►
From YouTube: 2021-02-11 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).
A
A
B
I
don't
know
why
lolita
actually
said
you
know,
there's
no
agenda
item.
I
can
ping
her
if
she
wants
to
come.
Maybe
she
has
something.
A
A
Yeah,
I
think
you
already
know
that
now,
the
upstream,
the
open
telemetry,
is
going
to
release
1.0
for
some
language
sdk.
But
I
think
let
me
check
my
notes,
but
the
javascript
and
the
go
long
would
be
delay.
A
C
A
So
on
our
side,
our
work
plan
is
at
the
end
of
this
month.
We
would
release
aws
distro
for
donate
java.
A
And
from
the
beginning
of
march,
we
will
fully
focus
on
lambda
support
for
language
javascript
and
java
and
kaiser,
because
this
three
language
will
cover,
I
think,
more
than
70
of
lambda
users,
so
yeah.
So
now
we
are
waiting
a
lot
of
design
proposal.
I
know
the
every
detail
of
this
design.
Just
we
are
waiting
a
lot
to
show
this
stock
to
public
and
basically,
as
I
mentioned
before,
we
will
have
folders
for
collector
and
every
language
sdk.
The
structure
will
be
like
a
flag
and
the
something
new
from
my
side
is.
A
I
remember
in
last
meeting
alex
asked
a
question:
how
do
we
really
release
the
public
landauer
right?
So
at
that
time
my
answer
was
in
upstream
in
open
television
report.
We
don't
release
any
public
landmark,
but
the
public
london
layer
should
be
maintained
by
downstream
records.
For
example,
in
aws
we
have
aws
availability.
A
This
downscreen
vehicle
will
directly
focus
from
upstream,
let's
answer
in
last
segment,
but
after
that
I
had
some
discussion
with
my
team
with
a
laurita,
so
I
said
there
are
two
reasons
why
we
don't
want
to
release
lambda
in
upstream.
One
reason
is
because
the
size
of
a
collector
right,
because
we
cannot
create
a
full
collector
to
cover
every
user
case,
because
that
size
is
too
huge.
A
The
second
reason,
because,
as
the
we
want
to
replace
our
aws
version
of
collector,
which
is
aoc
so
my
proposal
here
is,
do
you
think
can
wait
for
our
aws
version
of
lavalier
called
upstream
I
mean:
can
we
publish
aoc
version
of
lambda
layer
in
open
telemetry.
A
B
A
A
Good
question
me:
are
you
here?
Yes,
I'm
here
yeah,
could
you
help
his
plan?
What's
the
difference
between
the
official
collector
and
the
aws
version
of
character,
why
would
you
wait
online.
D
Yeah,
I
think
most
of
people
already
know
that
one
benefits
I
can
see
to
doing
this
way
because,
right
now,
the
aw
digital
collector.
We
have
the
testing
framework
set
up
on
in
github.
That
is
ready
and
also
all
the
compo.
I
mean
all
the
components
that
we
supported
are
verified.
D
So
by
doing
data
we
can
migrate
our
existing
testing
framework
against
those
supported
components
directly
into
the
lambda
ripple,
so
which
means
whatever
we
are
supporting
in
the
lambda
layer
for
the
collector.
We
have
the
test
case
or
the
integration
test
covered
and
also
we
can
share,
because
our
integration
test
also.
D
Can
use
our
aws
resource,
so
we
can
probably
can
also
vent
a
account
there
for
the
community
to
run
the
test
directly
on
the
in
in
the
repo
so
yeah
a
few
of
those
things.
A
From
my
personal
perspective,
we
can
put
simply
we
can
see.
Aws
version
of
clapper
is
a
subset
of
a
facial
collector,
so
we
remove
some
of
component
only
certified
component,
we
will
be,
they
will
be
added
into
aws
version
of
clicker,
so
the
size
of
aws
version
of
twitter
is
is
smaller
than
original
version,
and
we
added
some.
We
have
the
icd
workflow
to
verify
to
run
the
enter
and
test
for
this
component.
A
B
Yeah
yeah
yeah.
I
see
that
I
I'm
my
question
was
a
slightly
a
bit
of
a
different
question.
I
think
we
I
was
like
thinking
like,
since
it's
going
to
be
based
on
the
collector
that
we
test
and
verify,
wouldn't
it
make
sense
to
distribute
it
from
the
a
dot
repo
rather
than
like
from
a
repo
in
upstream.
