►
From YouTube: 2021-01-22 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
C
D
D
D
C
Oh
yeah,
what
do
we
do
here
as
I
I
share
and
meeting
notes
and
sorry,
I'm
a
little.
Actually
I
just
woke
up
from
a
nap.
I
will
admit
it.
C
B
Well,
the
topic
of
the
day
for
me
is
going
to
be:
how
difficult
will
it
be
to
put
in
a
flag
to
allow
span,
nesting,
quote-unquote,
illegal
span,
nesting
for
clients
and
servers
and
etc,
etc?.
C
Yeah,
so
if
you
look
at.
B
C
So
like
here,
we
check
if
it's
in
a
client
span,
so
maybe
this
I
think
this
would
more
or
less
be
where
you
want
to
say
false
or
maybe
rename
to
something
that
makes
sense,
that
it's
configured,
suppress,
client.
B
C
Okay,
yeah,
maybe
you
can
at
the
same
time,
consolidate
us
to
use.
I
mean
I
think
that
it
should
be
calling
within
client
span
and
with
client
span
everywhere.
Well,.
B
C
B
B
I
may
I
may
tackle
that
tomorrow,
or
at
least
I'll
put
in
a
pr
that
will
refactor
everything
to
use
to
use
that
method
and
rename
it
to
something,
suppress
yeah
or
something
like
that
and
then
maybe
make
it
optional
in
this
in
a
separate
step.
B
D
It's
true
I
mean
x-ray
breaks
with
nested
coin
spans.
Oh.
B
D
C
D
C
Totally
supported
that
change,
it
made
a
lot
of
sense
for
our
usage.
Also
we're
the
application
map
is
like
our
holy
grail
so
having
you.
A
C
A
clean
application
map
is
with
better
naming,
is
very
helpful.
Yeah.
B
D
B
B
C
Yeah
at
some
point
I
like
I
want
to
make
a
option
in
here
to
make
internal
spans
optional.
So
just.
C
C
Have
customers
who
are
very
we
charged
by
the
ingestion
gigabytes,
so
they
a
surprising
number
of
them
care
about.
Look
at
the
telemetry
and
they're
like
this
is
not
that
important.
I
don't
wanna.
D
B
B
B
B
Right
right,
yeah,
yeah
yeah.
Actually
I
was,
I
was
thinking
that
one
of
the
one
of
the
knock-on
effects
of
that
bug
that
I
found
today
I
mean
bug.
I
don't
know
whether
it's
a
bug
or
I
don't
know
whether
it's
a
bug
or
anyway.
So
it
certainly
was
surprising
and
weird,
as
I
wonder
whether
we
really
should
separate
propagation
from
span
creation,
because
that's
a
case
where
we
really
wanted
to
do
propagation,
but
we
wanted
to
maybe
suppress
the
second.
The
second
client
span.
B
B
D
B
D
C
B
D
B
B
D
B
D
D
D
B
D
That
doesn't
end
that
only
can
add
attributes
and
read
from
it
right.
So
this
was
one
idea,
which
is
the
idea
I
like
the
best,
but
we
were
going
to
discuss
it
on
a
spec
issue
which
it
was
december
and
it
was
too
lazy
to
do
it.
But
now
it's
january
we
can
discuss
that
a
bit
more
of
how
to
properly
handle
these
cases.
B
Yeah,
well,
especially,
I
just
saw
yeah
matteo,
so
was
it
mateo
she
put
in
a
pr,
no
was
jacob
put
in
the
pr
to
do
aws,
sqs
handling,
which
creates
some
other
weird
nesting.
That
is
yeah
not
in
the
spec.
D
C
Honorag
did
we
ever
understand
the
issue
with
the
the
signing
the
aws
signing
of
the
http
request,
because
that
for
a
while,
we
thought,
maybe
that
would
so
john
we
we
had
so
we
have
some
weird
stuff
in
the
neti
instrumentation
that
if
it's
an
aws
call
to
aws,
we
specifically
don't
like
we
check
if
it's
an
aws,
backend
or
something
url
and
we
will
not
propagate
because
the
whole
http
request
is
signed,
and
so
we
breaks
the
segment
we
break
the
signature.
Oh.
D
Like
our,
I
think
we
tried
removing.
I
brought
it
back
in
because
because
the
net,
because
in
a
sense
it's
not
a
different
thread,
we
can't
propagate
context
or
something
like
that.
