►
From YouTube: 2021-08-10 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
Oh,
it's
so
laggy,
I'm
gonna
switch
over.
I
was
testing
this
with
edge,
for
some
reason
seemed
like.
B
C
D
C
C
D
C
Cool
well,
we
got
the
item
number
one
already
cleared
up.
That
explains
why
it's
so
laggy
to
edit
this
document.
C
It's
just
too
big,
all
right,
since
we
have
ludmila
here.
Let's
talk
about
suppressing
instrumentation
and
material,
could
you
explain
the
the
spring
integration
rabbit
mq
issue
like?
Is?
It
is
spring
integration?
Creating
a
server
span
and
rabbitmq
is
creating
a
consumer
span.
B
Detects
whether
there's
already
some
sort
of
local
root
span,
and
if
it
if
there
isn't
it
adds
a
new
custom
consumer.
E
B
B
C
Actually
humor
and
is
it
consumer
like
then
receiving
pulling
a
message
down
and
and
processing
it?
No.
B
It's
so
spring,
integration
is
what
was
it
called
enterprise
something
patterns
and
the
rest
integration
pattern
is
library
right,
so
it
kind
of
works
like
message
boxes
connected
with
pipes
and
for
it
it
doesn't
matter
whether
it's
on
the
receiving
or
on
the
sending
side.
The
code
always
looks
the
same,
so
we
decided
that
it
should
only
propagate
context
and
create
consumer
expense
if
there's
none
already
on
the
receiving
side.
B
But
if
you
have
the
spring
into
spring
integration,
instrumentation
plugged
in
in
both
sides
or
in
all
of
your
applications,
then
obviously
the
instrumentation
will
will
get
triggered
every
time
we
use
in
spring
integration
and
if
you
have
sending
sites,
for
example,
you've
got
a
some
kind
of
an
http
endpoint
that
does
something
and
puts
something
else
onto
a
queue
using
spring
integration
and
the
same
code
as
on
the
consumer.
Side
will
be
executed
and
possibly
it
will
create
a
consumer
span.
B
That
is,
it
would
create
a
consumer
span
if
I
hadn't
added
this
server
or
similar
check.
So
what
this?
What
this
pr
actually
did
was
remove
that
check
and
change
the
string
integration
instrumentation
to
just
look
up.
If
there's
any
spam
in
the
context,
because
if
it's
only
supposed
to
add
a
new
local
response,
we
we
could
as
well
check
if
there's
any
spam
in
the
codex.
If
there's
none,
then
yeah,
we
can
add
any
one.
E
So
I
don't
know
how
spring
integration
works,
but
presumably
doesn't
look
into
rabbit
and
q
message
headers
and
if
ribbit
and
q
instrumentation
would
kick
in
and
create
a
consumer
stand
with
the
context
from
the
message
and
spring
integration
will
create
something
else.
The
server
spent
without
context
is
this
the
case
here.
Can
it
be
this
problem
that
we
will
break
the
trace
because
we
didn't
create
the
proper
consumer.
B
I
don't
think
so
because
if
so,
the
spring
integration
instrumentation
is
kind
of
a
fallback
instrumentation
like
if
you
didn't
have
a
properly
working
gravity,
instrumentation
and
it
didn't
create
a
consumer
spam.
Then,
when
that's
the
situation,
when
spring
integration
comes
into
action
right,
so
it
takes
the
headers
from
the
message
and
creates
a
new
consumer
spam.
But
if
there
is
a
properly
working
messaging
instrumentation,
whether
it's
rabid
or
kafka,
or
something
else,
it
will
create
a
consumer
span
on
its
own
and
when
the
spring
integration,
instrumentation.
B
C
C
B
B
So
that
was
just
my
example,
but
imagine
that
you
before
the
spring
integration
producer,
you
have
some
kind
of
an
http
endpoint,
but
it
might
be
as
well
another
messaging
consumer,
so
your
rpc
endpoint,
it
doesn't
really
matter.
You
have
a
another
sort
of
top-level
thing
that
has
some
business
logic
and
actually
tries
to
put
the
message
into
the
queue
right.
What
I
wanted
to
prevent
with
with
this
server
or
consumer
check,
was
for
this
spring
integration
producer
part.
B
C
B
Yeah
so
because
spring
integration
is
only
supposed
to
create
a
consumer
span
right,
so
essentially
a
sort
of
local
root
span.