A
Yeah,
we
can't
do
that.
The
thing
is
for
aws
collector.
We
maintain
a
school
module
file
right
in
this
school
method
file.
We
will
record
which
exact
component,
which
exact
version
we
are
using,
so
we
will
extract
this
blue
module
to
a
lambda
component
module.
So
for
us
it's
easy
to
reuse
this
dependency.
We
don't
have
to
add
this
dependency
separately
in
collector
layer
just
for
the
use,
since
this
this
component
already
verified
the
aoc
precisely
workflow,
which
actually
user.
E
Yeah,
I
don't
think
it's
weird
for
amazon
to
maintain
amazon's,
specific
components
like
lambda,
I
mean
lambda
layers
and
things
like
that.
That
seems
pretty
straightforward,
though
you
do.
I
do
keep
hearing
that
there's
like
a
lot
of
like
collector
tests
and
like
security
stuff,
and
that
does
to
me
sound
like
things
that
should
probably
get
moved
upstream
if
possible.
A
B
There
are
like
some
actual
very
you
know,
difficulties
like
the
security
reviews,
for
example,
you
know,
has
like
pen
testing,
which
will
which
actually
requires
people
to
write,
like
you
know,
thread
models
and
like
security
team.
You
know
spending
some
like
time
and
you
know
we're
trying
to
like
do
it
with
you
know
releases.
B
It
cannot
be
a
continuous
thing.
So
that's
like
what
the
difficulty
is
like.
We
would
be
blocking
the
releases
of
upstream
if
we
don't
have
a
downstream
distribution,
so
I'm
still
not
super
sure
like
what's
the
best
way
to
go
like,
I
think
the
main
goal
was
not
to
you
know
slow
down
the
upstream
development,
although
you
know
all
the
security
findings
that
we
find
we're
contributing
them
to
upstream,
because
the
distribution
is
really
a
thin
packaging
like
it
doesn't
have
anything
different
than.
B
Components
so
it's
more
of
like
you
know
what
works
for
both
projects.
Sort
of
thing
like
we
don't
want
to
block.
You
know
upstream
releases.
A
D
Yeah,
I
think
that's
a
good
point.
Yeah.
That's
that's
probably
the
concern
I
also
have
so
if
we
want
to
make
aoc
as
the
official
version
to
support
the
nanda
yeah,
I
think
the
security
review
is
a
blocker.
I
would
say
it
is
a
blocker.
So.
E
Yeah
I
mean
I
just
don't
think
we
should
be.
I
wouldn't
consider
it
to
be
blocking
the
release
of
a
collector
to
be.
You
know
actually
like
testing
it
properly.
You
know
like
if
there's
something
I
would
rather
like
a
security
vulnerability
not
go
out
than
it
go
out
and
then
a
downstream
repo
find
it,
and
then
we
go
back
and
patch
it.
You
know
if
it's
slowing
it
down,
it
means
like
we
need
to
work
on
our
our
approach
to
releasing
the
collector.
E
If
it's
more
like
it's
would
be
slowed
down
because
of
some
like
bureaucratic
thing
around
like
managing
the
github
repos,
like
I
kind
of
wonder,
if
that's
maybe
part
of
it,
just
a
concern
that
if
it
went
upstream,
then
it
might
be
harder
to
do
just
because
there's
there's
more
groups
involved,
but
I
do
wonder:
if
that's
a
thing
we
could.
We
could
figure
out
because
yeah.
B
That's
a
good
point,
like
you
know,
the
organizational
side
of
it
is
not
a
big
problem
because
we
still
have
to
rely
on
upstream
right,
like
if
there's
any
issues
that
anybody
flags
we
still
need
to
like
merge
it
upstream.
So
it
would
be
nice
to
actually
like
be
a
part
of
the
upstream
release
cycle.
So
we
can,
you
know,
collaborate,
and
I
think
it
will
be
much
faster
for
us
to
be
able
to,
you
know,
contribute
some
of
the
changes
as
we
see
is
coming
by
the
way.
G
A
We
are
talking
about
if
moving
our
current
aws
lambda
repo
pull
upstream,
I
mean
fully
module
there.
We
want
to
publish
public
lambda
layer,
okay,
it's
a
it's
a
little
bit,
weird
that
it's
like
a
circle
you
see.
