natively
on
the
project
itself,
specifically
the
collector
and
and
the
component
you
know
being
integrated
as
an
receiver
on
the
collector
itself.
Again,
I
don't
think
it's
scope
creep
because
it
actually
adds
value
to
the
collector
itself.
A
H
A
With
splunk's
team,
I
do
think
there
is
convergence.
You
know
in
the
in
the
future.
D
A
Addressing
the
you
know
why
not
host
downstream
versus
on
the
project?
Again,
let's
go
into
the
next
topic,
but
certainly
I
would
like
to
see
it
on
the
project.
E
D
Okay,
well,
let's
talk
about
bandwidth,
it
is.
It
was
not,
I
think,
previously
serviced
at
gc
that
it's
that
it's
not
that
the
maintainer
bayless
crisis
is
not
only
contribute,
but
that
there
are.
It's
also
impacting
work
in
the
main
repos.
A
Yeah
I
mean
I,
I
have
actually
been
working
closely
with
google's
team
and
microsoft's
team
to
figure
out
a
proposal
that
we
would
like
to
make
in
terms
of
you
know
not
only
alleviating
maintainer
deadlock,
but
also
working
closely
with
specifically,
you
know
the
collector
maintainers,
as
well
as
the
language
maintainers,
for
just
being
able
to
have
an
easy
ramp
up
process
and
for
other
you
know,
folks,
to
graduate
and
also
work
with
how
we
can
you
know,
scale
out
the
ability
for
more
contributors
to
contribute
to
the
project,
because
these
are
you
know
these
work
closely
together.
A
I
do
intend
to
make
a
proposal.
You
know.
Early
next
week
I've
been
working
on
that
so
I'll
submit
that
to
the
tc.
I
also
request
that
the
dc
consider
you
know,
formulating
a
small
work
group
which
is
not.
You
know
there
right
now
to
kind
of
focus
in
on
on
this
area
and
be
able
to
actually
help.
You
know
with
this
whole
process
of
not
only
expanding
the
maintainer
pool,
but
you
know
also
working
with
the
expansion
of
and
managing
baseline
infrastructure
for,
more
more
contributors
and
contributions
to
come.
E
And
and
for
contrib
there's
an
otep
that,
like
yuri
and
I
and
a
few
others,
have
reviewed
just
about
clarifying
that
we
could
have
like
contrib
specific
maintainers
for
each
language,
as
opposed
to
just
having
the
main
language.
Maintainers
do
both
the
main
part
and
contrib
yeah.
H
E
I
tend
to
agree
I'll
leave
it
like,
I
think,
having
more
maintainers
and
approvers
will
help
a
lot
as
well
as
as
just
managing
infrastructure
for
them
a
little
better.
Daniel,
probably.
D
D
I
think
the
main
problem
that
is
at
least
as
outlined
by
by
yana
and
and
by
folks
from
from
google,
is
that
you
know
it
is
even
if
approvers
are
approving
changes
quickly.
Even
if
there
are
enough
approvers
the
fact
that
a
maintainer
has
to
click
the
merge
button
right
is
is
a
problem
and
we
may
want
to
revert
to
what
we
were
doing
before,
which
is
that
approvers
can
click
the
merge
button,
but
a
maintainer
can
and
do
it
if
there's
a
problem
right
and
can
revoke
approval,
privileges,
yeah.
A
And
sloppy
I
mean
listen,
that's
what
we
intend
to
propose
and
and
again
you
know,
there
is
also
a
proposal
that
you
know
tigran
has
been
putting
together
which
which
I've
been
working
with
him
on.
So
I
do
intend
to
make
a
more
comprehensive
proposal
and
as
a
no
tap
and
then
we
can.
I
would
like
to
ask
for
our
group,
you
know
which
I
would
like
to
participate
in
on
the
tc
so
that
we.
D
A
A
C
H
C
Like
what's
the
right
way
to
ask
this,
I
mean:
is
it
going
to
ever
be
able
to
like
hook
into
the
trace
context,
or
is
it
just
going
to
be
emitting
like
kind
of
appearing
event?
