►
From YouTube: 2022-04-14 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
B
Yes,
yes,
the
what
like
around
tuesday
around
noon,.
B
It
could
have
been
worth-
oh
also,
it
was
weird
the
power
came
on
too
early
tuesday
morning
at
like
3
a.m,
for
a
couple
hours
and
then
went
back
out
which
actually
proved
to
be
very
convenient
because
the
heater
went
got
to
run
for
two
hours
and
nice
so
tuesday
morning
didn't
suck
quite
so
badly
and.
B
So
I
will
go
through
for
the
release:
hey
laurie
I'll,
go
through
the
open
pr's
here
see
if
there's
what
we
want
to
get
in
would
love
your
eyes
on
this
one
lori
just
in
case
there's
something
I'm
missing.
D
I
looked
at
it
and
couldn't
come
up
with
anything
better.
I
think,
okay,
I
initially
thought
that
it
would
break
while
fly,
but
but
there
is
a,
but
it
takes
a
different
branch
and
wildfly
and
and
so
it
keeps
working,
I'm
not
quite
sure
like.
Maybe
there
is
some
some
other
clever
way
to
start
wi-fi
like
like
whatever
clustering
option
they
have.
D
I
don't
know
whether
then
you
have
the
chapel's
home
set
or
whatever
this
current
logic
behave
like
depended
on.
B
B
I'm
assuming
that
that
delaying
I
know
at
least
in
our
reap
in
our
distro
wildfly
breaks.
If
I
don't,
if
I
don't
delay
that
branch,
because
our
we
initialize
some
stuff
like
whether
it's
m
beans
or
ssl.
D
B
D
B
Oh
yeah
yeah,
I
have
been
merging
that
I've
been
updating.
Our
distro
had
a
couple
other
problems,
updating
our
distro
not
not
relating
to
upstream,
so
I
haven't
been
able
to
get
numbers
from
any
customers
yet
on
startup,
but
still
looking
forward
to
that.
C
John,
do
you
want
to
talk
about
that
that
issue
that
you
that
was
transferred
to
the
java
repo.
A
I
don't
well,
I
guess
I
don't
understand
enough
about
prometheus,
to
really
understand
what's
going
on,
but
it
sure
seems
like
we
need
some
clarification
from
the
spec
about
how
we
should
be
writing
a
prometheus
metrics
exporter,
because
I
think,
as
you
commented,
we're
actually
doing
things
correctly.
They're
two
different
two
different
meters
with
two
different
two
different
meter
names
that
happen
to
be
exporting
with
the
same
exporting
the
same
metric
with
different
tags
on
it.
C
Yeah
prometheus,
just
in
general,
lacks
the
the
concept
of
a
meter
where
instruments
are
like.
You
know,
nested
under
a
meter,
so
that
hierarchy.
B
C
Yeah
trask
this
the
bottom
comment
right.
There
indicates
the
problem,
so
we
we
have
two
instruments
that
are
both
called
cue
size
and
they're
under
different
meters.
One's
part
of
the
batch
log
processor
one's
under
the
batch
span,
processor-
and
you
know
they
have
the
same
instrument
name
and
yet
they
don't
aggregate
together
because
they're
under
different
meters
and
so
prometheus-
that's
not
allowed.
C
C
But
you
know
we
shouldn't
have
to
do
that
because
that's
like
this
is
all
perfectly
allowed
within
open
telemetry's
metric
data
model.
C
This
is
just
unspecified
like
what
should
happen
in
this
case,
so
I'm
not
I'm
not
sure
what
whether
the
back
end's
expected
to
handle
it
or
not.
The
rules
in
terms
of
translating
the
open,
telemetry
data
model
the
prometheus
are,
are
lacking
in
this
category.
There's
been
some
attempts
to
like
solve
it,
but
it's
not
it's
not
obvious
what
to
do
like.
For
example,
you
could
prefix
the
in
the
metric
name
with
the
name
of
your
meter,
so
you
could
have
like
you
know.
B
B
B
C
B
C
So
one
suggestion
from
there's
a
pr
that's
closed
at
the
spec.
One
suggestion
is
to
add
the
meter
name
as
a
prefix
to
the
instrument,
but
that
obviously
isn't
great
in
all
scenarios,
because
some
back
ends
rely
on
the
instrument
names
and
not
having
that
prefix.
So
adding
the
meter
prefix
would
break
some
some
implementations.