Clamper
is
the
upstream
of
our
aoc
right
and
everything
becomes
upstream
of
open,
telemetry
lambda.
So
it's.
C
A
A
circle,
so
this
is
a
little
bit
of
weird,
but
it
it
brings
some
benefit
for
us.
For
example,
the
lambda
layer
readability
is
good.
User
can
find
this
layer
directly
in
open
telemetry,
community
and
yeah
the
size
of
collector.
The
aws
layer
is
small
because
we
remove
lots
of
useless
components
so
currently
the
size
of
collector
layer,
I
mean
the
ws
version
is
less
than
39.
D
But
if
I
I
just
my
two
cents,
if
we
are
using,
if
we're
going
to
move
the
you
know
the
aw
digital
collector
to
upstream,
and
that
will
be
our
official
release
to
support,
because
nematode
is
especially
for
aws
right.
So
if
there
is
a
channel
for
us
to
use
that
channel
to
support
the
official
or
official
version
of
aws
digital
of
nanda
support,
I
think
the
property
security
is
a
lot
of
problem
right
anyway.
D
G
D
The
the
aw
digital
collector
to
directly
into
the
lambda
the
oxygen
lambda
ripple.
So
I'm
saying
if,
if
it
is
lambda,
is
it
other
stuff?
If
we
we
have
the
responsibility,
because
once
we
release
that
through
the
repo,
it's
mostly
for
the
adopted
customer,
so
we
need
to
make
sure
it
is
pat.
It
passed.
The
security
review
passed
all
the
tests.
Yes,.
G
G
G
You
know,
do
all
the
development
and
testing
in
obviously
in
hotel
right,
because
we,
even
if
the
layer
and
the
technology
you
know
the
layers
built
in
the
technology
is
aws,
but
still
you
know,
the
development
can
all
be
done
in
hotel,
right
and
and
the
testing
and
any
kind
of
integration
tests
that
we
need
to
run
etc.
But,
given
that
you
know
this
gets
bundled
into
the
destroy
aws
distro,
you
know
downstream.
G
If
we
still
as
menacing,
you
know
need
to
run
through
our
security
reviews,
our
performance
testing
and
any
other
guarantees
that
we
need
to
provide
right.
So
is
there
you
know,
I
mean
again,
I'm
trying
to
understand.
Is
there
a
conflict
in
in
that
model
or
jana?
I
know
you
were
mentioning
and
you
know
earlier:
do
you
see
any
issues
there.
B
No,
what
we
were
discussing
is
like
can
we
make
security
perform
some
of
this
confortance
test,
a
part
of
upstream
releases?
Would
that
make
more
sense
or
would
it
make
sense
to
keep
it
as
a
part
of
the
distro?
That
was
what
the
conversation
was.
B
There
are
some
pros
and
cons.
Do
you
know
both,
but
you
know
like
given
the
lambda
is
going
to
rely
on
the
our
distribution
collector.
We
should
we
can't
really
distribute
it
from
the
upstream.
That's
that's
sort
of.
G
Like
my
take
currently
I
mean
in
the
way
that
it
is
built
right
now
and
structured
right
now.
I
don't
think
we
can
actually
put
out
a
distribution
from
hotel
right.
I
mean
again
just
given
the
dependencies
that
we
have
and
and
and
the
to
your
point.
You
know
that
the
security
or
some
of
the
performance,
testing
or
integration
testing
should
be
done
in
hotel.
G
Yes,
of
course,
you
know
many
much
of
that
can
be
integrated,
and
we
certainly
can
do
that
and
intend
to
that's
why
we
want
to
you
know,
house
the
code
there
so
that
we
can
actually
start
doing
that
in
a
way
that
you
know
is
is
usable
by
the
project
and
at
the
same
time,
if
there's
anything
additional
that
we
need
to
do
for
the
meeting
the
requirements
aws
has
or
you
know,
security
team
has.
Then
we
can
do
that
downstream.
In
addition,
right.
E
E
It
seems
like
a
bad
cycle
to
to
release
a
security,
vulnerability
and
then
consume
that
downstream
and
then
absolutely
and
it
I
imagine
for
a
lot
of
this
stuff
it
it
would
be
cool
to
just
have
it
be
like
I
assume
most
of
this
is
black
box
testing
against
the
binary,
which
means
it's
sort
of
like
you
machine
out,
you
build
a
collector
binary,
and
then
you
run
it
against
this
test
and
this
test
suite
and
should
be
able
to
do
that
with
you
know
the
official
collector
that
has
the
whole
kitchen
sink
of
like
everything,
bundled
in
or
like
more
streamlined
ones
like
like
the
aws
distro,
for
example,
where
there's
a
bunch
of
stuff
you
know
you're,
not
gonna,
compile
in
or
then
the
the
and
then
a
step
past.