I
can
simplify
this
check,
this
server
consumer
and
just
check
if
there's
any
spam
in
your
contacts,
because.
E
B
Then
it
means
that
something
else
has
already
instrumented
this.
This
will
say
threat
local
threat
of
protection,
and
we
probably
don't
need
to
propagate
the
message
from
the
no
cooperate.
The
context
from
the
message,
because
some
something
else
has
probably
already
done
that
or
it's
a
server
span
or
I
know
scheduled
job
that
did
this
and
there's
a
need
to
create
a
new
customer
spam.
C
Right
so
from
a
suppression
perspective
I
mean
okay,
it
does
the
spec
even
allow
nested
servers,
consumers,
consumers
servers.
Do
we
nested.
D
C
So,
from
a
nesting
perspective,
so
okay,
so
we
don't
want
so
we
are,
though,
in
here
we
are
still
suppressing
nested
consumers.
E
E
B
C
C
Do
you
does
a
local
route
I
mean?
Is
this
not
even
related
to
this
topic.
C
So
we
have
so
my
thought
would
be
that
that
could
potentially
replace
the
the
server
enrichment,
in
particular
updating
the
span.
Name
like
we
have
a
lot
of
instrumentation
that
gets
the
route
and
updates
the
span
name
of
the
server.
E
C
Which?
Users.
D
B
D
C
Okay,
so
let's
talk
about
what
is
remaining
any
open
things
remaining
for
this
pr.
E
C
Oh,
did
you
update
those
tips?
Oh
nice,
our
build
the
github
actions
have
been
drowning
all
day,
we're
like
10
that
queued
for,
like
10
hours
back.
E
C
Cool
one
thing
that
so
what
I
had
kind
of
called
the
renamed.
This
outgoing
propose
renaming
this
outgoing
key
versus
client
key
to
cover
both
client
and
producer.
C
If
that
makes
sense,
yep
the
other
is
the
span
key.
I
guess
we
do
have
more
than
just
okay.
E
C
C
And
that's
our
existing
behavior
already
to
suppress-
and
I
know
our
our
back
end
will
get
more.
We
will
prefers
just
having
a
single
outgoing,
so
I
I
think
we
could
have
the
with
the
layering.
Now
we
have
that
option
to
get
the
multiples.
B
C
C
So
in
that
case,
but
this
one
felt
a
little
odd
to
me
if
we
put
this
on
the
instrumenter
builder
and
like
it's
opt-in
per
instrumenter
builder,
when
it
feels
more
like
a
global
like
everybody
needs
to
be
in
alignment.
C
Oh
yeah,
and
then
I
was
wondering
on
the
name
here
like:
is
it
gonna
be?
Is
it
outgoing
span
so
like
does
this?
I
think
this
only
applies
to
outgoing
spans
or
client
producer
spans.
Is
that
right,
ludmilla.
E
Yeah,
at
least
that's
my
understanding
that
it
should
be
the
piece,
so
I
can
change
it
to
something
like
that.
Outgoing
spend
suppression.
E
E
C
E
C
C
Okay,
oh
yeah,
the
span
suppression
strategy.
C
C
Oh
the
multiple
attribute,
extractors
honorable,
I
know
we
had
talked
about
the
precedence
ordering,
but
it
actually
seemed
fairly
straightforward
to
just
take
the
list
of
them
and
for
if
there's
multiples,
then
for
suppression
only
suppress
it
if
all
of
the
types
exist
already
and
then
when
we
add
it
to
the
context,
just
add
it
to
the
multiple
span
keys.
D
And
get
it
so
this
is
assuming
that
one
instrumentation
might
generate
both
the
db
span
and
http
spanner.
C
Right
so
when
making
the
decision
to
suppress,
it
checks
both
of
those
and
if
either
one
is
missing
from
the
parent
context,
then
don't
suppress
it
and
when
storing
it
into
context,
it
stores
the
same
span
against
in
both
of
those
on
both
http
and
db.
So
it
will
suppress
both
of
those
types
further
down.
D
D
D
C
Well,
because
we
have
a,
we
have
a
problem
right
now
where,
because
the
azure
sdk
yeah,
so
we
have
a
problem
but
we're
working
on
it.
D
D
C
D
C
So
yeah
so
john
has
been
trying
to
come
up
with
a
better
library,
instrumentation
that
actually
works
for
okay
gdp
and
it
sounds
like
the
magic
is
3.12.