You
know
what
I
mean.
E
This
is
the
very
first
thing
that
I
asked
them,
so
they
had
said
like
it
doesn't
today,
right
like
like
the
the
metrics
and
logs
and
other
things
that
it
generates
have
are
not
correlated
with
traces
at
all.
This
was
the
very
first
question
that
I
asked
them
upon
meeting
them.
They
believe,
and
I
don't
see
why
they
couldn't
that
they
can
parse
out
the
standards
like
w3c
or
zipkin
style,
like
like
the
variety
formats
like
trace,
ids
from
network
traffic
and
then
to
correlate
there.
I
I
wish.
C
E
E
C
H
C
E
Instrumentation
has
been
putting
this
into
like
first
off
just
open
sourcing
it
and
then,
and
then
putting
it
into
open
telemetry.
I
understand
that
this
is
like
their
next
big
thing
that
they
want
to
tackle.
E
It's
the
one
that
I've
been
pushing
on
because,
like
what
I
want
to
see
is
is
ours
and
everybody's
apm
products
is
when
you
have
a
span
actually
being
able
to
delineate
like
within
a
span
that
that
represents
like
a
single
service
processing,
something
the
actual
like
network
time
or
network
latency
is
part
of
that
yeah
and
to
me
this
is
the
probably
the
best
way
to
do
that.
I
I
do
wonder
if
you
bake
this
into
network
proxies
right
like
envoy
and
nginx
is.
E
I
think
that
would
probably
work
well,
but
for
for
applications
that
are
doing
their
own,
that
are
not
using
the
network
proxy
that
are
doing
their
own,
like
ssl,
then,
and
my
impression
perhaps
I'm
wrong.
But
my
impression
is
right
now
that
most
apps
do
their
own
ssl,
rather
than
not
doing
ssl
and
relying
on
envoy
to
do
it
for
them:
yeah,
yeah,
but
but
you're
right
like
that
that
that's
one
that
we
discussed
our
at
least
at
splunk.
E
Language
run
times,
maybe
at
some
point
well,
that's
that's
sort
of
what
I
was
I
was
getting
at,
but
that
requires
now
per
language
integrations,
but
that
seems
like
if
there
is
an
opportunity
to
do
it
cleaner,
that's
probably
the
way
to
do
it,
but
it's
just
going
to
require
like
language,
specific
code
paths,
right
yeah,
you
could
probably
get
it.
I
assume
this
thing's
written
in
c
plus,
so
you
could
probably
get
it
into
the
c
plus
plus
stuff
faster
than
anything.
I
Speaking
of
ebpf
collection,
I
think
another
area
that,
like
facebook
invested
a
lot
is,
is
combining
sort
of
like
collecting
all
kinds
of
performance
metrics
from
cpu
cycles.
Memory
allocations
pertaining
to
the
request,
and
those
are
pretty
hard
to
do
like
in
across
various
runtimes.
So
like
facebook
didn't
do
it
folder
on
times,
because
it
has
like
fairly
homogeneous
stuff
yeah.
But
that's
the
sort
of
the
thing
that.
H
A
And
kernel
performance
and
yeah,
so
there's
a
good
good.
You
know
complement
complementary
feature,
set
thing.
E
I
think
a
lot
of
it
is
done
by
giving
it
like
python
scripts.
I
I
don't
know
the
architecture
as
well,
but
so
I
don't
know
what
it
captures
out
of
the
box
versus
it's
just
highly
extensible
and
you
give
it
some
some
configuration
or
scripts
that
actually
tell
it
to
go.
Do
something
it's
highly
flexible,
whereas
the
the
flow
mill
implementation
is
much
more
opinionated
about
just
like
we're
going
to
capture
these
things
from
the
network.
A
H
A
For
for
the
evaluation,
because
there
are
complementary
feature
sets
and
again,
let
me
know
if
you
know
I
can
reach
out
to
you.
I
mean
work
with
morgan,
of
course,
to
get
through
mail
to
demo,
but
also
pixie
to
walk.