►
From YouTube: 2022-04-15 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).
C
Yeah
this
is
when
I,
during
the
pandemic
realized
I
was
going
to
be
working
from
home.
I
needed
to
have
greenery
around
oxygen
need
oxygen,
nice.
C
Off
and
on
for
a
decade
or
two,
no
two
decades
I
lost,
I
had
a
fish
tank
in
the
house.
I
lived
in
before
this
one.
I
lost
everything
due
to
not
having
a
good
handle
on
portland
water
chemistry
and
how
it
shifts
during
the
year.
Oh
weird,
the
water
chemistry
in
portland.
So
in
the
spring.
C
No
in
the
so
in
the
summer
when
it's
really
dry,
basically
we're
pulling
water
out
of
the
reservoirs.
They've
absorbed
a
lot
of
minerals
out
of
the
reservoir
or
the
earth,
and
then,
when
it
starts
raining
in
the
fall,
we
get
all
of
our
waters
fresh
rain
water,
which
is
extremely
soft.
So
we
shift
from
extremely
hard
water
to
extremely
soft
water,
and
I
was
a
planted
tank
and
was
injecting
carbon
dioxide
into
it
for
the
plants,
carbon
dioxide,
when
is
not
buffered
by
minerals,
drops
the
acidity
of
the
water
very
very
fast.
C
C
Yeah
it
was,
it
was
kind
of
a
disaster.
So
then
I
gave
up
that
aquarium
for
a
while,
until
I
moved
here
and
then
like
three
years
ago,
built
a
new
tank
and
four
years
ago,
that's
a
new
tank.
C
C
I
mean
these
days,
it's
super
super
stable,
so
I
measure
it
once
a
month
or
so
I've
got
my
little.
I
got
my
I
have.
My
test
sticks
right
here.
A
You
might
like
this.
I
have
a
hydroponic
system.
Next
to
me,
I
grow
like
lettuce
and
stuff,
and
so
I
bought
this
device
that
monitors
water,
nutrient
concentration
and
acidity
and
temperature,
and-
and
I
monitor
it-
with
open,
telemetry,
nice
and
report
it
to
new
reality,.
C
C
C
B
Cool,
we
talked
about.
B
Oh
yeah,
you
saw
this
this
one.
This
was
a
great
find
I
thought
just
as
far
as
because,
as
we
start
thinking
to
use
this
now,
I'm
surprised
that
well,
john
you're,
not
in
the
split
in
that
group
anymore.
It's
fun!
I
wonder
if
the
splunk
java
agent
is
not
logging.
These
warnings.
C
So
the
microphone
java
agent
sends
micrometer
metrics
not
to
hotel
that
sends
them
via
the
old
signalfx
metrics
ingest.
B
Okay,
so
the
not
not
using
the
upstream
micrometer
instrumentation
yeah,
so
injecting
the
got
it
well.
That.
C
B
Yeah
but
as
we
start
thinking
about,
because
I
was
starting
to
think
about
using
the
upstream
micrometer
instrumentation
in
our
distro
but
yeah,
this
would
be
users
would
ask
what
what
the
heck
that
they
love
asking
me
about,
or
what
our
different
warning
messages
mean.
B
So
I
don't
know
there
was
oh,
there
was
a
comment.
B
A
A
Yeah,
so
both
of
these
kind
of
are
related.
I
probably
should
have
posted
this
comment
on
the
other
one,
it's
more
directly
related
to
the
other
one,
but
in
both
cases
it's
related
to
yeah.
I
gotta
move
that
comment
this.
This
comment
is
in
the
wrong
place.
One's
related
to
prometheus
and
one
is
related
to
to
micrometer
in
the
micrometer
bridge.
D
C
C
We
could
have
two
instrumentation
libraries
reporting
metrics
of
the
same
name
that
aren't
necessarily
combined.
So
I
think
we
need
some
sort
of
prometheus
clarification
exported
clarification
on
the
spec
side
about
how
we
should
actually
be
doing
this.
