►
From YouTube: 2022-12-12 meeting
Description
Open Telemetry Meeting 1's Personal Meeting Room
C
A
A
A
Like
I
get
the
occasional
drop
spans,
The
Collector
was
not
healthy,
especially
on
my
eks
cluster.
When
I
started
throwing
the
demo
in
there
at
like
a
third
or
fourth
time,
it
got
worse
and
worse,
I
I
think
it
was
just
it's
actually
a
network
on
a
host
and
then
I
finally
realized
like
something
is
weird
anyhow.
D
E
C
C
C
C
F
B
B
B
E
D
A
great
warm-up,
so
I
put
together
kind
of
an
agenda
list
items.
If
anyone
else
has
any
other
items,
I
want
to
discussed,
please
just
go
ahead
and
add
them,
but
the
first
thing
I
wanted
to
call
out
was
this:
is
the
last
meeting
of
the
year
I
think
we
could
potentially
meet
next
week,
but
I'll
be
out
of
town,
so
I
guess
we
can
decide
if
we'd
like
to
meet
next
week,
but
my
kind
of
sense
was
that
the
holidays
would
probably
end
a
lot
of
things.
D
So
any
sick
thoughts
around
that
at
least.
D
D
Yeah
everyone's
gonna
be
coming
back,
no
point
yeah
and
rushing
for
that
and
then
I
think
we
discussed
it
a
bit
last
week
and
I
see.
Oscar
is
here
and
then
here
as
well,
so
we
have
I
kind
of
accumulating
pile
of
front-end
issues,
so
I
don't
think
we
need
to
take
care
of
it
before
the
holidays.
But
it's
something
to
kind
of
keep
in
mind
like.
D
Can
we
clean
up
some
of
these
things
and
get
the
front
end
in
a
better
State
I
know
when
you
add,
like
the
metrics
Juliano,
has
an
entire
list
of
kind
of
small
item
and
that
that
could
be
improved
and
there's
a
couple
bugs
as
well.
So
there's
a
lot
of
attention
probably
needs
to
be
directed
towards
the
front
end
in
the
first
month
or
two
of
the
new
year.
B
D
D
E
It
has
been
a
pretty
busy
week
at
racist,
so
yeah,
sorry
about
that.
But.
G
E
Yesterday,
I
started
looking
at
two
of
the
issues,
so
I
I
have
the
changes
ready
for
a
Carpio,
not
summing
sense,
so
that
that
should
be
a
shooting
up
here
today
and
I'm
about
to
actually
have
the
the
JavaScript
metrics
enabled
as
well.
So
that's
something
that
is
going
to
be
very
unappear,
probably
is
20
minutes
after
this
call,
and
after
that
I
don't
know
which
one
is
has
higher
priority
from
the
other,
the
rest
of
the
stuff.
D
D
G
D
And
Oscar
sorry
I
missed
the
first
PR
you're
gonna
open.
So
before
betrics.
What
was
the
other
item?
The.
E
First,
one
is
to
fix
a
bug
around
for
got.
D
D
She
had
gotten
the
demo
running
on
her
happy
tour,
which
I
think
is
like
a
specific
type
of
compute
or
something
like
that,
but
essentially
she
was
looking
to
just
make
it
available
to
the
public
and
some
sort
of
either
like
contrib,
folder
repo
or
something
like
that
for
the
demo,
of
course,
from
like
a
Sig
perspective,
I
don't
think
we
want
to
like
take
on
ownership
of
these
resources
necessarily,
but
at
the
same
time
it
would
be
really
cool
to
Spotlight
some
kind
of
interesting
things
you
can
do
with
the
demo
and
then
maybe,
if
we
have
some
sort
of
like
code
owners
or
like
roles
around
it,
we
could
deflect
a
lot
of
the
work
from
coming
into
the
demo
Sig
itself
so
curious
to
hear
some
thoughts
around
us
foreign.
F
Ly
would
like
to
have
vendor
stuff
in
a
country
demo
in
a
country
folder.
No,
no,
sorry,
not
folder.
In
a
country,
repo.
C
C
Vendor
specific
contribs
that
are
maintained
by
vendors
versus
and
then
it's
like.