D
D
C
D
C
C
Yeah,
I
love
that
you
guys
are
starting
to
get
bored
over
there
in
the
sdk
land,
closing
everything
down.
B
Well,
it's
because
I
want
to
keep
things
stable
for
1.0
and
interesting,
so
I
don't
know
if
anyone's
interested
in
metrics
or
if
we
want
to
save
that
to
the
end.
But
some
interesting
outcome
from
today's
metric
sig,
but
maybe
jason's,
is
probably
more
important
than
talking
about
that.
C
D
C
Oh
yeah,
this
was
the
one.
I
think
that
where's
the
one
oh,
this
is
the
one
you
were
asking
about.
D
B
Well,
is
this
the
is
this
the
tls
issue.
D
B
D
C
C
Yeah!
It's
weird
that,
because
I
feel
like
we
wouldn't
have
we
didn't
murder
like
it
feels
like
something
changed
on
the.
A
The
only
other
thing
I
had
was
whether
or
not
there's
been
any
talk
of
apache
cxf
has
that
ever
come
up.
A
A
B
D
C
It's
been
there
a
lot,
I
feel
like
it's
been
there
a
long
time.
It's
like
the
soaps,
isn't
it.
I
thought
it
was
a
soap
thing.
Xml,
soap
thing,
yeah,.
A
C
D
B
So
I
missed
the
first
15
minutes
of
the
meeting,
because
I
always
do
for
some
reason.
I
don't
I
don't
know
why.
But
I
have
a
hard
time
remembering,
even
though
the
little
thing
is
in
the
upper
corner,
the
alert
is
in
the
upper
corner
of
my
screen.
I
somehow
ignore
it,
but
I
asked
tyler
yawn
midway
through
the
meeting
like
what
did
I
miss
and
it
sounds
like
fallout
from
last
week's
marathon
workshop.
B
B
B
B
D
B
Anyway,
so
that's
that's,
my
update
is
that
I
think
we're
still
quite
a
ways
away
yeah,
and
so
I
think
we
have
there's
a
there's
still.
I
guess
it
was
a
pretty
good
question
and
I
actually
brought
this
up
in
jason's.
Pr
like
it
would
be
really
nice
to
actually
record
metrics
about
the
size
of
that
queue
or
the
size
of
the
cache
right
like
if
somebody's
using
this.
If
we
have
this
cash
in
here,
it
might
be
really
nice
to
have
some
supporting
like
supportability
metrics
about
the
size
of
caches.
A
B
I
mean,
I
think
we
you
can
make
up
whatever
you
want
cool
yeah.
I
don't
think
there's
any.
I
don't
think
we
have
very
many
metrics
naming
conventions
at
this
point.
D
C
A
A
B
C
B
B
B
D
D
I
wonder
if
we
think
it's
more
explicit,
like
like
I
forgot
about
the
open
tracing
with
0.18
and
all
that
stuff,
but
people
did
use
those
a
lot
and,
like
tommy,
brought
that
up,
and
I
was
like.
Oh
that's
right.
I
forgot
about
that.
People
do
use
it
and,
like
I,
almost
just
don't
want
to
publish
anything.
B
D
D
C
B
C
I
I
also,
though
I
like
you
know
I
like
logging
stuff
at
startup.
I
do
think
a
lot
of
people
see
that
and
if
you
are
seeing
these
kinds
of
requests,
that
would
probably
be
you
know
my
guesstimate
would
be.
That
would
cut
it
in
half.
B
B
D
B
D
B
Speaking
of
not
that
but
getting
ready
for
1.0
did
you
see
my
comment
on
the
1.0
discussion
about
the
current
state
of
the
bomb
and
whether
that
looked
okay
to
you.
D
D
B
Like
do
we
have
that's
one
thing
where
I
just
don't
know
if
we
have
any
users
yet
and
I
feel
like
the
other
things
like
we're
all
using
all
the
time
like.
I
know
that
jaeger
is
being
used
a
lot
in
the
proto
and
resources
and
all
these
things
are
being
used
a
lot,
so
the
kotlin
one
is
the
only
one.
I
just
don't
know.
D
I,
like
the
acre
remote
like
again,
we
would
support
anything
as
long
as
we're
sort
of
confident
the
api.