C
B
There
was
you
had
to
open
another
one
that
was
interesting
to
me,
because
I
was
thinking
of
at
some
point
soon.
I
want
to
switch
over
to
the
the
upstream
micrometer
instrumentation.
C
Whereas
in
hotel
land,
an
instrument
and
description
has
like
is
the
top
of
the
hierarchy,
and
then
you
have
many
sets
of
distinct
attributes
underneath
that
and
so
in
what
you
get
in
spring
boot.
Is
you
get
instruments
that
are
registered
with
like
the
same
name,
but
many
different
descriptions,
and
this
is
an
example,
so
they
say
down
on
line
79.
C
And
the
so
that
that
works,
I
guess
it's
not
ideal,
because
you
lose
some
of
the
metadata.
That's
important
to
understand
the
the,
but
what
it
does
by
you
is
is
like
it
gets
rid
of
the
logs,
but
suppose
that's
fine
suppose
you,
you
agree
with
dropping
the
description.
C
Well,
you
can
do
the
same
thing
with
with
units
as
well
like
so
like
in
micrometer
land.
Each
distinct
set
of
tags
can
have
its
own
unit
associated
with
it,
which
is
kind
of
nonsense
to
me,
but
I
guess
they
allow
it.
I
didn't
see
any
instances
where
the
same
instrument
name
actually
had
different
units
for
spring
boot,
but
in
that
scenario
like
you
can't
easily
drop
the
unit,
because
the
unit
I
think
is
involved
in
the
prometheus
writing
rules.
C
B
Could
we
somehow
concatenate
the
oh
materia
just
in
time
to
talk
about
hey
good
morning,
micrometer
nice?
Have
you
seen
this
issue
already.
E
Yep
I
have,
and
I
think
that
we
could
either
like
skip
the
descriptions
altogether
or
maybe
maintain
some
kind
of
cache
or
a
map.
You
know
kind
of
a
previous
description,
so
if
the
previous
description
for
the
same
meter
name
was
different,
then
we
can
skip
it.
If
not,
we
can
preserve
it.
It's
a
bit
hacky,
but
in
case
somebody
actually
uses
it
correctly
in
the
open,
telemetry
sense.
We
will
preserve
them.
C
B
And
I'm
assuming
it's
immute,
the
description
is
immutable,
so
there
wouldn't
be.
I
was
wondering
if
we
could,
like
concatenate
the
the
the
multiple
descriptions
together
and
send
that
as
like
a
big,
you
know
description
at
least
in
you
know,
grafana,
then
you
would
see
all
the
potentially
different
descriptions
that
applied.
E
Yeah,
if
it's
different,
we
could
just
use
the
first
one,
which
might
be
a
bit
confusing.
I
mean
if
you're
passing
more
than
one
description
but
yeah
in
case
you're,
using
in
case
you're,
setting
the
same
description
for
for
all
like
metrics
with
the
same
name.
It
should
be
fine.
C
B
Is
would
it
be
prohibitive
to
to
raise
that
raise
it
at
the
spec
like
to
try
to
allow
descriptions
per,
I
mean,
I
guess
everything
like
otlp.
Everything
is
so
already
built
around
the
fact
that
description
is
mapped
to
the
meter.
B
C
And
it's
probably,
it
seems
like
it's
more
of
like
an
oddity
of
micrometer
like
I
think
this
is
a
probably
a
miss
from
their
api.
I
I
don't
actually
think
that
you
want
the
description
to
vary
for
each
set
of
tags.
I
think
that
you
know
the
attempt,
maybe
is
to
explain
what
this
distinct
set
of
tags
means,
but
you
know,
I
think
that
I
I
I
don't
like
that
as
a
concept.
E
I
think
that
micrometer
folks,
probably
or
inspired
too
much
microfuse,
where
you
know
each
separate
instance
of
tax,
is
a
separate,
separate
meter,
basically
in
the
primary
use
output.
B
Yeah
I'd
love
to
get
jonathan's
thoughts
like.
I
wonder
if
this
is
something
that
they
would
consider
in
micrometer
2o,
changing.
A
B
Asking
for
his
thoughts
on
this
and
if
there's
a
way
for
us
to
play
nicely
in
the
future
kind
of
maybe
outline
what
your
or
materials
thought
you
know
kind
of
the
best
we
can
do
today
and
asking
if
there's
he's
got
ideas
for
a
better
something
better
in
the
future.