Why
is
that
not
in
the
org
right
and
then
that
gets
into
weird
places,
but
I
also
feel,
like
you
know,
there's
here's
a
really
good
example,
like
it's
kind
of
silly,
that
we
don't
just
have
a
first
party
AWS
one
right
like
there's
enough
people
that
are
using
AWS
or
gcp
or
Azure
pick
your
Cloud
here,
but
like
it,
you
know
we
have
the
deployment
stuff
to
throw
this
into
eks.
C
It
would
probably
be
really
interesting
to
see
what
this
looks
like
in,
like
ECS
or
fargate
or
like
a
kubernetes
like
thing
that
is
more
conventional
with
AWS.
C
C
B
C
I
mean
there's
like
a
bunch
of
ways.
You
could
do
it,
but.
D
G
G
C
C
Another
example
would
be
if,
like
Pierre,
for
example,
has
a
bunch
of
deployment
stuff
to
get
this
running
on
AWS
right.
They
give
eks
scripts
or
whatnot,
or
maybe
cloud
formation,
I,
don't
know,
but.
G
A
Makes
sense,
I,
I
think
what
you
can
do?
Is
you
instill
standards
with
each
one
and
they
have
to
have
a
readme
and
you
have
to
have
works
with
and
you
know
tested
with.
I
should
say
where
the
person
who
is
in
the
code
owner
can
they
themselves
contest
it
with
the
latest
version
of
whatever
and
they
could
update
their
readme
to
say
we
tested
it
with
version
1.2
all
the
way
through
to
1.5,
whatever
that's
at
the
top
of
the
readme.
So
when
somebody
walks
in
they
could
recognize
like.
C
C
A
C
C
D
C
C
A
That's
what
you're
saying
is
I
should
guess
my
my
Helm
chart
for
AWS.
C
Well,
I
would
I
would
actually
say:
I
was
because
I
just
want
to
reinvent
about
this,
because
I
I
did
a
stream
there,
where
I
showed
the
hotel
demo
off.
C
But
you
know
it
was
just
pure
eks
and
someone
was
like,
oh
well.
Why
aren't
you
know?
Why
aren't
you
using
right,
the
the
drop
in
Amazon
replacements
for
some
of
these
things
so
like
I,
think
there's.
C
That
would
obviously
be
useful
to
people
that
are
in
that
ecosystem
and
if
people
on
this
in
the
sound
of
my
voice
have
like
an
interest
in
doing
that,
then
I
don't
see
a
reason
why
we
shouldn't
do
that
where
we
shouldn't
have
something
you
know
if
Pierre
or
me,
or
anyone
wants
to
go
out
and
say
like
oh
well:
here's
like
the
gcp
version
or
here's
the
AWS
version
or
here's
the
Nomad
version
in
Adriana's
case.
Like
cool
go
for
it,
you
know,
and
we
as
the
projects
say.
G
C
C
C
D
All
right
so
I
think
Gary
put
this
item.
I
moved
it
up.
Just
so,
he'd
make
sure
to
spend
some
time
on
it.
Essentially,
in
the
the
rust
service,
it
doesn't
look
like
there's
much.
D
We
can
do
by
replacing
the
components.
Bella
Gary
speak
to
those
specifically.
B
Yeah,
so
the
biggest
chunk
that
I
kept
noticing
was
that
the
compilation
I
mean
we've
talked
about
how
we
can
optimize
times
for
all
of
these
longer
running
and
longer.
Compiling
services
and
I
noticed
that
the
updating
crates.io
index
took
the
majority
of
the
time
for
compilation
like
dramatically
more
than
compiling
any
of
the
crates,
and
when
we
take
out
specific
crates
like
I
know,
we
had
spoken
in
a
previous
session
about
grpc
to
http.
B
What
kind
of
Advantage
would
we
get
zero
because
the
grpc
implementation
is
necessary
for
the
open
Telemetry
collector
to
actually
receive
the
stuff
that
we
get
so
for
a
lot
of
technical
reasons
that
I
don't
have
to
get
into
in
the
summary
I
I.
Don't
think
that
removing
dependencies
or
changing
the
implementation
is
going
to
dramatically
increase
the
compilation
time.