That's
the
most
important
yeah
and
jaeger
remote
sampler
right
after
taking
a
look.
D
I
don't
yeah
exactly,
I
don't
see
the
api
breaking
much
and,
if,
like
there
are
bugs,
we
already
know
the
bugs,
but
bug
or
no
bug
is
not
a
condition
for
1.0.
It's
api
stability
and
so
kotlin.
Also
in
like
that's,
when
I
looked
at
it
in
that
lens,
like
the
unit
tests
seem
fine,
there
might
still
be
bugs,
of
course,
but
the
api
I
mean
we
do
need
an
as
context
element
somewhere
like
the
api,
seemed
same
so.
D
B
Yeah,
there's
a
there's.
This
guy,
named
marshall,
who
I
met
at
j
crete
a
year
and
a
half
ago,
was
a
big
kotlin
user
who
helped.
He
wrote
a
bunch
of
custom,
new,
relic
instrumentation
to
support
kotlin
and
I
know
he's
been
I'll.
Try
to
ping
him
on
getter
or
somewhere,
because
I
know
he's
he's
around.
D
D
B
C
And
then
logging
for
sdk
stability-
I
mean
this
does
not.
This
is
not
considered
a
telemetry
guarantee
kind
of
thing
for
the
logging.
B
Yeah
yeah
we're
not
publishing
it
as
a
I
mean
people
could
might
they
may
use
it
as
a
source
of
telemetry,
but
we're
certainly
not
publishing
it
as
a
source
of
telemetry.
B
D
It
has
metrics,
so
we
went
with
the
compile
only
yeah,
so
yeah
I
was
going
to.
We
should
merge
this
trace
into
otlp.
We're
not
publishing
any
hyphen.
D
B
Yeah
I
mean
that
one
originally
we
split
up
just
is,
I
think
the
otp
is
just
a
yearly
dependency
right.
Yeah.
B
B
D
B
B
C
C
B
A
D
B
C
Yeah,
I
don't
think
this
is
the
only
way
I
think
in
this
case
there's
it
feels
like
there's
I'm
a
little
confused
why
it
needed
to
go
this
route,
so
I
actually
need
to
clone
and
play.
D
C
So
we'll
get
we'll
get
a
we'll
we'll
converge
on
a
span
name
for
servlets,
eventually.
D
D
A
D
C
And
then
oh
yeah
kind
of
a
recap
of
the
discussion
that
we
had
on
to
on
tuesday
when
your
wednesday
morning
about
java
agent,
versioning
java
agent
1.0
and
what
it
means.
A
C
C
D
I
think
so
why
is
it
so?
Common
span
names
yeah,
so
spam
name.
We
talked
about
that
the
hardest
one.
C
C
Yeah,
I
think
that
is
yeah
yeah
edition.
I
do
yeah,
hopefully
we'll
cancel.
D
D
C
So
just
other
thoughts,
for
you
know,
I
mean
if
we
think
we're
going
to
bump
major
version
constantly.
Then.
A
D
B
C
D
Think
since
their
preconditions,
breakage
they're
more
vigilant
about
not
doing
that
ever
again,
so
they're,
like,
I
would
say,
they're
somewhere
yeah
in
the
sense
that
that's
the
minor
version
that
thing
that
they
treat
as
the
major.
D
C
D
C
B
I
don't,
I
would
wonder,
you
know
the
people
who
could
answer.
This
is
the
people
who
did
open
tracing.
What
is
that
open
tracing
agent
called.
B
D
D
C
D
C
C
C
One
of
the
major
like
the
drivers
were
sort
of
like
not
breaking
dashboards
and
alerts.
D
D
Like
it's
more
transparent,
like
yeah,
we're
not
going
to
break
their
dashboards
in
a
minor
version,
but
I
think
the
point
of
that
was
to
the
user
experience,
not
the
versioning
scheme
that
we
use,
and
so
I'm
trying
to
understand
what
the
impact
on
the
user
is
like.
If
we
bump
the
major
every
three
months.
D
D
Yep
and
if
there's
no
way,
of
course,
there's
no
way,
but
we
could
like
telemetry
bridge,
like
I
mean
I'm
sure
many
many
if
statements
can
make
anything
possible
somehow
I'm
guessing
flags
yeah
like
a
million
flags,
so
which
might
be
a
worse
user
experience
than
major
versions.