E
That
might
be
a
lambda
one.
That's
that's
stripped
down
even
further.
The
reason
I
I
bring
that
up
is,
for
example,
you
know
like
you
might
want
the
splunk
plug-in
if
you're
talking
to
splunk.
G
We
do
right,
but
I
mean
the
point:
is
that
and-
and
you
know,
every
distribution
will
adapt
over
time,
even
if
it
is
certified,
you
know
with
the
project.
It
will
adapt
over
time
to
the
specific
use
cases
that
it
is
serving
right,
because
again,
it's
completely
driven
by
what
is
useful
and
required
and
requested
by
you
know
its
users
and
and
that's
where
it
will
adapt
to.
G
But
that
said
there
are,
you
know,
a
basic
set
of
tests,
whether
that
integration
performance
security,
you
know,
and
just
the
overall
hardening,
if
you
will
of
what
is
being
built
that
can
be
incorporated
into
the
project
and
then
and-
and
then
you
know
in
anything
in
addition
is-
is
above
and
beyond
right,
and
that
goes
for
I
mean
that's
the
way
I
I'm
I
my
understanding
is
and
again
jana.
Maybe
you
can
give
us
some
insight.
G
B
B
So
we
can
still
do
these
things
in
parallel
to
make
sure
that,
like
you
know
upstream
releases,
are
going
well,
that's
something,
but
I
I
think
we
will
need
time
to
think
about
it,
because
you
know
the
the
way
that
everything
works
in
terms
of
like
going
through
the
process
of
like
having
these
reviews
and
so
on.
It's
such
a
big
couple
of
like
it
takes
weeks,
and
you
know.
C
B
A
lot
of
work
like
I
was
actually
very
surprised,
so
you
know
like
that
process.
I
think
shouldn't
really
affect
the
upstream
work,
but
yeah.
G
Exactly
and
and
ted,
actually
that's
the
intent
that
we
have
here
that
we
that's
why
we
are
proposing
that
you
know
all
the
language
implementations
for
the
lambda
layer
as
well
as
any
addition
additional
features
we
developed.
G
You
know
in
the
hotel
under
the
hotel
framework,
so
that
automatically,
then
that
you
know,
gives
us
the
reasons
and
and
the
platform
to
be
able,
to
you
know,
add
the
integration
tests
and
and
and
and
again
the
you
know,
just
going
back
to
the
overall
hotel
vision
of
you
know
providing
operational
guarantees
over
time.
Security
and
performance
are
very
big
parts
of
that
right.
So
we
do
want
to
use
this
opportunity
to
kind
of
build
that
out,
for
you
know,
components
vet
that
out
what
works.
G
What
doesn't
work
you
know
I
mean:
how
do
we
provide
a
more
well-thought-out
security
vulnerabilities
reporting
structure
in
the
project?
This
could
be,
you
know
and
beachhead
for
that
and
and
then
you
know,
build
that
out
further
right,
because
there's
no
reason
for
us
to
not
be
able
to
report
pzeros
on
the
project.
There's
no
reason
for
us
not
to
be
able
to
flag
any
kind
of
and
and
provide
any
patches
that
are
being
you
know,
available
for
the
core
components
or
the
dependencies.
E
Right
totally
yeah
I
mean
that
makes
sense.
Yeah,
I'm
not
proposing
that
the
entire
internal
awg
process
needs
to
be
moved
down.
No,
I
mean,
but
we
we
do
need
to
make
the
as
long
as
we're
on
the
topic
of
the
collector
like,
I
do
think
we
need
to
make
that
that
collector
test
and
build
pipeline
more
robust,
not
just
moving.
E
You
know
getting
more
performance,
testing
and
vulnerability
testing
other
things
up
there,
but
something
lei.
You
had
mentioned
earlier
around
like
a
build
pipeline
that
lets
you
pick
what
what
plugins
were
installed
essentially
seems
like
something:
that's
that
would
be
interesting
to
people.
