►
From YouTube: 2022-02-08 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
A
A
B
A
C
Well,
it's
like
still
the
morning
here,
you
probably
shouldn't
drink
coffee
in
your
timezone,
though
it's
a
bit.
A
I
don't
think
we
have
a
lot.
Oh,
let's
see,
let
me
make
a
note.
A
A
B
B
A
B
A
A
Oh
right,
oh
yeah,
I
scrolled
through
this,
but
I'm
going
to
give
you
the
the
honor.
C
A
Nice,
nice
killing
off
of
some
groovy
code.
B
B
A
B
B
A
B
B
A
Was
that
your
you
have?
Is
this
the
issue.
B
B
A
B
C
A
Pr
keeps
failing
why
the
build
keeps
failing
on
this.
I
hope
it's
not
something.
A
Yeah,
let's
see
maybe
after
the
release,
nothing
nothing
before
the
release.
A
But
I
could
see
just
I
might
just
try
it
out
after
the
release
and
see
how
it
goes
for
a
week.
A
A
Yeah
so
anurag.
I
finally
understood
this.
So
what's
your
thought
on
not
having
it
because
I
mean
we
do
have
a
lot
of
these
unregister
tests.
B
I
mean
we
should
flush
in
these
tests,
but
in
general,
like
at
the
end
of
every
test,
we
clear
all
the
exported
data
just
to
make
sure
it's
not
left
over
to
the
other
test.
That's
not
a
thing
for
metrics,
though
so,
ideally,
that's
a
no-op
in
almost
all
cases,
so
we
don't
want
to
do
it
on
every
test.
I
think.
A
Yeah
separately,
not
clear,
not
calling
clear
data
yeah.
B
A
C
Okay,
I'll
try
to
use
flash
like
everywhere
in
my
computer
tests
in
some
separate
vr,
it's
it
and
rag.
Is
your
pr
that
added
flash
mesh
already.
B
Flush,
the
grpc
diameria
one,
but
I
stopped
so.
I
deleted
it.
So
it's
not
there
anymore,
yeah,
okay,.
C
B
B
A
A
Oh
yeah,
I
wanted
to
get
under
your
thoughts
on
so
the
client
name,
so
somebody
wants
to
customize
the
or
the
spam
name
for
grpc
library,
instrumentation,
and
so
the
two
options
are
one
just
you
know
simple,
full
override
and
the
other
is
similar
to
the
armeria
status.
Extractor.
Where
you
can,
you
can.
B
B
Sounds
fine
one
thing
that
came
like
I
did
skim
this
a
bit
and
after
this
pr
like
it
will
be
good
to
split
out,
maybe
not
even
just
this
little
summary
and
other
ones
splitting
out
the
server
inclined
entry
points
just
to
prevent.
Having
these
two
said:
server
blah
blah
blah
blah
blah.
I
think
it
does
get
a
bit
weird.
B
I
think
that
tends
to
keep
like.
I
still
have
my
three-month-old
pr
that
stupid
builder,
but
that
wouldn't
work
with
the
sort
of
patterns
I
think
splitting
and
enforcing
the
what
properties
are
for
each
given.
Instrumentation
is
probably
a
good
idea.
A
And
how
about
that
on
the
using
the
army,
the
the
function,
the
chaining
the
optional
chain,
like
that.
B
B
A
A
I
was
going
to
search.
I
was
on
that
the
experimental
properties
it
was
your
pr
materia
the
config
for,
and
I
can't
even
remember
what
I
just
reviewed-
micrometer
one.
It's
probably
the
most
recent
configuration
the
base
time
yeah
unit
yeah,
I
wanted
to
say
experimental,
but
then
I
was
like
well,
what's
the
point
of
calling
it
experimental
in
an
alpha
module
yeah,
I
think
we've
called
some
other
things
experimental
in
alpha
modules
that
I
mean
maybe
that
those
could
get
used
in
the
job
agent
right.
C
B
A
That's
a
good
question,
so
we
don't
technically
have
to
based
on
our
stability.
B
A
A
Yeah,
but
we
also
changes
to
the
telemetry
produced
by
the
instrumentation.
A
Yeah
yeah
because
there
are
also
properties
that
affect
like
the
working
like
the
extension.
I
think
this
was
sort
of
the
point
was
there
were,
like
extension,
ones,.
A
B
B
B
In
the
context
of
all
that
sql
commenter
stuff,
like
that's
something
I
would
have
expected
to
be
experimental,
even
though
it
only
affects
one
instrumentation,
I
don't
know
if
propagation
internet
skill
stream
is
considered
telemetry
or
not-
I
guess
not,
but
then
at
that
point
it's
like.
Are
they
really
all
that
different
concept,
though.