It's
not
really
clear
and
if
we
really
want
to
do
it
right,
it
also
needs
to
include
the
version,
and
then
things
get
super
hairy
and
the
metric
games
and
prometheus
will
be
super
ugly.
A
A
One
of
the
approaches,
I
think
that
they're
trying
to
go
with
right
now
is
by
expanding
instrumentation
scope
to
include
attributes,
and
then
they
would
have
an
attribute
called
like
something
like
a
short
name
like
a
short
name
for
the
meter,
and
the
idea
would
be
that
if
the
short
name
is
present,
then
you
include
that
as
either
prefix
or
an
attribute
on
the
prometheus
metric
and
I'd
like,
so
that
the
folks
that
are
in
that
kind
of
the
prometheus
metrics
working
group.
That's,
I
think,
what
they're
aligning
around
right
now,
but
it's
not.
E
D
C
D
Yeah,
like
I
don't
changing
the
name
that's
experienced
beneath
this.
That
just
seems
like
a
no-go
for
many
reasons,
and
so
in
the
meantime,
I
think
we
probably
should
still
drop
the
description.
The
conflict
and
descriptions
to
at
least
have
something-
and
I
suspect
the
real
fix
is
going
to
be
way
way
harder
than
we're.
Imagining.
A
That's
that's
correct.
That's
yeah,
that'll
show
up
when
you
use
just
spring
boot
at
all
and
that's
because
in
micrometer
world
each
distinct
set
of
attributes
can
have
a
different
description
and
so
that's
kind
of
incompatible
with
the
hotel
data
model
or
not
maybe
incompatible.
But
it's
inconsistent.
B
D
A
A
Just
so
what
matteo,
what
he
suggested-
and
I
I
just
want
to
confirm
this-
is
the
same
thing
as
what
you're
saying
so
the
first
metric
that
comes
around.
Let's
say
it
has
a
description
called
like
foo
for
the
for
the
instrument.
Jvm
thread
states,
another
instrument
comes
around
and
registers
under
jvm
thread
states
as
well,
but
its
description
is
is
is
bar
this
time
and
so
because
it
recognizes
that
we
already
registered
a
instrument
under
jvm
thread
states.
We
use
the
existing
description.
A
D
E
E
D
A
Yeah,
so
that
that
yeah,
that
would
fix
it
on
the
x
port
side,
but
you'd
still
get
all.
D
A
Yeah,
but
that's
kind
of
related
to
this
problem
of
you
know
different
er,
the
instruments
with
the
same
name
under
different
meters.
You
know
that
it's
not
necessarily
true
that
they
represent
the
same
thing.
A
You
know
they
could.
They
could
represent
two
completely
different
measurements,
and
so
it's
can.
You
combine.
D
D
Yeah,
it's
still
just
sharing
the
description,
though
I
don't
think
the
description
is
in
practice
really
used
that
much
anyways
people
just
writing
from
ql
using
the
name
and
attributes,
and
so
we
just
need
to
get
them
out.
Somehow
I
can't
think
of
any
like
changing
the
name.
I
think
that's
worse,
because
people
are
writing
from
kill
based
on
the
name
anyways.
Oh
yeah,.
D
D
D
A
Did
you
open
a
timestamp
issue
with
with
the
spec,
or
are
you
just
talking
about
that
that
issue
that
someone
opened
with
us.
B
B
A
B
B
D
Yeah,
I
I
mean
I
I
personally
never
used
it
like,
I'm,
not
even
sure
if
it
shows
up.
I
know
that
I've
entered
a
name
to
on
the
dashboard
myself,
because
I
never
tried
to
use
the
description
before
and
I
because
even
the
description
is
pretty
long
right,
so
you're
not
going
to
name
your
dashboard
with
that,
I
think
so.
D
A
D
D
B
And
yeah.
D
B
D
B
D
D
B
On
this,
one
just
wanted
to
confirm
with
all
three
of
you
so
the
because
I
think
honorary
discussed
a
little
bit
back
about
the
because
probably
this
would
end
up
in
the
sdk
extensions,
the
set,
or
at
least
the
spect
samplers.