B
The
way
that
we
might
be
able
to
do
it
is
by
creating
a
cache
index
in
the
docker
image,
but
that's
like
a
very
different
problem
than
the
one
that
we
started
with
so
I
wanted
to
bring
it
it
to
the
group
before
I
started
like
thinking
about
that
and
talking
about
how
that
might
be
possible.
I
went
down
a
couple
of
rabbit
holes
before
I
came
to
this
conclusion.
B
One
of
the
biggest
ones
that
any
any
rust
station
will
probably
be
aware
of
is
that
you
can
actually
have
it
skip,
checking
the
package
index
when
you're
doing
this.
But
when
you
do
that
it
can't
find
any
of
the
packages.
So
we
can't
skip
the
package
index
because
there
is
no
package
index
in
the
docker
image.
B
That
was
basically
what
I
wanted
to
get
across.
And
what
I
wanted
to
just
chat
about
in
this
group
open
to
thoughts
and
suggestions
and
feelings.
A
B
D
A
C
C
Oh
boy
I
think
we
can
probably
use
GitHub
cash
because
there
is
like
a
Gilda.
There
is
GitHub
actions
do
have
does
have
cash
that
can
put
us
across
build.
So
I
think
this
is
just
going
to
be
a
matter
of
running
something
to
Cache
the
crate
index
to
that
and
then
fetching
from
fetching
that
is
like
a
volume
or
whatever
so
I
think
that'll
be
I'll,
go
ahead
and
write
up
an
issue
on
this
I.
B
Because
in
locally
checking
it
like
the
index
is
not
even
like
50
meg.
So
it's
just
that
it
needs
to
know
where
all
of
those
go
and
it
will
make
sure
that
it
knows
where
it
goes.
D
A
Thank
you.
This
is
the
current
list
as
I
have
it
today.
What
do
I
do
this
I,
don't
know
as
I
run
into
things
I.
Just
it's
easy
for
me
to
throw
it
into
a
text
file
then
write
up
a
whole
GitHub
issue
and
if
something
bugs
me
enough,
I'll
radiating
you
out
of
it
I
think
we
should
probably
discuss
this
right
here.
D
A
There
you
go
see,
Miko
is
all
we
should
just
be
using
what
we
go
does
yeah,
and
so
this
one
here
Carter.
Well,
let
me
start
with
this:
one
I
don't
know
if
y'all
notice
in
the
demo
there's
like
that
yellow
vocal
thing
that
shows
up
right
here,
yeah
the
little
little
bar
yeah
yeah,
yeah
I
tried
setting
the
environment
variable
to
other
variables,
so
they
even
looked
in
the
source
code
for
front
end
for
what
variables
it
uses
and
it's
and
it's
enum.
It
doesn't
work.
G
G
F
I
have
the
Hipster
shock
the
the
old
project
deployed
in
two
different
environments
here
in
dynatrace,
and
we
do
use
that
I
I
haven't
applied
in
the
demo
yet.
But
it
is
nice
because
sometimes
we
we
just
forget
in
which
one
we
are
and
then
we
have
it.
A
F
C
A
Yeah
yeah,
it's
I'll
also
come
out
and
say:
can
it
not
be
an
enum
I
would
appreciate
if
it
just
could
be
a
string
and
I'd
plug
in
whatever
I
want
in
that
string?
So
I
might
say
this
is
AWS
eks
demo.
This
could
be
whatever
it
is
different
things.
So
I
I'll
vote
like
like
okay.
Let's,
let's
allow
us
to
pass
this
variable
in
just
other
people
whatever
and.
C
A
Carter
we
have
copyrights,
I,
think,
there's
like
front
end,
doesn't
have
them
and
then
there's
a
couple
code.
Files
like
some
of
our
services
have
more
than
one
code
file
and
we
don't
have
the
copyright
notices
on
all
the
code.
Files
and
I
think
that's
what
I
was
meaning
by
this.
D
Yeah
I
did
a
rough
and
dirty
license
Edition
before
the
V1
released,
but
I
don't
think
we
got
many
eyes
on
it.
