►
From YouTube: 2021-08-31 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
A
C
D
A
That
doesn't
mean
it's
hard
to
work
around
by
fixing
yourself
script
or
by
setting
otel's
service
name,
because
that's
usually
what
people
are
setting.
That's
what
I
did.
I
just
changed
hotel
service
name,
which
is
a
new
variable,
oh
great
yeah.
So
that
was
my
work
around.
So
I
actually
don't
need
a
patch
for
myself.
B
I
mean
the
we,
I
don't
think
we
need
to
cherry
pick
this
change
again,
because.
A
A
A
B
Yeah,
I
have
no
issue
hitting
the
hitting
the
patch
button
for
it.
It
fits
if
it's
our
definition
for
patching
and
yes,
it
will
I'm
very
curious
if
this
is
going
to
work,
because
the
one
of
the
topics
in
this
morning's
maintainer.
A
B
B
A
B
Yeah
the
so
like
what
I
did
here
is
I
overrode
and
I
removed
the
cla
protection
and.
B
B
Is
that
it
because
the
cncf
is
not
replying
to
daniel
they're,
not
giving
him
an
answer?
I
don't
know
why
this
page
always
they're
going
to
take
a
vote
for
temporarily
to
say.
Yes,
it's
fine
to
create
these
for,
like
gh
pages
and
potentially
release
branches,
so
yeah.
B
No,
I
looked
it
up
because.
D
B
Also,
okay,
star
star
is
supposed
to
be
any
number
of
paths-
oh
yeah,
but
then
the
slash.
So
the
idea
would
be
that
it
has
at
least
one
slash
in
it,
but
this
was
definitely
applying
because
I
needed
to
do
the
gh
papers
to
work.
A
It's
a
bit
different
than
gradle,
I
think
do
you
know
if
you
want
v
startup
start
at
x
would
then
take
precedence.
B
It's
gonna
give
me
a
unicorn.
I
swear
this.
I
don't
know
something
about
this
page.
Just
is
oh,
yes,
I
got
one
okay,
quick,
the
star
star.
B
A
v
star,
I
think,
is
all
we
need
and
we're
not
going
to
have
any
rules.
Yes,
I
am,
I
say
now
come
on
no
unicorn.
B
Oh
hey,
it
did
create
it
though.
Okay
and.
C
Thus
one
asterisk
match
exactly
one
character
or
one
or
more,
because
it's
if
it's
exactly
one
that
we
probably
don't
have
two
letter
branches.
B
Yeah
yeah
so
yeah.
I
think
you're
right
like
because
I'm
wondering
if
it's
like
that
refs
heads
thing,
but
then
I
don't
know
why
my
gh
pages
one,
this
branch
is
covered
by
this
star
star
v.
Star
applies
to
star
star
star
star.
I
just
wondered
if
it
updated.
A
B
All
right,
I
can
mess
with
this
tomorrow
and
try
to
test
out
yeah
before
I
click
the
patch.
B
B
Because
I
noticed
the
this
in
the
java
sdk,
it
looked
like,
we
only
were
capturing
the
one
attribute
and
I
think
there
were.
B
Md
there's
a
few
others.
That
would
be
interesting,
but
my
understanding
is-
and
I
don't
know
if
either
of
you
know-
that
the
these
other
ones
would
require
making
a
like
a
connection
to
a
local
service,
local
kubernetes
service,
to
scrape
the
data.
A
A
B
Was
looking
at
was
maybe
azure.
D
B
Has
something
similar
then,
because
I
was
looking
at
some
dotnet
code
that
was
scraping
endpoint
some
like
api
rest.
B
B
Me
no!
No!
I
I
want
I'd
like
to
collect
I'd
like
to
enhance
the
the
sdk
container
resources
to
capture
that
now.
You
said
that
ecs.
So
do
you
have
that
in
the.
A
And
you
see
to.
B
B
So,
do
you
have
a,
is
it
only
java,
11
or
you
have
an
alt.
B
B
B
B
A
look
at
the
I'll
take
a
closer
look
at
that.net.
Basically,
some
somebody
did
something
for
net
not
for
hotel,
but
for
the
azuremonitor.net
sdk.
Somebody
asked
if
we
could
do
something
similar
for
java.
A
B
B
And
what
did
I
I
put
this
on
the
agenda
while
remove
unnecessary
url?
Wait.
B
Put
one
your
one
link
to
a
different,
let's
see,
but
we
can
probably
get
there.
This
was
the
one
today
that
I
closed
by
accident.
That
was,
I
thank
you
matt.
A
I
had
my
idea
of
always
capturing
url,
but
I
was
suspicious
that
you
don't
like
that
idea
for
performance
reasons.
Maybe
that's
one
point
to
chat
about,
so
I
had
my
at
the
end
of
my
comment.
I
think
there
were
like
I
like
either
three
or
four
four
is
the
easy
one,
but
that's
the
one
that
has
performance
stuff.
A
C
There's
this
sort
of
like
hierarchy
of
preferred
ways
to
have
these
attributes
in
server
and
client
and
in
this
in
the
client.
It's
http
url
it's
in
the
first
place
and
the
server
is
host
target
and.
A
B
It's
okay,
I
mean
I
haven't,
did
I
see
I
think
it
left
my
well
didn't
show
up
as
a
particular
hotspot
after
that.