B
Oh
mateish
for
the
release.
I
was
starting
to
go
through
and
just
mark
a
couple
things
that
I
was
gonna
try
to
get
merged.
I
saw
your
comment
I'll
have
to
look
at
the
spring
kafka.
B
It's
a
nightmare.
Unfortunately
yeah
I
was
hoping
to
get
this
in
just
since
it
seems
to
be
causing
a
couple
of
user
problems.
But
may
I
may
punt
we'll
see
this
one.
Our
review
today.
B
It
today
I
think
it's
it
would
be
good
to
get
in
just
so.
We
can
get
the
deprecation
train,
rolling.
E
Yeah
and
there's
a
lot
of
code
there
I
mean
a
lot
of
lots
of
changes,
but
it's
it's
actually
pretty
simple
and
like
95
percent
of
it
is
just
inline
tests,
so
copy
paste,
cool
and
then
this
one
I
think
this
one's,
probably
a
bit
too
complicated
to
you,
know,
merge
it
immediately.
E
B
E
B
You'd
be
surprised,
I
get
a
lot
of
support
tickets
because
of
weren't.
Just
you
know
a
warning
message:
it's
not
affecting
like
everything
it's
working
perfectly
for
the
for
the
user.
B
A
lot
of
so
I've
kind
of
shifted
so
that
the
hotel,
a
lot
of
the
hotel
stuff,
only
gets
enabled
if
we
turn
on
debug
logging.
B
B
Logging
is
a
bit
of
a
mess
yeah,
the
one
that
I
do
still
get
that
I
that
you
had
helped
with
recently
was
the
at
debug
level.
The
those
type
not
found
exceptions
during
transformations.
D
D
D
Yeah,
that's
that's
super
annoying,
but
they
have
been
doing
it
for
years,
but
it's
actually,
I
think,
it's
possible
to
turn
it
off
programmatically
by
patching
one
method
in
some
internal
ibm
class
that
make
it
return
a
different
layout.
D
It
really
isn't
that
bad.
Like
on
my
previous
agent,
we
used
that
I
think.
E
Yeah
about
this
warnings,
usually
java
agent,
just
looks
like
two
lines
like
this
version
and
like
pretty
much
it,
and
if
somebody
suddenly
sees
a
warning
message,
then
maybe
you
know,
since
it's,
it
will
be
pretty
much.
The
only
thing
that
agent
will
block
in
normal
non-debug
mode
it.
Maybe
it
will
be
visible
to
the
users.
E
B
Yeah,
but
this
kind
of
requires
remembering
to
do
this.
B
All
right,
oh
kafka,
mates,
let's
chat
briefly,.
B
Yeah,
it
was
so
it
was
weird
that
finally
rang
a
bell
like
I
had
opened
this
issue
forever
ago
and
then
yeah
forgot
about
it,
but
yeah
pretty
sure
that
this
is
what's
causing
problems.
B
So
is
it
I
was
gonna
ask:
is
it
just
the
the
consumer
records
iteration
or
is
there
even
other
parts
to
that.
E
Three
yeah
there's
producer
size
is
probably
working
flawlessly,
because
it's
it's
much
simpler
to
instrument,
since
you,
you
know
just
send
a
message
and
visit.
The
consumer
side
is
still
completely
broken,
though.
E
Yeah
yeah,
so
I
I
think
we've
talked
about
it
several
times
and
there
was
a
idea
that
we
could
instrument
like
all
the
higher
level
apis
like
spring
kafka
reactor
kafka.
Vertex
stuff-
I
don't
know-
maybe
there's
more.
E
E
Batch
processing
scenarios,
the
regular
like
single
record
processing,
still
relies
on
the
kafka
client's
instrumentation
because
it
works
in
that
case.
So
I
didn't
touch
it.
B
And
so
also
vertex.
E
Yeah,
I
think
vertex
kafka
is
like
80
percent
sure
that
it
works
and
the
reactor
kafka
is
almost
center,
certainly
broken,
so
that
would
be
a
net
new
thing.
E
E
I
don't
know
I,
I
think
it
uses
some
kind
of
hard
to
instrument
and
hard
to
understand
stream
api,
so
it
might.
It
might
not
be
that
easy.
B
Yeah
so
like
with
this
pr,
I
mean
it's.