A
Yeah
makes
sense,
let
me
let
me
look
through
see
where
we're
at
what
our
current
state
is
and
I'll
open.
B
C
Oh
yeah,
the
next
one
is
mine,
so
somebody
opened
an
issue
about
their
usage
of
the
micrometer
bridge
with
the
from
vs
exporter
and
a
couple
of
bits.
Later
it
turned
out
that
they
are
configuring.
C
The
distribution
summaries
in
micrometer
with
their
own,
like
buckets
their
own
custom
bucket
settings,
and
it
turned
out
that
they
get
overwritten
by
the
default
histogram
bucket
settings
provided
by
the
auto
sdk
when
you're
using
prometheus
exporter,
because
the
the
names
for
the
buckets
and
the
like
bucket
gauges
that
micrometer
produces
for
all
histograms
they're
the
same
so
apparently
our
exporter
just
merges
them
together
and
hotel
buckets
win
for
some
reason.
C
Oh,
I
think
that
that's
the
that's
a
good
command
so
in
if
in
these,
in
this
text
boxes,
if
you
scroll
a
bit
right,
then
yeah
and
that's
the
same
one
for
the
one
below.
So
you
can
compare
the
tags.
C
Okay,
so
the
ones
the
upper
box
come
from
they're,
the
ones
that
are
generated
by
the
java
agent,
if
you're
using
micrometer
and
as
you
can
see,
all
those
less
or
equal
than
100,
250,
500
and
so
on.
These
are
the
default
packets
that
are
used
in
all
histograms
and
if
you
don't
create
a
view
with
a
custom
regression.
C
This
is
what
you
will
get,
and
this
is
what
happens
when
this
user
instrumented
the
application
with
java
agent
and
if
they
didn't-
and
they
just
used
that
confused
with
the
registry
from
my
computer,
then
the
results
from
the
bottom
or
the
lower
box
happens.
So
you
can
see
that
they
have
some
custom
custom
markets
like
two
thousand
hundred.
C
Yeah
exactly
and
what
happens
underneath,
I
think
is
that,
because
what
what
does
micrometer
do
when
it
like?
How
does
it
represent
the
buckets
it
for
every
histogram?
A
C
A
C
Yeah,
so
you
can,
when,
when
you
obtain
a
meter,
you
can
pass
the
distribution
config
and
just
say
what
kind
of
buckets
and
what
kind
of
percentiles
you
want
there
and
the
micrometer
will
create
a
separate
gauge
for
each
of
these
buckets.
So
what
happens
in
our
own
bridge?
Is
we
create
an
total
histogram
for
the
timer?
C
And
we
and
if
you
use
like
custom,
micrometer
histograms
and
you
you
pass
the
custom
distribution,
config
settings
or
whatever
it's
called,
then
we
will
just
fall
back
and
allow
my
computer
to
create
a
set
of
gauges,
hotel
gauges
for
each
bucket.
So
you
have
like
double
histograms.
You
have
the
auto
histogram,
with
its
own
custom,
wheels
on
bucketing
from
coming
from
auto
sdk,
and
you
have
a
set
of
gauges,
but
they
have
the
exact
same
names
yeah
and
they
don't
necessarily
have
the
same
exact
names.
C
The
other
way
around-
I
don't
know
the
history
for
that,
but
yeah
it
so
it
turns
out
then,
when
you're
using
a
prometheus
exporter,
it
gets
the
auto
histogram
and
creates
a
prometheus
gauge
for
each
bucket,
and
these
turn
out
to
have
the
same
name
as
the
micrometer
gauges
that
we
also
export
and
they
basically
complete
with
each
other.
C
That's
one
idea,
although
it's
the
about
that's
kind
of
terrible
for
everything,
that's
not
confused
right!
So
because
you're
not
like
translating
the
calls
to
an
equivalent
instrument,
you
know
for,
like
I
think,
if
you're
using
like
otlp,
you
will
have
hotel
histograms
working
correctly
and
then,
in
addition
to
that,
you'll
have
the
micrometer
histograms
that
you
can
fall
back
to
if
you're
like,
depending
on
them.
B
C
Second
view:
what
what
I
think
should
happen
like
ideally
here
is
the
micrometer
histogram
settings.
The
packaging
basically
should
get
translated
to
auto
buckets,
but
that
can't
happen
right
now,
because
the
api
doesn't
allow
you
to
set
the
buckets
that's
only
available
in
the
sdk
and
only
during
the
sdk
meter
provider
setup.
So
you
can
register
the
packets
later.
When
you
actually
create
an
instrument,
you
can
register
the
view
that
has
the.
C
But
you
have
to
do
it
first,
for
example,
what
what
could
probably
solve
it
here
is.
