►
From YouTube: 2021-06-25 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
Hit
like
the
highest
I
ever
remember,
I
mean
I
should
just
look
up
what
the
highs
are,
but
I
don't
ever
remember
it
being
over.
A
So
why
is
weather.com
is
so
bad
where.
A
B
Only
although
it
says
monday
is
still
going
to
be
106.
in,
I
think
in
accuweather,
which
is
what
I
have
on
my
phone
since
dark
sky
isn't
on
android
anymore.
Oh,
it
says
what
does
it
say?
It
was
saying
something
not
quite
as
bad
on
monday
only
98
on
monday,
but
it's
got
112
on
sunday.
B
A
C
C
C
C
C
C
All
right,
I
need
a
home
for
the
xp
remote
sampler
you're
reading.
B
We
talked
a
bunch
about
it
this
morning
and
transferred
it
out.
Past
pointed
out,
I
think
something
that's
more
important,
even
than
code
complexity
or
any
of
that,
which
is
that
it's
really
not
useful
for
anyone,
but
an
aws
apm
user,
like
an
x-ray
user,
and
I
think
that
does
kind
of
break
the
rule
of
vendor
apm
vendor
code
in
the
in
the
main,
repo.
C
A
Maybe
we
should
move
like.
C
C
B
B
A
B
I
mean,
I
guess,
there's
not
I
mean
it
would
be
out
of
the
bomb,
no
matter.
What
there's
I
mean,
there's
no
reason
in
particular
couldn't
keep
the
same,
coordinates
from
the
contributor.
B
C
C
C
I
mean
just
one
interesting
thing
I
noticed
is
that
since
we
don't
charge
for
the
sampling
api
technically,
you
can
use
this
remote
sampling.
Even
if
you
don't
use
x-ray
yeah.
I
wonder
who
would
it's
a
sort
of
nice
centralized
remote
sampler
but
yeah?
I
know
splunk
people
don't
even
sample
yeah.
What
is
the
sampling
thing?
I
don't
even
know.
A
C
B
B
C
C
A
C
A
Is
there
oh?
No,
we
know.
Actually
somebody
just
opened
an
issue
two
days
ago
wanting
a
rate
limit,
because
the
their
app
generated
four
million
exceptions-
and
you
know,
triggered
a
fairly
large
bill.
C
B
A
Like
a
the
ingestion
date
account,
but
actually
so
actually
they
don't
do
it
wanted
across
all
of
the
apps,
because
they
didn't
want
one
app
blowing
out
of
the
water
and
then
not
getting
capability
in
any
of
their
other
apps.
It
actually
did
want
it.
I
think
there's
some.
C
A
C
C
C
A
Yeah,
I
kind
of
think
you
might
probably
need
to
add
that
for
this,
this
user
request.
Basically
something
like
that,
and
I
mean
it's
gonna
come
with
that
same.
You
know
caveat
that
you
know
your
your
back-end
metrics
aren't
gonna
be
correct
if
it
happens,
because
we
still
calculate
metrics
based
on
ingestion
and.
B
A
We
keep
the
sampling
percentage,
so
we
multiply,
but
if
we
cap
it
you're
not
gonna
get
that.
But
so
you
do
serve
the.
A
No,
no,
we
send
the
sampling
percentage
that
was
used
in
the
spin
yeah
in
the
span
so
that
the
server
knows
how
many.
A
Yeah
and
there's
some
weird
logic:
that.net
sdk,
not
open
telemetry,
but
the
previous
azure
monitor
net
sdk
had
something
called
a
like:
an
adaptive
sampler,
which
would
every
minute
it
would
sort
of.
B
A
Itself
based
on
the
last
minute,
so
that
it
would
and
then
it
would
use
so
that
it
could
change
the
sampling
percentage
that
it
used
and
that
way
it
could
still
send
the
sampling
percentage.
A
A
Yeah,
hopefully
we'll
just
get,
I
mean
once
we
have
metrics,
then
we'll
do
metrics
on
the
client
side
and
we
won't
need
to.
C
C
B
C
A
Sure
what
the
reason
mean
like
I
could
think
of
like
you,
have
pretty
like
sort
of
like
that
exemplar
concept
of
you
want
to
know
like
something
errored
and
approximately.
How
many
that
would
have
been,
but
I
feel
like
metrics
gives
you
that
also.
C
A
B
Yeah,
it's
a
it's
a
net,
it's
like
6,
000
lines
of
new
code
and
twelve
thousand
lines
of
code
deleted
because
fpr
at
the
moment,
but
he's
gonna
break
that
up.