B
Worse
for
sure,
because,
like
it's
not
really
server,
instrumentation
like
you
could
be
using
that
inside
of
you
know
some
courts,
job
or
spring
schedule,
job
or
yeah.
I.
E
B
E
I
I
think,
if,
if
we
instrument
that
the
higher
level
libraries
and
had
many
sources
of
some
sort
of
manual
utility
for
instrumenting,
like
like
the
one
you
posted
here
as
an
example,
then
we
could
maybe
turn
the
kafka
client's
instrumentation
off
by
default.
But
there's
a
lot
of
work
to
cover
all
those
all
of
these
libraries.
B
All
right,
well,
I'm
at
least
happy
with-
I
think
this
will
for
all
for
most
users
should
for
most
usage
patterns,
should
help.
D
B
B
I
know
in
in
our
distro
we
don't
enable
internal
spans
by
default
those
controller
spans
by
default
anyway.
So
I'm
trying
to
that's
why
I
had
sent
that
one
pr
to
change
the
jax
rs
common
instrumentation
name,
so
I
can
suppress
that,
but
I
do
think
those
controller
spans
are
useful.
Some
people,
like
them.
D
Like
the
operation
information,
I
think,
and
the
other
one
contains
the
actual
method.
That's
involved.
D
Which,
I
don't
remember,
maybe
just
update
this
the
server
span
name,
it
definitely
doesn't
have
a
server
server
span
on
its
own.
E
Yeah
it
does
update
server
responding.
I
remember
not
adding
http
right
there.
D
Yeah
because,
like
they
don't
have
like
a
natural
world
because
they
just
get
some
kind
of
poster
quest
and
just
take
some
stuff
out
of
the
booster
quest
to
somehow
identify
like
what
the
request
would
have
been
doing.
D
I
guess,
if
you
try
to
do
the
framework
instrumentations,
then
it
probably
is
going
to
be
a
bit
different
than
what
we
have
now
what
we
are
now.
Basically,
there
is
a
span,
that's
like
only
for
this
like
invoked
method,
but
then
we
have
to
like
widen
the
scope
a
bit
and
probably
just
use
the
reflection
method
or
whatever
to
create
the
spam.
D
B
Yeah
this
is,
I
noticed,
because
I
was
playing
around
with
disabling,
that
or
running
the
framework
specific
instrumentations
without
the
common
one,
and
one
of
the
changes
I
noticed
was
that
right
we
don't
get
the
exception
on
the
server.
Well,
we
never
got
the
exception
on
the
server
span.
I
think
in
jax
rs
anyway,
because
it
doesn't
bubble
up,
and
so,
if
we
suppress
the
internal
span,
that
kind
of
sucks
you'd.
Never
you
don't
capture
that
exception,
then.
D
B
Yeah
another
I
mean
one
option
is
to
just
disable
them
by
default
and
document
that
you
know
you
can
use
it,
but
it
could
impact
startup
time.
D
D
We
do
have
some
like
pretty
ancient
stuff
that
we
are
instrumenting
like
like
the
cereal
2.2
instrumentation.
B
Yeah
this
sounds
good
to
me
for
that,
given
the
ancientness
of
it,
I
mean
it's
like
it's
still
there.
If
somebody.
B
Really
wants
it
or
we
just
delete
it,
given
that
it's
only
internal
spans,
any
oh,
but
it
doesn't
have
its
span
name
also,
which
is
important.
B
So
this
so
jack's
rs
jax
ws,
the
these
are
annotation
based
ones.
We
still
have
our
open
telemetry
api
instrumentation.
The
width
span.
D
Those
have
a
different
issue.
I
think
I
tried
to.
I
created
the
issue
for
bite
body,
but
the
bite
by
the
altar
seem
to
be
a
bit
reluctant
to
change
the
behavior
which,
like
basically,
he
has
a
nice
api-
that
kind
of
behaves
like
the
java
api.
When
the
annotation
is
missing,
then
the
api
just
doesn't
report
it,
but
to
determine
that
the
annotation
is
missing.
D
I
I
have
some
changes
like
stashed
away
like
it's
possible
to
come
over
it,
but
it
won't
look
pretty.
D
Yeah
I
had
to
wrap
some
stuff
and
use
a
trade
local
to
detect
whether
whether
an
annotation
is
being
loaded
and
then
somehow
return
fake
results.