B
So
the
question
was
sort
of
should
we
land
it
in
contrib?
First,
if
it's
gonna
land
in
the
sdk
repo
anyways.
So
I
put
the
question
out
and.
B
Atmar
replied
that
some
of
the
sampler
implementations
are
not
in
the
spec.
B
So
I
haven't
got
no
objection
to
landing
it
here.
First,
just
wanted
to
check
in
with
you
all,
since
it
probably
would
eventually
make
its
way
to
the
core
repo.
D
D
A
Yeah
I
haven't
looked
at
this
at
all,
so
I
it's
hard
to
have
an
opinion
except
for
at
the
high
level,
which
is
you
know
this.
It's
a
spect
component,
it'll,
eventually
land
in
there.
D
A
B
I
I
actually
need
actually
have
need
of
it,
so
I
will.
I
will
take
an
another
pass
and
try.
D
D
B
D
D
B
Yeah
so
I
mean
we
just
have
like
we
put
an
item,
count
like
a
multiplier
on
all
the
spans
and
then
there's
just
some
user
experiences
that
are
kind
of
sort
of
that
are
driven
by
that,
but
yeah
I
mean
and
part
of
it's
historical
in
yeah.
We
want
to
go
to
more
metric
stuff,
eventually.
B
But
even
then,
I
think
that
it
has
some
place,
although
for
more
sparsely
for
sort
of
your
long
tail
of
of
attributes
that
you
don't
necessarily.
B
What
else
we
talked
about
kafka,
so
yeah
mateusz
was
thinking
you
know
if
we
could
wrap
things
if
we
could
instrument
all
the
higher
level
components
kind
of
like
our
plan
for
nettie
that
never
materialized.
B
D
B
Yeah
yeah,
that's
a
good
point.
If
you
don't
make
a
comment
in
the
issue,
I
will
about
that
because
yeah,
I
think
we
pretty
much
agreed
that
yeah.
I
that
we
should
get
rid
of
I'm
just
like.
D
B
Yeah
so,
but
at
least
yeah,
if
we
had
some
good
docs
and
some
good
sort
of
templates
on
how
to
do
it,
that's
pretty
yeah.
That's.
C
B
D
B
Account
not
duration,
but
right
yeah,
just
the
count
just
like
the
count
of
the
number
of
messages
being
processed.
B
Do
you
think
it's
worth
people
using?
I
mean
like
in
that,
like
the
text
map
getter
like
having
a
usable
or
just
copy
page
or
just
telling
people
copy
paste,
it.
D
D
D
D
B
D
D
D
Yeah,
maybe
we
need
to
look
into
this,
but
I
mean
if
library
instrumentation
is
doing
the
right
thing,
then
of
course
the
job
agent
just
needs
to
be
rewritten
to
use
that.
But
I
thought
it's.
The
job
agents
is
trying
to
also
intercept
the
message
processing
itself
and
that's
that
would
probably
be
an
issue
for
library
users
as
well
see.
B
B
Cool
and
then
we
talked
about
jax
rs,
the
matchers,
the
slow
matchers.
B
B
D
B
Yeah,
so
that
will
be
a
reason
to
kind
of
probably
push
back
on.
That
request
on
that
feature
is
the
matchers,
and
so
if
these
three
jacks
rs1
and
jxws
are
sort
of
legacy
well,
jax
rs1
is
legacy.
Xws
is
fairly
legacy,
so
we
could
probably
disable
just
disable
them
by
default
and
then
jax
rs
lori
is
going
to
look
into
if
we
can
add
framework
specific
instrumentation
for
to
capture
those
internal
spans
that
the
annotation
driven
instrumentation
is
capturing
today,.
B
Jeff,
sorry,
it
means
jersey,
rest,
easy
and
cxf.
B
So
actually
I'm
I'm
playing
around
in
our
distro,
since
we
don't
capture
internal
spans
by
default
anyway,
with
killing
off
the
annotation
jax
rs
annotations
and
just
getting
the
the
route
from
the
framework
specific
one.
So
we'll
see
how
that
goes.
Also,.