It
was
really
me
doing
my
best
across
the
services
to
scope
out
where
the
the
proper
license
should
be
so
more
than
happy
yeah
to
have
these
taking
a
look
at
again
and
also
have
a
decision
on
what
the
gcp
notices
and
then
like
multiple
licenses,
Food
Services
too.
C
Don't
we
do
we
not
have
the
license
Checker
I?
Guess
we
don't
have
that
in
the
make
file
daily
and.
A
Hey
I
guess:
should
we
be
changing
the
notices
away
from
gcp
to
be
all
open
Telemetry
at
this
point
and
give
you
know
accolades
to
Google
for
the
wonderful
starting
point
they
provided
us
with.
C
A
A
C
A
Okay,
so
I'll
write
up
an
issue
to
encapsulate
this
and
I'll
go
through
the
last
few
files
that
I
think
we're
missing.
I'll
do
another
big
copyright,
PR
oud
I'm,
not
going
to
talk
about
this
just
yet
I
have
a
cool
chart
to
show
y'all
Austin.
Is
this
on
you
next
time
you
release?
We
just
don't
include.
C
A
For
images,
don't
have
these
in
there
I
want
to
get
into
it.
Software
has
these
images?
Do
not
I,
don't
want
to
talk
about
this.
It's
up
foreign.
A
C
Can
you
actually
there
actually
is
an
open
issue
on
release
process,
so
we
should.
C
C
We
it
is
in
my
head
right
now:
I
love.
A
D
D
A
Value
add
yeah
what
I'm
doing
here.
These
plus
signs
are
just
telling
me
to
create
issues
out
of
these.
That's
all
yeah,
okay
performance,
I,
actually
yeah,
stupid,
zoom
bar.
That's
this
one
right
here,
so
I've
been
playing
with
this.
This
is
running
now
for
like
a
day
and
a
half
about
of
time.
What
are
we
looking
at
I'll
be
whenever
this
happens,
I
mean
something
restarted.
A
A
I
will
show
you
what
it
looks
like
the
bottom
chart
is
actual
consumption
or
I'm.
Sorry
not
concerned
utilization,
and
you
can
see
it
does
some
weird
funky
things.
I
don't
know
if
it
needs
more
memory,
but
when
I
see
a
chart
that
looks
like
that,
I'm
not
really
inclined
to
give
it
more
memory,
I'm
more
inclined
to
what
the
hell
is
wrong
with
it.
A
Yeah,
so
I
do
believe
we
have
a
bug
here.
Another
way
to
look
at
this
is
when
I
try
to
sort
this
by
percentage
I'm
trying
to
Target
80
percent
memory,
utilization
am
I
being
weird.
Should
I
Target
higher,
lower
I,
just
kind
of
pulled
that
that
number
out
of
thin
air.
C
A
Think
that
makes
sense
so
Jaeger,
our
spans
got
bigger.
Our
traces
got
bigger
is
what's
happening,
we're
adding
more
instrumentation
and
more
spans.
So
a
memory
assumption
we've
made
for
B1
no
longer
applies
I'm,
going
to
give
the
Edgar
a
little
bit
more
memory
on
the
next
Go
Go
pass
on
this
see
I'm
running
this
I've
been
running
for
a
while.
C
C
Well,
also
elastic,
so
let's
talk
about
a
license
as
we
like
I.
A
I
should
mention
I
have
yet
to
figure
out
what
to
do
with
Prometheus.
It's
still
uncapped
and
it
looks
like
this
I'm
not
really
fond
of
that
yeah.
C
C
A
C
But
like
from
a.
E
F
Just
I
just
need
to
put
a
URL
and
token
it's
fine
I
just
go
and
configure
in
The
Collector,
but
if
I
need
the
front
end
exposed
in
the
way
that
we
are
exposing
at
the
moment,
we
need
to
rebuild
the
front
end
or.
C
C
I
think
there's
an
argument
to
be
made
that
we
could
support
like
multiple,
like
we
could
support
multiple
things.
Here
we
could
say
like
okay,
here's
like
the
like
most
Libre
version,
but
then
here's
the
like
grafana
version
right
like
here's,
the
LG
TM
version
of
this,
where
everything
is
wired
up
into
can.
A
Yeah
yeah
yeah,
because
I
think
that
actually
dovetails
well
into.