The
user
could
create
an
extension
that
could
get
loaded
with
the
java
agent
and
the
extension
could
specify
views
for
all
histograms
into
his
application,
but
that's
quite
terrible
yeah.
C
So
I
think
that
there
are
probably
two
solutions
here.
One
is
the
one
is
bogdan's
idea
of
using
a
meter
metric
producer.
C
For
instead
of
apa
to
api
translations,
but
I
don't
like
that,
since
it's
very
extremely
complex
when,
when
you're
using
it
with
the
java
agent,
then
it
doesn't
really
work
with,
like
context
correlations
with
baggage
and
exemplars
or
you.
We
could
probably
wait
for
a
hint
api
right.
If
you,
if
you
could
be
able
to
pass
the
bucket's
configuration
as
a
hint
to
the
instrument,
then
if
you
do
not
have
a
view
that
changes
that
they
could,
you
could
just
take
the
buckets
from
the
hint.
C
Numbers
the
second
number
is
the
time
stamp.
I
think
oh.
B
C
Okay
and
the
time,
apparently,
the
timestamp
also
breaks
something
for
that
user,
but
I
have
no
idea
if
it's
our
problem
or
graph
on
a
problem
like
it's
a
final
problem,
not
sure,
I'm
not
sure
about
that,
though,.
C
A
thousand
yeah,
it
was
the
reason
why
I
added
that
base
unit
pr,
because
we
ex
before
that
pr,
we
exported
everything
in
microseconds,
and
I
confused
that
and
told
the
user
to
name
it
seconds,
even
though
it's
microseconds,
okay,
no
milliseconds
what
well
yeah
so
that
should
be
kind
of
sort
of
fixed.
After
I
mentioned,
my
printer
appears.
B
B
C
B
A
Would
we
still
be
duplicating
the
effort,
though,
if
we
had
those
buckets
in
hotel
and
our
micrometer
bridge
is
creating
gauges
or
would
be.
C
We
actually,
if
we
had
like
the
back
buckets
in
hotel,
we
probably
wouldn't
have
to
create
gauges
for
for
buckets,
but
micrometer
also
supports
the
calculating
percentiles
on
the
client
side,
and
you
get
a
separate
set
of
gauges
for
them
and.
A
B
C
Yeah
are
think
that
we
should
probably
say
that
hey
try
to
continue
to
juice.
For
now,
since
it's
experimental-
and
we
should-
I
mean
it's
not
going
to
work
properly
at
this
point
from
if
you
use
until
we
have
the
spec
and
the
implementation
for
hints,
which
we
need
anyway.
So.
A
C
A
C
All
kilps
or
sorry
yeah
for
otlp
and
kind
of
forward
compatibility.
I
mean,
if
maybe
if
somebody
starts
to
expect
that
micrometer
timers
are
translated
to
a
set
of
open
symmetry
gauges.
Then,
if,
when
suddenly
histograms
start
supporting
all
of
these,
and
we
change
it
to
histogram,
it
will
break
right.
C
Yeah-
I
don't
know
that
too,
but
I
think
that
everybody
that
uses
these
both
things
like
the
micrometer
bucket
configuration
and
prometheus-
will
probably
run
into
this
at
some
point.
C
So
well,
it's
probably
nothing
that
we
can
fix
right
now,
but
it's
still
a
problem
that
we
should
be
aware
of.
B
Maybe
what
is
the
naming
issue
here
was
that
the
is
that
a
problem
with
our
prometheus
exporter,
here's
the
underscore
seconds
or
something.
C
Oh,
I
think
the
custom
rename
code
that
I
suggested
to
the
user
probably
adds
one
seconds
too
much.
A
C
C
Yeah,
that's
that
would
fix
that
issue.
That's
I
think
it's
a
pretty
good
idea
for
now.
C
So
right
now
it
could
cause
changes
in
naming
and
the
instrument
type,
but
when
we
start
supporting
kids,
then
we
could
change
it
so
that
it
will
only
change
its
name
since
auto
histograms
would
contains
and
things
will
do
it
properly.
Yeah.
It's
a
good
idea.
I
like
that
I'll
try
to
fix
that.
I
probably
won't
manage
to
do
it
for
this
release,
but
yeah.
C
A
A
B
A
A
A
B
B
B
B
A
A
Yeah,
I
would
ping
laurie
if
he
was
you
know,
or
just
mention
in
the
pr
that
there's
a
you
know
that
there's
a
it's
core
dumping
and
point
to
the
core
file
and
if
he
doesn't
pick
it
up
I'll,
take
a
look
at
it
tomorrow.
B
A
A
B
A
B
A
B
A
A
not
push
it
for
the
release
if
we
can't,
if
it's
failing.