B
Lori
mentioned
it's
part
of
the
problem
with
the
annotation
matchers.
Is
that
even
what
the
width
span
suffers
from?
B
Is
that
when
bite
buddy
looks
it
reads
all
the
annotations
and
it
actually
looks
up
those
annotation
classes,
as
opposed
to
just
being
able
to
say
that
oh
they're
here
and
they
match,
and
he
said
he
opened
an
issue.
B
Yeah
and
it
said
that
yeah
there
was
yeah
because
it
doesn't.
B
The
current
behavior
of
bite
buddy
matches
like
javascript,
it's
like,
and
so
this
would
be
sort
of
different
and
I
think
it
sounded
like
laurie
had
done
some
spiking
work
in
fight
buddy
like
that
it
you
know,
works
for
our
needs,
but
may
not
be
sounds
like
it
probably
wouldn't
be
upstream
upstreamable.
So
I
don't
think
we're
thinking
about
that.
At
this
point,.
D
D
C
B
The
problem
with
that
one
is
that
it
relied
on
our
instrumenting
class
loader
design
class
in
order
to
trick
it
to
load
the
parent.
First.
Sorry.
D
B
I
don't
think
that's
typically,
what
like
buddy
would
something
that
relies
on
yeah,
but
it
could
be
interesting
to
even
just
post
our
experience
over
there
with
that,
because
that
was
a
good.
That
was
a
good
optimization.
D
Yeah,
they
generally
feel
as
if
it's
ideal,
if
the
api
can
be
kept
as
simple
as
possible
for
sort
of
the
common
case
and
then
handle
non-common
cases
with
more
effort
like
these
optimizations
seem
to
be
not
like
skipping.
The
batch
call
back
if
the
instrumenters
aren't
enabled
or
something
it
seems
like,
not
so
common
usage
of
the
api.
D
So,
ideally,
we
wouldn't
then
need
the
stuttering
if
it's
possible.
That
was
basically
my
like,
for
example,
it
could
be
optional
right,
like
if
there's
something
registered,
then
they're
checked.
If
there's
nothing
registered,
the
batch
public
was
always
run.
That
seems
like
one
way
to
still
have
the
simple
mechanics
without
also
losing
the
other
ones.
A
Yeah
yeah,
that's
interesting.
I
think
I
think
then
like
instrumentors,
like
folks,
writing
instrumentation
code,
for
example
in
java
instrumentation,
would
want
to
always
you
know,
use
the
expanded
version
just
because.
D
D
C
A
There's
always
the
chance
that
end
users
would
have
to
want
to
instrument
custom
things,
just
I
kind
of
have
the
question
in
my
head
of
like
how
how
how
often,
though,
if
they
have
to
write
this
once
or
twice
for
an
application,
it
doesn't
really
bother
me
as
much
yeah.
I
have.
D
D
A
Yeah,
so
that's
part
of
what
I've
implemented
here
is
in
is
you
you
can't
record
to
any
instruments
that
aren't
registered,
and
that
was
actually
true.
That's
ex
that's
true
for
the
existing
async
instruments
today,
like
you,
could
do
something
funky
in
your
callback
for
a
single
instrument
today,
where
you
take
the
observable
and
set
it
in
like
an
atomic
reference
and
then
record
to
it
outside
of
your
callback.
A
You
could
do
something
like
that
today
and
we
wouldn't
do
anything
to
stop
you
so
this.
This
tightens
that
up
as
well,
so
that
you
can
only
record
those
and
we
just
log
in
I'm,
not
I'm
not
too
concerned
about
that
case.
D
A
They're
locked
by
default
and
then
right
before
you
call
them
back
you,
you
unlock
the
storages,
call
the
callback
and.
A
A
A
A
Yeah
yeah
they're,
so
if
you
scroll
down
in
the
test
directory
on
the
left
side,
there's
one
that
indicates
it
pretty
well
so
batch
callback
test
that
that's
just
like
you
know
a
simple
demonstration
of
it.
So
this
this
demonstrates
recording
to
like
six
different
instruments
in
a
single
callback,
and
so
you
register
them.