E
One
thing
we're
starting
to
see
like,
for
example,
I
mean
I
don't
want
this
to
be
a
problem,
but
one
thing
we
do
see
is
plugins
getting
out
of
date
or
ceasing
to
be
supported,
for
I
think,
like
the
stackdriver
plug-in
was,
was
causing
problems
as
well
as
the
map
yeah,
and
that's
like
a
problem
where
that's
that's
like
the
opposite
issue:
it's
not
we're
putting
too
much
attention
on
this.
So
it's
taking
a
long
time.
It's
like
no
one's
putting
attention
on
it.
E
B
B
I
actually
suggested
that
thing
you
know
in
the
collective
sig
meeting
today.
E
I
think
that'd
be
awesome.
Actually
I
mean
I
mean
like.
I
think
that
would
honestly
be
a
great
marketing
tool
to
have
something
like
that
on
the
website,
but.
E
Then,
just
this
practical
thing
of,
like
I
mean
we're,
gonna
see
this.
I
think
in
all
the
repos
of
like
as
we
get
more
and
more
plug-ins
and
instrumentation.
All
of
that,
like
we're,
gonna
try
to
maintain
all
of
it,
but
the
idea
that
if
one
little
corner
isn't
maintained
the
whole
release
gets
halted
is
something
we're
gonna
have
to
we're.
Gonna
have
to
solve
this
problem
like
across
open
telemetry,
because
everything's
going
to
be
affected
by
it,
not
just
oh.
G
Absolutely
I
mean
we're
just
kind
of
making
dudes.
No,
there
is
no
way
to
scale
this.
I
mean
as
jana
brought
up
even
earlier,
and
the
collectors
think
today
we
need
to
be
able
to
do
development
for
the
prometheus
components
and
don't
want
to
necessarily
you
know,
have
merge
conflicts
later.
So
it's.
G
E
I
was
going
to
say
the
most
basic
thing
I
can
think
of.
There
is
sort
of,
like
you
know,
a
two-stage
build
of
the
collector
where
stage
one
is
you
build
a
collector
with
just
each
contrib
component
and
then
that
thing
has
to
build
properly
and
pass
its
component
specific
tests,
and
if,
if
that
happens,
it's
sort
of
like
all
of
the
ones
that
pass
that
round
of
ci.
E
That
happens
in
parallel,
just
the
components
that
pass
then
get
machined
out
into
into
a
collector
build,
and
likewise
for,
like
maybe
python
contrib
or
like
other
places
where
it
gets
released,
where
what
happens
is
like
that
first
round
just
drops
things.
So
it's
not
that
you,
you
fail
the
whole
build.
It's
that
you
get
a
build.
That's
that's
missing
the
things
that
didn't
pass.
That
might
be
a
like
a
good
start.
G
And
I
think
you
know
one
of
the
discussions
we
had
also
just
going
back
to
the
ci.
You
know
even
fernanda
as
we
move
this.
G
The
this
you
know
component
into
hotel
is
that
we
want
to
have
incremental,
builds
every
every
day
and
this
came
up
in
the
prometheus
discussions,
also
where
we
want
to
have
daily,
builds
and-
and
you
know
how
do
we
move
towards
that,
because
we
do
not
really
have
that
structure
set
up
today
for
any
of
the
work
that
is
happening
across
the
project,
and
you
know
if
we
were
to
do
daily
builds,
then
it
actually
needs
a
whole
parallel
thinking
around
you
know,
that's
a
test
and
development
environment
that
build
developers
are
testing
with,
and
it
doesn't
necessarily
have
to
be
stable.
G
It's
you
know,
and
yet
it's
not
experimental
either
right.
So
it's
work
in
progress
and
and
how
do
we?
How
do
we
sustain
that
and
that's
super
useful
in
the
lambda
environment
too,
because
you
know
again,
there
might
be
different
parts
in
flight
at
the
same
time
which
we
want
to
test
against
right.
G
Yep,
absolutely
so
going
back
to
the
lambda,
you
know
repo.
We
had
suggested
a
monorepo
right
and
is
that
something
that
you're
still
proposing
yes
and
did
you?
You
saw
that
proposal
right
from.
G
But
ted
had
you
and
alex:
did
you
guys
have
a
chance
to
look
at
that?
A
A
Yeah,
so
a
logical
help
creates
this
issue
in
community.
Basically,
we
want
to
create
a
land
variable
yeah.