Can
we
drop
Docker
because
then
you
could
do
stuff
like
presets
in
a
Helm
chart
right
and
you
could
just
say:
do
you
want
to
install
the
default
or
do
you
want
to
install
an
open
search
version?
Do
you
want.
C
B
A
Okay,
so
that
was
a
long
way
of
me
saying:
email,
Services
memory
leak,
okay,
and
we
should
probably
fix
that.
The
next
one,
Carter
I,
think
we're
going
to
talk
about
is
k8s
right.
We
got
like
four
minutes.
D
Yeah
we
have
that
Docker
and
to
kubernetes
as
an
option,
so
we're
gonna
go
ahead
and
discuss
it.
The
wire
group
right
now.
G
D
G
A
G
B
C
Like
because,
if
you're
trying
to
reduce
complexity
for
local
dev,
then
the
the
easiest
way
to
reduce
complexity
for
local
Dev
is
drop
Docker
entirely
and
just
run
the
you
know
run
the
applications.
Obviously
that's
slightly
challenging
in
terms
of
tool
chain,
because
now
you
need
literally
every
language
Under.
The
Sun,
but
I
I
feel
like
if
you're
worried
about
that
like
there
is
a
lot
of
options
for
local
kubernetes.
We
can
make
it
very,
very,
very
straightforward.
C
We
can,
you
know,
bring
in
just
like
kubernetes,
manifest
files
and
say
like
hey,
you
just
keep
cut
all
Dash
F
Dot
and
everything
kind
of
deploys
like,
and
it
means-
and
it
gives
us
a
much
easier
and
better
way
to
handle
things
like
getting
metrics
out
getting
logs
out
like
you
can
incorporate
Argo.
C
G
G
I
think
this
comes
down
to
the
goals.
A
I
think
if
we
could
make
the
local
dev
be
you
brew,
install
or
you
install
a
couple
binaries
on
on
Windows
being
the
probably
need
home,
tooling,
Argo
and
a
local
k8s,
and
we
really
like
it's
got
to
be
down
to
it's
simple
for
anybody
get
started
and
we
have
to
add
the
docs
for
it
and
then,
as
long
as
the
dev
experience
is
great,
the
benefits
of
dropping
kubernetes
is
we
could
do
more
open,
Telemetry
things
with
this
demo.
D
A
C
D
Can
replicate
that
experience
in
kubernetes
I'm
open
to
it,
but
I
think
that's
the
challenge,
because
we
can
do
that
in
Docker
and
it
is
accessible
and
people
do
know
it
so
I'm
gonna
defend
the
dummies
like
me
and
I
know
that
Docker
is
simpler,
while
kubernetes
is
more
technically
elegant.
So
we
should
keep
in
mind
we're
building
for
the
the
bulk
of
people
here.
So.
G
Yeah
yeah
I
agree
with
that
and
also
I
think
I
wrote
another
comment
on
the
brainstorm
for
product
roadmap.
We
couldn't
get
to
that,
but
basically
what
if
we
sort
of
think
as
one
big
application,
what
if
we
have
like
multiple
applications,
which
each
demos
like
one
aspect
of
open
telemetry
like
I,
think
right
now,
what
we
are
trying
to
do
is
like
like
have
one
big
application
and
deploy
it
and
get
it
out
like.
G
Maybe
this
may
not
be
a
bit
of
approach
to
get
all
the
things
open.
Telemetry
supports
like
spam
links
like
what
about
histogram
metrics
like.
C
Yeah
I
mean
I,
think
there's
I
think
there
are
like
a
good
example,
is
sort
of
the
thing
that
came
through
about
open
symmetry
operator,
Auto
instrumentation
of
like
Java.
It's
like
okay.
You
could
definitely
see
a
use
case
for,
like
hey,
there's
a
separate
like
completely
uninstrumented
demo
app
and
then
it's
the
demo
is
just
like
deploy
spring
prep
clinic
and
use
Auto
the
instrumentation
operators
to
inject
instrumentation
to
it
right,
like
that's
a
separate
application.
C
I
think
that's
definitely
a
good
thing
to
kind
of
circle.
Back
on
next
year,.