A
You
have
to
you
have
to
create
those
instruments
before
you
register
the
callback,
because
you
need
like
handles
to
record
values
to
okay
and
that's
what's
happening
there.
So.
D
Now
I
see
the
double
long
issue
you
talked
about.
This
does
look
really
hard.
I
wonder
if
there's
a.
A
Way
to
work
around
this,
we
could
give
them
a
super
interface
or
an
interface
that
you
know
they
all
extend
yeah.
C
D
A
B
What
that
was
actually
my
question,
what
you
guys
were
just
talking
about
about
the
intermediate
common
interface,
but
now
I
get
it
right
yeah,
I,
like
the
you,
know,
call
it
just
once
per
because
this
this
is.
I
don't
really
see
too
much
advantage
to
avar.
I
assume
this
is
firearms
here.
Yep.
C
Add
honorary
to
your
comment
like
if
you
don't,
if
you
did,
if
you
didn't
call
ad
that
we
wouldn't
enforce,
I
wonder
if
it
would
be
worth
just
having
an
explicit
method
that
says:
enforce
registered
instruments
only
or
something
like
that.
That
would
you
would
turn
it
on
and
then
it
would
be
enforced.
Yeah.
D
But
after
hearing
the
implementation
of
the
locks
and
stuff,
I'm
not
sure
if
we
can
actually
implement
that
well
anyways.
So
maybe
we
don't
do
that
actually,
but
then
we
would
have
to
unlock
every
single
storage
before
the
callback
and
then
lock
it
back
up.
Then
that
seems
a
bit
yeah
intense
yeah,
so
that
one
might
just
be
too
hard
to
implement
yeah.
I
can
see
the
batch
being
pretty
rare
friend
user.
So
if
it's
a
bit
tedious,
it's
okay,
I
think.
B
Just
tried
a
really
nice
javadoc
on
it
to.
A
A
Maybe
maybe
just
including
a
an
outright
example
on
the
javadoc
to
show
usage,
so
maybe
that
would
clear
it
up.
A
Okay,
we
could
flip
it
too
to
make
so
you
could,
you
could
say,
batch
callback
builder
and
pass
the
runnable
when
you
create
the
builder
and
then,
when
you
click,
when
you
say
build
you
could
we
could
force
you
to
say
all
the
instruments
that
you
want
to
record
do
to
make
it
less
easy
to
miss.
A
C
A
D
A
And
the
go
implementation
of
metrics
considers
this.
This
feature
absolutely
essential,
because
if
they
want
to
use
that
to
you,
you
know
monitor
all
sorts
of
system
level
stuff
in
the
conductor.
A
A
Out
of
this
right
now,
that's
what
I'm
thinking
I
I
I
could
use
some
additional
tests
as
well
like
it's
fairly
well
tested,
but
it
needs
kind
of
some
like
additional
test
that,
like
the
higher
level
testing
things
end
to
end,
but
it's
it's!
It's
really
close
and
then
tweaks
to
the
api,
based
on
anything
that
we
decide.
B
Oh,
I
will,
I
need
to
write,
release
notes,
we'll
see,
might
all
tap.
It
might
all
happen
tomorrow,
which
means
release
the
instrumentation
release
might
not
happen
until
monday,
but
that
is
on
its
way.
D
D
Yeah
we
could
because
one
of
those
tricky
things,
because
we
don't
bundle
jackson
and
our
distro.
We
just
bundle
the
agent
I
could
rebundle
jackson
and
just
to
catch
it
on
our
side.
But
I
don't
want
to
do
that
because
it
sounds
like
a
lot
of
work,
but
in
general
I
guess
if
it
seemed
important
enough
for
aws
to
catch,
then
it
would
be
important
enough
for
upstream
to
patch-
and
I
would
just
push
for
that-
probably
which
I
didn't
understand.
B
D
B
D
D
B
Well-
and
you
like
so
the
way
that
we've
basically
taken
our
application,
insights
team
and
those
are
the
people
who
are
involved
in
open
telemetry
and
those
are
the
customers
and
those
are
which
would
be
like
the
x-ray.