B
Now,
if
we
I
mean
so,
how
is
the
jax
rs
annotation
stuff
different,
because
I
know
as
soon
as
we
need
to
inspect
any
annotations,
then,
like
so
say,
we
have
both
open
telemetry
api
and
jax
rs
on
the
class
path.
D
The
difference
between
them
is
that
the
chucks
rs
and
chuck's
ws
they
search
for
annotations
in
the
class
hierarchy,
so
they
search
annotations
and
super
class
methods.
D
But
bitspam
it
do
researches
on
the
in
the
current
class,
which
is
actually
fine.
The
only
like
nasty
side
effect
is
that
it
wants
to
locate
the
actual
annotation
type
in
all
the
classes
in
all
the
methods,
although,
like
for
efficiency,
it
would
be
like
it
could
just
match,
based
on
the
annotation
name,
without
locating
any
additional
class
files.
D
Yeah,
I
also
also
saw
it,
but
I
hope
that
we
won't
implement
that.
B
B
B
D
D
B
Oh
yeah,
I
I
I
mean
I
will
say
that
it's
very
convenient
to
annotate
interface
methods.
I
did
that
in
glow
root
a
lot
and
yeah
definitely-
and
I
had
logic
to
detect
like
if
it
was
implemented
in
if
it
was
implement.
If
the
method
existed,
but
it
the
interface
didn't
get
pulled
into
lower
yeah.
There
were
a
lot
of
edge
cases,
but.
B
I
mean
without
one
nice
thing
that
I
like
about
so
the
the
way
that
glow
root
dealt
with
this
is
it
kept
the
cache
of
everything
like
a
kind
of
thin
cache
of
all
the
classes
that
had
been
loaded
so
that
it
could
go
and
look
quickly
at
relook
quickly
at
the
class
hierarchy,
but
that
was
not
free
in
terms
of
memory
that
was
generally
like
about
20
or
30
megs
easy,
and
so
I
really
like
when
I,
when
I've
seen
you
know
that
we
didn't
have
that
problem
here.
B
B
B
Yeah,
especially
the
internal
spans,
I
could
see
people
I
mean
our
our
users
by
default,
we
don't
capture
internal
spans
and
most
people
actually
prefer
not
to
be
cluttered
and
pay
for
the
internal
spans.
B
Do
you
also
sample
not
by
default,
but
there's
a
percent?
I
mean
that's
generally,
the
cost
control
that
users
apply.
B
B
Okay,
yeah
cause
like
jax
rs.
This
gets.
This
is
a
pretty,
is
used
by
corcus
and
a
bun.
Basically,
my
my
thinking
is,
if
you're
not
using
spring
you're,
probably
using
jax
rs.
D
D
D
D
D
So
trust
like
have
you
thought
about
those
this
http
client
that
we
tries
and
the
redirects
should
create
separate
spams.
B
E
Now
couple
of
days
ago,
I
counted
how
many
http
client
instrumentations
we
have
and
there
is
30
different
hdb
client
attributes
gather
implementations
in
the
agent.
So
I
think
that's
at
least
a
good
month
or
two
of
work
on
these.
B
Yeah,
that's
true:
it's
gonna
be
really
hacky.
The
java
agent
instrumentation,
I'm
afraid,
because
it'll
yeah
require
really
low
level.
D
E
Yeah,
I
remember
reviewing
something
that
did
with
the
retro
did
something
with
the
retrace,
and
I
think
he
had
to
do
some.
B
B
B
I
had
a
we
got
one
minute
left
laurie
I
had
one
just
remembered,
http
client
made
me
think
of
you
know
the
change
you
made
to
the
http
url
connection
recently
to
simplify
the
double
check.
D
B
Because
I
had
somebody
report
in
our
distro
infinite
recursion
there
and
it
was
still
on
the
old.
It's
still
on
the
old
code.
My
double
check
thing
and
I
keep
staring
at
it
and
can't
figure
out
how
it
happened,
but
I've
given
them
the
latest
snapshot
with
your
changes,
I'm
hoping
that
maybe
that
solves
it.
But
I
was
wondering
if
that
happened
to
have
been.
You
would
happen
to
happiness.
D
Like
I,
I
guess
my
first
question
with
those
is
always
like:
can
this
be
like
reliably
reproduced.
B
B
We'll
I
will
see
you
all
yeah,
so
I
will
try
to
kick
off
the
release
today
or
tomorrow.
Things
are
looking
good,
everything's
awesome.