It's
already
here
right,
alex
help
us
rename
the
existing
lambda
collectors,
and
under
this
repo
we
would
have
a
very
flat
structure
so
for
each
folder
it
will
support
functionality.
For
example,
we
have
a
folder
for
collector
layer.
A
If
customer
just
want
to
use
open,
parameter
connector,
they
just
need
this
folder
interface
folder.
We
will
provide
a
one
click
script,
so
user
can
build
their
custom
layer
on
demand,
for
example,
if
they
just
want
a
spelunk
as
exporter,
they
can
build
their
smaller
and
also
we
would
provide
public
lambda
layer
here
in
release.
Notes,
as
I
mentioned
before,
this
layer
will
based
on
aws
version
of
collector,
and
the
remaining
folder
is
for
language
sdk.
A
A
G
Yeah,
the
the
reason
why
we,
you
know
wanted
to
kind
of
clearly
call
this
out
in
my
discussions
with
lei.
Again,
we
kind
of
thought
through
whether
it
was
you
know
useful
to
have
a
monorepo
or
request.
You
know
language
implementations
to
be
housed
and
for
every
language,
and
we
kind
of
and
alex
I'd
really
like
to
get
your
you
know.
G
Thoughts
here
is
that,
given
you
have
you
have
an
existing
repo,
would
it
make
sense
to
actually
reuse
that
same
repo
or
or
you
know,
take
a
more
scatter
shot
approach
of
embedding
the
lambda
implementation
and
contrib
for
every
language
I
mean
initially.
C
C
Would
go
with
the
monorepo,
as
as
that
discussion
progressed,
I
did
rename
the
repo
to
open
telemetry
lambda
and
I
plan
on
moving
all
of
the
current
code.
That's
in
there
into
the
I
guess
the
collector
directory,
since
that's
really
what
this
will
equate.
A
C
A
G
Right,
that's
what
alex
was
saying
that
you
know
you
guys
decided
on
the
monorepo
which
would
work
well,
okay,
alexa!
I
mean
that's
sorry.
I
didn't
have
that
context.
So
thanks
for
looping
me
in
and
yeah,
would
you
then
move
the
implementation
that
lightstep
has
done
into
the
collector?
G
You
know
folder
and
then
the
python
you
know
implementation
that
we
have
done.
We
would
move
in.
You
know,
move
it
there
and.
A
Yeah.
Sorry
because,
as
we
said,
we
want
to
publish
aws
version
of
collector.
G
G
So
I
mean
that's
where
I
want
to
better
understand
from
min
what
the
circuit,
if
there
is
a
circular
dependency
here
and
if
we
actually,
if
that
makes
sense
or
not,
because
again,
I'm
assuming
that
this
is
the
lambda
implementation
for
the
collector
and
nothing
else.
Right.
D
Yeah,
so
so,
as
we
already
know
for
for
running
the
collector
in
the
lambda
layer,
there's
a
lot
of
restriction
there.
First
of
all
the
you
know,
for
example,
we
need
to
limit
the
binary
size
I
unlock.
If
we
are
just
directly
using
the
upstream
one,
I
I
don't
think
it's
gonna
work
has
provided
the
customer,
a
very
good
experience.
That's
the
important
one.
The
point
two
is
so
so
we
already
have
most
of
the
common
components
supported
in
the
aws
digital
of
the
you
know,
collector
yeah.
D
For
the
you
know
the
lambda
release?
I
mean
yes,.
D
That
would
be
the
hotel,
collector
yeah.
I
think
those
are
the
my
rational,
my
my
points
of
using
this
collector
but
yeah.
I.
G
D
Yeah,
it
will
be
only
the
name
that
enable
the
components,
for
example
yeah.
So
so
we
need
to
yeah.
It
doesn't
support
all
the
components
right.
G
E
Right,
this
is
just
you,
you
have
your
own
build
pipeline
where
you
build
yeah
yeah
exactly
and
you
have
your
aws
collector
and
then
periodically
you,
you
update
the
lambda
repo,
which
is
what
the
latest
collector
after
it
goes
through.
The
builds
right.
It's
not.
C
F
D
Yeah,
I
don't
the
only
thing.
I
just
want
to
confirm
that
so
like
we,
so
we
only
in
the
current
aws
digital
collector.
We
have
really
limited
components
right.
D
D
They
have
more
needs
than
what
we
are
providing.