But
I
don't
know
if
we've
thought
it
through
enough
to
have
a
good
answer
for
that.
C
Yeah,
you
know
it's,
I'm
kind
of
I
haven't
seen
this
be
as
big
of
a
problem
like
I
know
at
our
with
our
agent.
I
have
changed
the
span
names
multiple
times
in
the
year
and
a
half
that
I've
been
here
and
haven't
gotten.
I
have
had
complaints
that
the
new
span
was
not
capturing
as
much
information.
C
Like
this
band
name
used
to
be
the
full,
have
the
full
url,
which
was
a
also
a
horrible
aggregation
experience.
So.
C
D
C
D
D
Yeah
see
just
having
those
anecdotes
like
if
many
vendors
that
have
used
agents
sort
of
like
they
have
that
experience
like
most
customers,
don't
have
their
user
experience
broken
by
these
sort
of
changes,
then
definitely
an
aggressive
major
version.
Bump
policy
should
not
be
a
problem.
I
think
I
just
don't
know.
B
B
C
B
B
D
Because
I
I
mean
I
do
remember,
having
like
I've
had
experiences
just
once.
I
think
where
I
think
I
was
using
armeria
and
I
had
my
nice
grafana
and
alert
setup
and
it
changed
the
metrics
and
I
didn't
notice
and
we
pushed
the
binary
and
all
of
a
sudden.
All
of
our
graphs
for
our
production
servers
are
dead
and
we
don't
really
notice
in
alpha,
because
we
don't
look
at
the
graphs
for
our
alpha
servers.
D
We
only
look
at
graphs
for
production
really
and-
and
there's
like
how
do
I
fix
my
grafana
and
and
then
after
a
few
minutes,
I
fixed
the
graph.
So
it
wasn't
a
huge
huge
deal,
but
that
was
a
few
minutes
of
production
unreadiness,
which
was
quite
scary
to
deal
with.
B
B
D
B
C
Know
you
know
what's
important
here
is
the
route
and
so
and
there
I
think
there
is
a
attribute
for
a
route
that
we're
not
capturing
currently
and
so,
ideally
like
that
kind
of
alert
could
be
on
the
login,
because
you
need
something
because
the
you
you
need
to
say
this
is
my
login
page
is
typically
based
on
the
route.
D
Yeah
I
mean
especially
if
our
discussion
right
now
makes
us
think
that
the
main
reason
we
want
to
bump
our
major
version.
So
much
is
the
span
name
and
then
definitely
our
suggestion
here
should
be.
We
want
to
treat
the
span
name
outside
of
the
stability
guarantees
and
hope
that
users
only
use
attributes
that
will
allow
us
to
provide
a
good,
stable
experience,
while
good,
descriptive
names,
because
band
name
is
supposed
to
be
sort
of
descriptive,
anyways
right
so
having
that
outside
of
stability,
sort
of
makes
sense.
B
I
mean,
I
think
the
the
most
important
thing
is
that
we,
I
think,
like
ted
calls
out
like
we
have
to
just
define
something,
and
then
hopefully,
vendors,
who
are
you
know
very
involved
and
think
that
open
source
telemetry
is
important,
can
take
that
into
account
and
make
sure
things
work
with
what
guarantees
are
provided.
D
B
D
B
Never
I
mean
I
could
imagine
that
would
be
really
rad.
If
someone
had
a
you
know
an
sre
set
setup
that
could
do
game
days
against
canaries
that,
like
automate
the
whole
thing
automated.
That
would
be
amazing,
but
I've
certainly
never
seen
anyone
invested
sort
of
level
of
automation,
all
right,
yeah,
I'm
gonna
get
out
of.
D
C
Yeah,
I
think
there
was
nikita's
point
was
just
that
we
should
probably
go
ahead
and
you
know
push
out
all
the
people.
C
We
need
some
some
to
put
a
stake
in
the
ground.
There.
D
C
And
potentially
we
haven't,
you
know
we
do
need
to
maybe
on
tuesday
we
could
use
next
tuesday
to
do
a
triage,
yep.
C
C
C
Any
I
am
behind.
I
still
haven't
reviewed
stuff
today
me.
D
C
D
D
D
C
C
C
Yes,
yes,
yeah,
that
would
be
cool.
Really,
let's
put
that
in
the
next
release
party.