B
It's
a
crazy
pr,
but
I
actually
started
looking
through
it
today.
It's
so
much
easier
to
understand
than
what
we
have
in
place
right
now.
So
that
makes
me
happy
has
a
much
simplified
api
that
still
supports
everything
that
the
old
api
does
so
yeah.
B
C
B
So
right
now,
there's
kind
of
two
sides:
there's
the
the
hint
api
which
is
for
instrumentation
authors
to
say,
yeah,
I'm
going
to
generate
a
whole
bunch
of
of
attributes,
but
these
are
the
three
that
are
probably
the
most
important
and
the
sdk
if
possible,
by
default,
if
a
hint
is
specified
should
just
just
aggregate
on
those
three
or
whatever,
on
those
three.
C
Yeah
I
mean,
if
you're
generating
your
own
metrics
completely,
but
even
that
could
be
modeled
at
this
like.
So
I'm
imagining
that
sdk
could
have
a
mapping
from
this
metric
name
to
semantic
conventions
about
the
metric,
which
includes
what
we're
proposing
to
have
in
the
apis,
and
so
it
doesn't
seem
like
you've
missed.
B
That's
a
really
interesting
idea,
because
you
might
be
well
like
when
we
get
to
semantic
conventions
for
metrics.
You
might
be
able
to
specify
like
here's
all
the
stuff.
We
probably
you
might
want
to
capture
and
here's
the
three
things
that
you're
most
likely
to
actually
care
about
from
a
like
actual
useful
data
perspective.
C
Yep
and.
A
B
C
A
B
B
Where
some
people
are
like
well,
prometheus
defines
a
default
and
it
seems
to
work
for
a
lot
of
people.
We
could
just
use
that
and
josh
mcdonald
like
well
there's
this
cool
histogram
from
open
histogram
that
basically
should
work
most
like
99.9
of
the
cases
will
just
work
for
everybody
like
you,
don't
have
to
configure
it,
it
does
exponential
bucketing
and
it
just
works.
B
So
we
should
just
use
that
for
the
default
and
other
people
are
like
that's
complicated,
blah
blah
blah
anyways
and
then
some
people,
both
people
are
like
well,
there's
no
way
we
can
ever
possibly
solve
every
problem
with
the
default
set
of
buckets,
so
we
shouldn't
do
it
and
we
should
force
users
to
specify.
C
B
And
vigorously
in
today's
metrics
meeting
so
today.
C
A
B
B
So
that
all
of
the
views
can
be
validated
to
be
against
valid
instruments-
and
I'm
like
I,
don't
think,
that's
possible
in
java,
like
there's
just
no
way,
you
could
possibly
do
that,
maybe
maybe
in
go.
If
you
require
everyone
to
to
declare
their
instruments
and
static
initializers,
maybe
maybe
in
c
plus
plus
you
could
macro
it
into
fondness.
I
don't
know
I
don't
know,
certainly
not
auto.
B
Yeah,
I
kind
of
had
to
explain
that
java
is
an
interpreted
language
and
it
only
loads
libraries
on
demand
so
like
there's,
no
possible
way
to
discover
all
possible
instruments
before
you
configure
the
sdk,
because
someone
could
always
just
load
a
library.
You
know
after
half
an
hour
of
running
the
code
and
there
it
would
go
so
anyway.
C
B
B
Yeah
anyway,
I
guess
the
way
go
deals
with
it.
Well,
actually,
no,
I
don't
even
know
they.
They
don't
do
this.
They
don't
do
this
particular
thing
anyway,
so
I'm
also
having
to
push
back
and
make
sure
that
it's
okay,
that
the
sdk
configuration
is
immutable.
You
can't
go
and
do
stuff
on
the
fly
like
crazy
all
over
the
place.
B
B
It's
not
ready
for
a
third
review,
he's
going
to
break
it
up
into
a
bunch
of
small
small
pr's,
but
I
wanted
to
kind
of
see
where
he
was
envisioning
the
end
state
just
to
kind
of
understand
where
it
was
going
and
provide
feedback
on
that
before
he
went
too
far,
so
yeah
so
he's
going
to.
Basically
the
first
thing
he's
going
to
do,
I
think,
is
convert
labels
to
attributes
yeah,
and
then
I
don't.
B
A
B
B
Which
is
not
a
good
thing
in
the
first
place
so
anyway,
anyway,
and
then
the
pr
for
example,
ours.