D
D
D
D
C
D
Yeah,
that's
why,
even
if
we
really
so,
I
think
the
more
interesting
thing
is
like.
Maybe
we
do
have
to
define
our
target
release
like
our
compatibility
guarantees,
because
two
years
old
enough
is
three
years
old
enough.
I'm
sure
we
can
define
some
number
of
years
and
just
always
refer
to
that
rather
than
making
this
call
every
single
time.
C
I
think
it's
gonna
be
hard
yeah,
because
it
depends
on
how
much.
D
E
C
Pretty
old:
well,
we
even
support
you
know
2-0
of
that,
but
but
it's
also
just
in
so
many
apps
and
for
library,
instrumentation
right.
Sorry,.
D
C
See
but
that's
not
true
that,
because
a
lot
of
auto
online
for
library,
instrumentation
users,
I.
E
D
C
This
was
only
out
for
a
month
before
one
six
came
out
yeah.
We
also
already
had
a
floor
so
like
from
a
support
perspective,
it's
easier
for
me
to
say:
hey.
We
have
a
floor
right.
I
already
have
to
deal
with
people
who
I
already
have
to
ask
customers.
What
version
are
you
running,
whereas
I
like
the
ok
http
story
is
better
for
me
from
a
support.
C
C
And
I
think
we
may
also
need
like
if,
if
we
did
want
to
go
forward
with,
you
know
bumping
this,
for
example,
I
would
definitely
ask
nikita
to
you
know,
get
sign
off
from
splunk's
product
management,
because
I
know
for
like
for
me.
I
would
probably
end
up
patching
it
back
into
our
distro
and
I'm
assuming
splunk
probably
would
have
that
pressure
also
from
customers.
D
A
C
C
Capturing
that,
then
we
could
answer
those
questions
that
would
be.
A
C
E
A
D
A
D
C
Do
you
know
android
what
the
plan
is
for
the
one
five
release
for
sdk.
D
C
Thought
yeah,
so
this
one.
C
D
B
I
don't
think
that
we
have
a
lot
of
copy
paste.
I
mean
most
of
these
methods
that
config
properties
define
are
already
present
in
our
config,
the
only
one.
That's
I
mean
we
have
our
own
list
or
map
property
parsers,
but
the
only
thing
that
the
only
thing
that's
missing
is
duration
parser,
but
I
think
that
that
would
make
sense
to
add
the
java
agent
too,
at
the
very
least,
because
the
spline
destroyer,
probably
wouldn't.
D
D
B
D
B
D
B
D
B
D
B
The
issue
that
this
pr
tries
to
solve
just
had
my
few
prepositions
how
to
solve
them,
and
then
john
just
said.
D
D
I
think,
on
top
of
this,
we
could
just
go
ahead
and
add
that
map,
so
that
we
can
reuse
the
default
config
properties
as
well
so
yeah,
I
think
so,
since
we
solved
this
other
issue,
I
don't
see
anything
wrong
with
merging
your
change
and
I
think
it's
a
good
change.
In
general,
we
could
probably
still
consider
adding
the
map
approach
as
well
just
to
reduce
the
duplication.
C
D
C
D
D
C
And
then
the
other,
this
one
would
be
nice
for
our
extension
stuff
because
we
can
drop
the
export.
I
think
we,
this
allows
us
to
drop
the
exporter.
C
C
D
C
Okay,
I
know
our
prs
got
all
backlogged
today
because
of
ci
anything
that.
D
Is
there
any
big
difference
in
approach
here?
Is
it
quite
similar,
because
I
know
at
one
point
we
were
like
not
using
configurations
and
then
we
had
to
backtrack
and
go
back
to
configurations,
I'm
just
sort
of
where
we
ended
up,
because
I
didn't
follow
those
pr's
in
detail
that
were.
D
A
A
D
D
A
But
before
we
had
several
classes
in
java
agent,
tooling
module
and
we
like
our
gradle
plugin,
switched
to
like
agent
touring
class
right
away.
It
load
lo
create,
creates
a
new
class
loader
and
switch
to
like
runtime
class
in
agent
tooling.
A
D
A
C
Well,
you
all
enjoy
your
days,
I
know
with
mill
and
I
are
late
at
night.
Yes,
yes,.