They
probably
need
to
build
their
own.
You
know
layers
instead
of
the
layers
we
publish
from
this
repo,
so
that
that's
my
you
know,
yeah.
G
D
Yeah
they
need
to
build
their
own,
so
basically
they
need
to
replace
these
imports
and
do
their
own
way
to
to
to
build
the
layers.
A
We
can
we
can
provide
another
script
to
let
customers
use
collector
builder
to
build
their
customers.
Yeah
yeah,
that's
another
thread.
Yeah.
E
E
You
know
a
stripped-down
collector
to
have
this
hosted
in
open
telemetry,
but
have
the
aws
distro
collector
be
part
of
it,
because
it's
I
mean
it's
it's
it's
not
like
a
branding
thing,
that's
just
how
we
want
the
lambda
layer
to
work
yeah
and
as
long
as
there's
like
a
build
script
included,
so
that
if
someone
you
know,
for
whatever
reason
wants
to
to
include
a
different
collector,
for
example,
they
might
want
to
strip
it
down
even
further
or
maybe
they've
got.
You
know
some
kind
of
span
processor
plug-in
that
they
want
to
run.
E
G
G
E
G
A
Yeah
next
for
for
language
sdk
take
an
example
of
python,
so
I
will
move
all
of
this
folder.
This
is
our
python
implementation.
I
was
it's
called
tokyo
for
this
folder.
Basically,
I
have
a.
I
have
a
one
click
script.
This
script
will
help
user
to
draw
a
send
command
to
build
the
layer
from
scratch
and
after
we
provide
a
public
lambda
layer
in
probably
it
may
only
not
may
in
march,
or
in
april
we
will
provide
public
layer
at
that
time.
A
G
G
So
alex,
are
you
what,
where?
What
is
your
plan
there
in
terms
of
ex
I
mean?
Are
you
guys,
looking
at
any
other
language
implementations
right
now,.
A
Oh
sorry,
a
lot
of
times
I
think
hollow
follow,
does
not
attend
the
meeting
today
he
is
working
on
donate,
okay,
okay,.
C
Yeah,
from
from
our
side,
we
haven't,
haven't,
had
any
any
cycles
to
spend
on
any
lambda
stuff.
Since.
C
Work
that
was
done
last
december,
but
I
think,
as
some
of
the
work
that
we're
doing
in
the
different
things,
kind
of
comes
to
an
end,
we'll
be
able
to
go
back
to
it,
but
we'll
see.
I
think
this
might
take
you
know,
maybe
in
march
we'll
have
more
time
to
do
this,
but.
G
C
G
Weeks
and
then
we
can,
what
I'd
like
to
kind
of
figure
out
is
if
we
can
maintain
and
roadmap
like
a
project
board
or
something
on
the
on
the
repo,
so
that
we
can
kind
of
build
and
track
the
different
language
implementations.
C
I
mean
it
might,
it
might
be
useful
to
create
just
just
to
create
issues
in
the
repo.
C
Can
we
create
an
issue
in
here
to
track
the
the
freezing
unfreezing
event
for.
E
E
Exactly
I
was
wondering
like
I
was
trying
to
figure
out.
I
was
like
okay,
where,
if
you
were
to
fix
this,
where
would
you
go,
and
I
was
curious
if
you
happen
to
know,
is
this
like
something
you
do
with
the
firecracker
player,
or
is
it
actually
a
shortcoming
in
like
kvm
itself,
or
is
it
something
to
do
with
how
you
would
have
to
do
it
all
kind
of
in
user
land?
E
So
your
you
know,
the
the
guest
operating
system
would
have
to
be
aware
of
some
kind
of
signal
being
sent
to
it.
G
Have
you
lei,
have
you
looked
at
firecracker
in
in
no,
I
thought
it
was
all
user
layer
based.
E
G
C
E
It
looks
really
fun
as
someone
who's
like
was
working
on
like
containers
and
scheduling.
It's
like
my
last
thing
like
this
is
really
cool,
but
just
realizing
that
I
mean
there
is
some
amount
of
being
able
to
signal
when,
before
a
a
guest
is
going
to
be
shut
down
right.
That
seems
like
a
signal
that
that
you
get
but
yeah.