I
wanted
to
push
off
exemplars.
I
looked
at
the
exemplar
stuff
and
I
think
it's.
B
I
think
it
needs
more
thought.
So
I
I
gave
him
the
feedback
already
off
offline
on
slack
so
anyway,.
B
B
I
think
is
very
clever
that
basically
every
builder
when
you
create
it
has
a
default
type.
So
counters
are
default
along
and
histograms
default
to
double,
but
you
can
change
in
the
builder.
You
can
say
no.
I
wanted
this
to
be
the
other
one,
and
then
you
get
a
basically
you
get
a
builder
back,
that's
the
other
type
and
you
can
keep
keep
working
with
it,
but
it
makes
the
api
surface
pretty
small,
and
you
don't
know
if.
C
B
The
normal
yeah
yeah,
so
that's
pretty
good
anyway,
there's
a
lot
of
crazy
generics.
I'll
say
that.
B
There's
not
really,
there
is
so
he
basically
to
accommodate,
pulling
bringing
the
context
along
with
each
measurement.
So
you
can
capture
exemplars.
B
Or
I
suppose
you
could
yeah,
I
think
the
real
reason
is,
or
you
could
capture
examples
or
you
could
potentially
pull
attributes
out
of
the
baggage.
I
guess
would
be
another
thing
you
could
do
with
that
feature,
but
he
he
created
a
measurement
object
that
wraps
every
measurement
along
with
the
context.
A
B
A
B
A
C
C
C
B
B
They
don't
they,
don't
they
really
don't
line
up
like
semantically
they're
different?
Maybe
they
mean
it
might
be
closer
now,
but
I
think
at
least
when
aaron
and
I
worked
through
a
whole
bunch
of
that
stuff
and
talked
it
through
like
it
ended
up
being
really
difficult,
and
I
think
the
gain
is
small.
I
mean,
I
guess
if
you
want,
and
if
you
want
to
send
your
data
to
prometheus
you've
already
got
a
prometheus
registry.
Like
you.
A
C
B
C
C
My
initial
impression
of
the
step
registry
might
be
something
like
the
api
like
you're,
only
giving
it
measurements
and
then
that
gets
aggregated
into
the
actual
registry
yep,
maybe
it
maps
sort
of
still
okay
to
hotel
api.
That
was
my
two
minute
impression
when
I
scanned
that
pr,
like
oh
step
register,
I
didn't
even
know
this
exists.
This
might
actually
work.
Pretty
well
was
my
first
impression
when
I
read
that.
B
Okay,
I've
implemented
one
of
those
for
when
I
was
a
relic
yeah,
just
mapping,
micrometer
metrics
to
new
relic
metrics,
and
I
mean
the
nice
thing
there
is
you're
dealing
with
the
then
you're
dealing
with
the
pre-aggregated
data.
So
you
don't
like
the
step
meter
registry
will
do
your
aggregation
for
you
and
then
you
just
need
to
know
how
to
map
those
the
aggregated
metrics
into
whatever
your
backend
wants,
which
is
why
I
was
thinking
we
could
just
map
them
directly
into
otelp
and
push
them
out
the
door,
and
it
should
be
relatively
easy.
B
Yes,
for
sure,
although
I
mean
the
the
otl,
this
is
what
the
other
thing
is.
Otlp
is
stable
right,
like
that,
actually
is
now
stable,
even
the
metrics,
so
you
could,
I
mean
logs,
isn't,
but
that
is
something
you
could
now
target.
Is
the
hotel
p
data
model
yeah
without
having
to
worry
about
the
stuff
inside.
C
B
I
mean
logs
is
definitely
not
stable,
but
but
I
think
they
have
definitely
declared
the
protobuf
stable.
Now.
C
B
C
B
Yeah,
I
think
I
think
so
I
mean
a
lot
of
what
he
does
ends
up
being
a
little
bit
rubber,
stampy
yeah,
so
I've
been
I've
been
waiting
a
little
bit
okay,
but
if
you
think
his
reviews
have
been
helpful
to
you,
then
I
will
I
support
it.
For
sure
I
mean
that's
definitely
weird
mean.
C
B
C
C
And
I
feel
that
that
could
be
sort
of
different
than
approver
and
maintain
our
news
like
just
having
that
is
still
good,
like
christian
hasn't
been
reviewing
like
he
was
reviewing
a
lot
for
a
while.
That
was
hopeful,
and
now
he
hasn't
been.
So
I
feel
as
if
kuba
was
just
sort
of
taking
over
that
helpfulness.