But
it
seems
like
it.
It
would
have
been
like
if
the
x-ray
people
were
supporting
that
right.
Where
you.
D
D
D
C
D
D
A
What's
what's
the
policy
on
on
patches?
Is
it
just,
but
do
you
ever
patch
back
to
versions
besides
the
latest,
we
could.
D
Done
it
once
yeah
like
it's
very,
we
don't
generally
do
it,
but
we
had
that
one
customer
user
who
said
they
don't
use
the
latest
version
as
their
best
practice
because
there
might
be
bugs
so
they
use
the
n
minus
one
or
something.
So
can
we
patch
the
n
minus
one,
and
we
did
do
it
that
time,
but
we
generally
try
to
avoid
it.
B
D
B
But
you
know
what
happened
to
me
yesterday:
secure
security,
I'm
running
a
spring
pet
clinic
on
my
machine
and
all
of
a
sudden,
and
you
know
I'm
monitoring
it
right
and
all
of
a
sudden,
I'm
getting
all
these
jndi
dns
requests
paths
urls,
hitting
my
local
test,
app,
wow
and
yeah.
B
A
It
was
on
my
laptop
what
was
your
laptop
like
if
you
know
whatever
accepting
connections
from
your
whole
network,
was
it
on
zero,
zero?
Zero,
zero?
I
think
I
think
spring
does
that
by
default.
Actually,.
B
B
Anyways
right
yeah
malware
malware,
but
I'm
like:
why
would
they
try
to
rc
a
remote
code
exploit
if
they
already
have
infected
my
machine
right?
That's
true!
Oh
my
gosh,
so
I
got
some
good
advice
from
my
teammate
to
send
it
to.
We
have
a
report.
It
now
system
at
microsoft,
and
so
I
reported
it
to
them
and
they
got
back
fast
and
we're
like
oh
yeah.
That's
us!
C
B
Yeah
they
were,
you
know,
just
testing
for
vulnerability,
vulnerable
java.
B
D
B
Yeah
I
started
I
spent
when,
after
I
saw
those
requests,
I
spun
up
my
my
virus
checker
and
did
started
a
full
scan,
but
well
first
I
did
a
quick
scan,
but
then
I
was
like
oh
fine.
I
got
malware.
I
was
so
mad.
B
B
Hours
so
I
was
like
forget
it.
I
canceled
after
a
while.
D
Speaking
of
hours
as
part
of
my
merging
those
projects
into
the
main
build,
I
did
look
at
the
benchmark
overhead
project.
Also.
My
experience
mentality
is
smart
enough
now
that
if
I
just
open
that
folder,
maybe
it'll
only
index
the
requirements
of
that
folder
and
first
I
was
wondering
if
it's
possible
just
open
a
subfolder
intellij,
and
it
was
so.
D
A
D
D
D
A
Yeah
I
mean
we,
we
haven't
done
anything
with
that
since
we
set
it
up,
I
mean
there's
yeah
the
whole
one
of
the
reasons
for
it
was
so
that
we
could
have
examples
for
the
agent
and
the
sdk
side.
D
D
You
need
to
do
it
someday
yeah
for
trust,
your
point,
since
the
javadocs
is
not
the
sdk
repo,
it's
also,
in
effect
an
independent
build.
So
it's
that
is
being
tested.
Is
that
a
quick
statement.
D
D
B
B
Cool
anything
else
that
we
needed
to
touch
on.
D
Is
this
also
instrumentation?
You
can
do
it
and
I
ask
mostly
because,
like
of
course
span
name
is
still
not
as
big
a
deal
as
root,
but
it's
also
supposed
to
be
relatively
low
cardinality,
so
we've
solved
the
problem
for
root,
but
not
spending,
and
if
that's
not
solvable,
maybe
this
is
just
not
appropriate
instrumentation.
That
was
the
thought
process.
C
B
D
D
D
B
D
B
D
B
Which
we
disable
in
our
by
default
yeah,
but
I
do
know
they're
popular.
They
can.