E
I
don't
know
which
one
it
is,
but
it's
probably
suspend
or
pause
happens
on
the
on
the
guest
and
yeah
I
I
dug
in,
and
I
could
not
find
anything
that
looked
like
like
a
signal
that
gets
sent
when
that
happens,
because
it
would
have
to
be
like
a
signal
and
then
some
kind
of
waiting,
but
it
doesn't
look
like
there's
because
it
there's
a
linux
concept
of
like
being
shut
down,
but
there
is
no
linux
concept
of
being
paused.
A
Yeah,
but
I
I
believe
that
the
lava
is
using,
say,
sql
freezer
the
design
concept
of
single
freezer
is.
It
does
not
want
to
send
any
signal,
because
it
believes
that
this
is
the
top
priority.
If
they
want
to
freeze
the
environment,
they
don't
want
to
wait
any
response
of
signal,
so
it's
thinking
we
should
not
send
any
signal.
That's
the
reason
why
we
cannot
receive
think
about
it
in
another
way.
If
lambda
want
to
freeze
your
environment,
why
lambda
want
to
get
confirmed
that
I'm
ready
for
b
frozen.
E
Well,
I
mean
for
for
this
reason
right,
which
is
when
you
go
to
freeze
one
of
these
things.
It
might
want
to
do
a
certain
amount
of
cleanup,
because
it's
not
it's
not
the
kind
of
hypervisor
that
that
can
you
know
it's
not
doing
like
managing
the
network
connections,
you're
literally
just
turning
it
off
and
bringing
it
back
later.
E
I
just
I'm,
I'm
just
you
know
running
this
black
box
guest
and
I
just
want
to
pause
it
and
what
you're
doing
with
lambda
is
actually
like
a
little
more
of
like
no
one.
I
don't
think
anyone
was
thinking
you
would
be
hitting
freeze
or
pause
on
a
virtual
machine
when
it
was
running
in
production.
E
E
They
would
be
like
what
are
you
talking
about,
but
but
if
you're
building
something
like
lambda,
it
makes
perfect
sense
to
do
that,
which
is
you
know
you,
because
you're
saying
I'm
going
to
shut
this
down
for
a
long
time
and
then
I'm
going
to
bring
it
back
up,
and
so
that's
the
savings
but
like
losing
like
a
second
on
shutdown,
isn't
necessarily
like
a
huge
deal.
I
mean,
maybe
if
you're
amazon,
that
that
does
add
up
to
like
a
lot
of
money
losing
a
second
there.
E
But
but
I
think
that's
that's
the
issue
right.
Is
that
yeah
it?
It
seems
like
all
the
way
down
at
the
kvm
level.
It
doesn't
work
like
that.
G
So-
and
I
mean
going
back
to
the
the
freezing
issue,
though
did
you
I
mean
please
file
an
issue
for
sure,
because
we
are
also
seeing
the
same
problem:
layered
flagged
that
and
and
we'll
we'll
figure
out,
you
might
have
to
experiment
a
bit
with
the
firecracker
implementation.
Also
to
see
if
we
can.
You
know
if
that's
an
alternative,
that
that
can
be
recommended,
but
we
have
to.
A
A
And
could
you
share
some
design
talk.
G
G
A
G
E
G
This
this
is
a
different
one.
We
also
have
an
internal
design
dock
that
we
have,
you
know,
want
to
kind
of
publish
and
just
make
it
available
for
everyone
to
kind
of
comment
on
and
see.
You
know
what
are
some
of
the.
What
are
the
some
of
the
findings
that
lay
had
and
also
you
know,
just
get
your
feedback
on.
A
E
G
E
G
Yeah,
we'll
just
put
it
on
an
issue,
and
then
we
can.
We
can
just
review
it
in
the
you
know,
please
feel
free
to
add
comments,
and-
and
was
there
anything
else
late
that
you
wanted
to
cover
today?
I
think
we're
almost
that
time.
G
Okay,
cool,
so
ted.
I
mean
I
just
wanted
to
take
two
minutes
of
your
time.
Min
did
you
have
any
other
questions?
Oh
you're,
good.
Okay,
awesome,
ted.
I
had
put
some
comments
on
your
on
your
blog
blog
post,
oh
yeah,
so
again
there's
your
suggestions.
Please
just
take
a
look
at
them.
E
You
thank
you
so
much
for
commenting
on
it.
No.
G
Not
totally
glad
to
be
able
to
help
so
thanks
again
guys
I'll
drop
off.
I
have
a
meeting
in
for
a
minute.