Yeah.
C
B
B
B
A
This
is
resolving
technical.
B
C
A
But,
like
I
don't
know
if
they
serve
if
they
have
a
term
the
way
governance,
the
governance
committee
do
there
are
no.
There
is
a
number.
A
A
B
A
App
dynamics,
I
think
jade,
the
two
of
two
of
them
were
from
app
dynamics
right,
john.
Yes,
so
they
were
just
checking
in
about
the
kubernetes
resource
detector.
B
Oh
yeah,
they
had
a
question
about
whether
oh
I
forgot
this.
We
should
probably
merge
this.
Have
you
had
a
chance
to
take
this?
Take
a
look
at
this
by
the.
B
But
I
think
jason
lowe
was
saying
he
thinks
that
maybe
it
could
supplant
the
eks
detector
that
it
kind
of
does
the
same
thing.
B
C
B
C
B
Anyway,
yeah
thanks
for
reminding
me,
I
completely
forgotten
that
that
was
brought
up.
I
got
I
got
wrapped
up
in
reviewing
metrics
or
I'm
looking
at
metrics
well
on
top
of
trying
to
deal
with
insurance
companies
about
our
totaled
car
and
refinancing
refinancing
my
wife's
building
and
just
trying
to
deal
with
too
many
things
at
the
moment.
B
Yeah
they
decided
to
total
our
car
after
the
after
the
body
shop
took
it
apart.
I
guess
I
don't
know
whether
they
found
like
a
crack
in
the
frame
or
a
bend,
a
bent
frame
or
something,
but
they
brought
it
brought
an
appraiser
out
again
and
decided
they
were
going
to
total
it.
So
now
we
have
to
buy
a
car.
B
But
there's
a
but
there's
another
there's
another
facet
to
it,
though,
which
I
didn't
realize
I
was
talking
to
the
adjuster
from
the
insurance
company
that
like,
if
there's
a
crack
in
the
frame
or
a
vent
frame
or
something
like
that,
it's
probably
repairable,
but
the
liability
about
like
it.
It
means
that
you.
A
B
Potential
liability
on
that
vehicle
is
quite,
it
could
be
quite
a
bit
higher,
and
so
it's
something
that
they,
I
think
when
you
get
to
a
certain
point.
They'll
just
say
now,
we'll
we'll
part
out
the
car
and
it'll
be
cheaper
for
us
to
just
buy
you
a
new
car
and
it's
actually,
they
don't
use
blue
book,
which
I
think
is
interesting.
B
They
use
comparables.
They
actually
search
for
comparable
cars
in
your
area
and
use
that
to
put
together
what
how
much
money
they'll
give
you,
which
is
good,
because
they're,
basically
giving
us
fifteen
thousand
dollars
and
blue
book
value
is
12
000,
but
with
the
rent
with
the
used
car
right
now,
you
can't
get.
You
can't
get
our
car
at
twelve
thousand
dollars.
A
B
Yeah,
it's
going
to
be
a
mess
all
right!
Well,
I'm
going
to
take
off
yep.
I
will
review
that
I'll
review
that
pr
first
thing
in
the
morning
the
x-ray
sampler
thingy
cool,
all
righty
have
a
great
one.
Stay
cool
stay.
A
C
C
C
A
You
did
say
you
have
ac
right,
yes,
yes,
john
does
not
john
does
not,
though,
so
he
was
saying
on
monday
when
he
asked
to
work.
He
was
probably
gonna
go
into
his
wife's
office
where
they
have
ac.
C
A
A
It
does
a
retry.
What's
this
mean.
C
C
C
C
C
A
A
Yeah,
I
do
because
we're
we're
not
at
the
max,
I
think,
we're
at
the
50
concurrent
plan
and
there's
like.
A
Concurrent
plan
plus
I
mean
well.
A
C
Yes,
I
haven't,
I
know,
I'm
pretty
sure.
I
know
what
the
duplicate
stuff
is.
I
forgot
to
exclude
the
transitives
in
the
exporter.
Bundle:
oh,
okay,
cool!
That's
what
that's
what
it
was
one
year
ago
and
I
think
I
fixed
it
and
then
we
broke
it
so
that
that
exporter
is
just
the
bundle
who
got
disappeared
completely
and
bringing
it
back.
Now,
it's
back
to
the
state.
A
year
ago,.
A
C
C
C
A
C
A
A
Interesting,
like
auto,
attach
kind
of
a
functionality
for
cloud
foundry,
I'm
guessing.