A
Yeah
because,
as
far
as
full
url
for
claims,
like
I'm,
remembering
the
focus
http
instrumentation,
which
has
that
extremely
gnarly
stuff,
but
either
decision
doesn't
fix
that
early
in
this
right.
That's
just
probably
we're
screwed
either
way.
A
B
A
A
We'll
add
we'll
add
two
classes
not
replace
right
away
just
to
prevent
a
huge
migration,
but
http
server
attributes
will
make
sure
root
is
also
not
part
of
the
client
like
we
can
fix
that
bird.
At
the
same
time
there
there
might
be
other
ones,
but
I
think
root
is
probably
the
only
one
where
it's
server
only
so
that
will.
C
C
A
B
Oh,
did
you
see
this
honorable?
I
was
curious
if
you
had
thoughts.
A
B
Right
right,
yeah,
just
because
I
I've
seen
that
that
array-based
context
a
lot
in
the
profiles
and
I'm
a
little
concerned.
That's
gonna
get
worse,
as
we
start
putting
more.
I
guess
we
won't
those
client
ones
at
least
those
those
low.
B
A
A
D
A
B
I
see
and
then
we
would
have
to
replace.
I
mean
it's
a
little
bit
more.
We
have
to
replace
that
whole
object.
Each
time.
A
A
But
again,
that's
still
one
lookup
and
we
could
get
down
to
zero
lookups.
If
you
put
it
back
down
to
the
context
that
might
still
be
worth
it
but
yeah,
but
no
that's
a
good
thought
for
at
least
it
will
improve
the
library's
implementation
performance
at
the
same
time.
So
technically
probably
want
to
try
that
first
and
then
go
from
there
for
the
agent.
A
C
A
A
B
B
B
B
Yeah,
it's
really
looping
through
that
array
and
testing
the
keys
and
yeah.
I
it's
so
hard
to
when,
like
with
micro
benchmarks
versus
sort
of
what
I'm
running,
is
a
little
bit
more
macro
benchmark.
B
Oh
this,
our
the
most
the
most
my
best
guess
is
this
one
onorag.
B
B
Yeah,
because
it's
the
and
it's
not
that
you
can
look
at
the
runs
and
it's
not
the
smoke
tests
that
are
the
long
ones.
It's
the
the
test
on
the
integration
tests.
A
A
B
C
They
were
exposed
in
some
non-internal
interfaces.
I
think
so
we
were
international
yeah.
I
think
I
remember
something
like
that:
yeah
yeah,
we
are
implementing
operators
as
subscribers
or
I
don't
remember
even
what
it
was,
but
it
was
like
a
standard
custom,
customization
interface
and
refer
to
some
internal
plus
yeah.
A
A
A
A
A
C
A
B
Well,
maybe
I
mean
at
least
maybe
there
can
be
a-
I
mean,
no
like
okay,
so
that
having
a
common,
you
know
where
we
have
those
copy
pastes.
In
some
cases
we
have
extracted
out
like
a
common
piece.
C
C
Well,
I
don't
remember
what
but
there's
a
large
part.
That's
a
common
and
we
they're
just
customized
by
different
advice,
classes.
B
In
some
cases,
yeah
like
the
netty
one,
I
think
it's
generally
good
it
does
make.
I
was
kind
of
wondering
well
when
we
want
to
split
out
the
contrib
and
we
want
to
say
we
wanted
to
only
keep
the
latest.
C
B
All
right
anything
else:
how's
how's
life
in
the
sdk
world.
Honorable,
I
saw
some
good.
I'm
excited
for
the
next
sdk
release,
with
some
good
performance.
A
Oh
yeah,
I
I've
been
going
through
the
flame
graphs.
I
don't
know
if
there's
anything
left,
that's
obvious.
I
think
I'm
sort
of
done
with
what
I
can
do
like
the
nano
tank
was
the
last
one's
like
that's
the
one
I
left
for
last
visit.
Isn't
that
important
and
now
even
that's
up
to
my
stuff.
B
A
B
A
B
B
So,
and
part
of
the
reason
is
because
the
with
the
macro
benchmark
right,
I
ideally
our
the
agent
code-
would
only
be
a
small
like
very,
very
small
fraction
of
the
the
profile,
and
so
it
it
tends
to
be
really
hard
to
find.
It.
A
D
B
And
so
yeah,
so
this
is
what
I
I
adapted
this
from
what
I
used
when
profiling
glow
root,
which
takes
the
jfr
file
and
extracts
out
the
interesting
agenty
things.
B
Not
a
it
doesn't
produce
a
flame
graph
that
would
be
cool.
It
just
produces
a
node,
a
a
tree
profiling
tree
and
dumps
that
out.
A
B
B
A
A
B
Here
we
can
see
that
there
it's
in
the
thread,
context
impul
yeah
as
usual:
yep
yeah
yeah,
starting
timers.
There
you
go
for
your
nano
time,
yeah!
That's
definitely.
A
A
B
D
B
Think,
like
I
had
one
bad
experience
with
it
on
some
linux
system
that
I
can't
remember
what
the
resolution
was.
I
was
doing
some
benchmarking.
I
think
I
was
doing
some
benchmarking
and
in
probably
ec2
there
was
something
weird
I
don't
remember,
but
yeah.
If
I
recall
this,
it
wasn't
bad.
It
was
like
30
nanoseconds
for
a
call
or
something
which
you
know
adds
up
if
you're
using.
A
B
Timers
yeah
yeah,
but
it
shouldn't
be.
It
shouldn't,
be
an
issue
for
spans.