►
From YouTube: 2021-01-08 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).
B
A
Yeah,
the
new
year
itself,
yeah
I
mean
I
went
to
bed
at
nine
o'clock.
It
was,
it
was
all
good.
D
C
Yes,
and
now
I
need
the
weekend
to
catch
up
again
like
I
was
almost
caught
up
from
the
holidays
and
then
yeah
and
then
I
lost
the
day
and
now
play
catch
up
again.
C
Right
right,
yes,
not
too
not
to
complain.
Ooh
devos,
just
resigned.
A
Oh
yeah,
I
mean
the
whole
cabinet
I
think,
is
trying
to
get
out
as
quickly
as
possible.
I
I
saw
that
there.
One
of
the
reasons
is
that
trump
is
wandering
around
the
white
house
like
trying
to
cook
up
bad
illegal
plans,
and
anyone
who
listens
in
on
any
of
those
is
potentially
liable.
As
in
under
conspiracy
laws.
B
E
D
B
C
B
E
A
I,
like
your
your
duct
tape,
vent
there.
It
reminds
me
of
my
house
a
lot
traffic.
D
C
So
this
is,
I
don't
know
if
it's
better
or
worse,
but
the
vent
is
actually
there
it's
just
in
the,
but
in
the
summer
I
tape
it.
I
tape
it.
I
tape
over
it,
so
the
ac
doesn't
come
in
because
it's.
C
D
E
Which
of
the
six
quadrants?
Are
you
in.
C
Got
it
north
wind
here
I
look
like
right.
That's
why
you
should
see
that,
because
I
forgot
to
put
that
over
there.
A
A
A
C
C
A
E
I
I
don't
know
how
nice
they
are
now,
but
they
were
nice.
A
couple
years
ago
I
mean
they're,
nice
they're,
very
slow,
yeah
yeah.
There's
a
lot
of
pedestrians.
A
C
And
that's
yeah:
there's
a
windy!
There's
the
old,
because
it's
it's
a
real
old
bike
path.
Most
of
it
is,
but
they
just
built
the
new
stuff
that
like
goes
connects
over
the
ross
island
bridge,
which
is
pretty
cool
because
now
now
from
my
parents
still
live
there.
So
you
can
bike
down
to
the
park
and
do
a
loop
around
ross
island
and
the
tilikum.
E
D
C
Of
out
of
the
us,
that's
still
a
pretty
good
place
to
be
our
our
yeah.
I
was
looking
at
our
our
covid
numbers.
Are
I
mean
the
coveted
us?
Numbers
are
horrific,
and
but
oregon
is
not
great,
but
like
one
of
the
best.
A
Spots,
we're
keeping
it
under
control.
I
have
multiple
neighbors,
who
are
doctors
in
hospitals
and
they're,
not
freaking
out
too
much.
D
C
C
C
We
did
have
a
meeting
this
morning.
We
did
manage
to
fill
it
up
with
people
had
stuff
to
chat
about.
C
Harry
yeah,
a
customer
like
a
end
user.
D
C
C
First,
first
meeting
of
the
year
so
for
honorics
benefit.
C
Oh
yeah
jason
wanted
us
to
check
in
on
this
issue.
I
said
it
makes
sense
to
me
in
fact,
actually
in
our
exporter,
I
filter
out
these
spans
that
don't
that
are
polling
that
don't
pull
any
messages
because
they're
pretty
noisy.
C
Yeah
so
anyway,
they
were
at
their
two
points.
One
is
they're
asking
for
us
to.
Please
make
release
notes
with.
B
C
C
Showed
them
where
it
is
in
maven,
central,
okay,
cool
and
then
the
other
question
pavel
had
was
about
how
other
people
can
request
a
patch
release
so
non
basically
non-maintainers.
How
would
they
drive
help
drive
that
so
john
wisely
suggested?
We
take
that
to
the
maintainers
meeting
on
monday
and
sort
of
ask
that
as
a
general
question,
if
there's
any
consensus
or
guidance
or
you
know
open
telemetry
wide
how
to
support
that.
A
A
B
Even
this,
like
the
main
reason
for
doing
this,
weird
patch
release
was
because
we're
still
not
1.0.
I
don't
think
it
was
good
practice
because
you're
not
supposed
to
release
new
features
and
patch
releases,
and
so,
if
they're,
just
bug
fixes
generally,
you
want
to
fix
the
bug.
So
you
release
the
patches
and
easily
fix
the
bug.
B
E
C
As
maintainers
to
be
like,
oh
yep,
it's
fixed
release,
a
patch
if
it's
a
bug
that
has
been
there
since
the
beginning.
It's
a
little
bit
more
dubious
of
whether.
D
A
C
So
yeah
we'll
we'll
we'll
circle
back
to
pavel.
I
think
in
particular
as
a
vendor
next
week
and
imagine
it'll
take
longer
than
just
a
week
to
kind
of
sort
that
out,
but
I
also
encouraged
him
to
take
a
look
at
the
release.
Our
our
release,
training,
sort
of
release,
process.
C
C
The
other
thing
I
thought
of
afterwards
was:
I
think,
that
approvers
in
our
repo
approvers
have
merge
rights,
which
means,
I
think
approvers
could
trigger
a
release,
which
is
good,
I
think,
and
that
I
was
gonna
mention
to
pavel.
Potentially
I
mean
potentially
he
would
be
a
approver
candidate
if
he
started,
you
know,
work
doing
reviews
and
that
kind
of
thing,
because
I
think
partly
I
mean
that's,
why
companies
are
investing
time.
C
C
A
C
C
C
Yeah,
so
we
should.
We
should
remember
anurag
when
we
see
that
pr
to
make
sure
that
he
adds
that
to
the
example.
The
demo
example
also
at
the
same
time,
I.
A
Have
some
comments
on
the
example
messing
around
with
it
today
I
put
put
some
of
it
in
the
channel,
but
it's
probably
worth
talking
about
it.
Here
is
a
little
bit,
and
that
is
mostly
with
us
working
hard
to
make
the
sdk
as
immutable
configuration
wise
as
possible.
A
The
tracer
configure
or
whatever
that
interface
is
called.
It's
probably
gonna
have
to
change,
because
you're
not
gonna
be
able
to
do
anything
with
the
management
interface
anymore
like
you
could
shut
things
down,
but
you're
not
gonna,
be
able
to
add
spam
processors.
You're
not
gonna,
be
able
to
do
that
stuff,
so
I
think
it
probably
may
my
initial
thought
is:
it
makes
sense
to
kind
of
turn
that
interface
around
and
have
it
return
an
sdk
rather
than
take
an
sdk
that
you
mutate
so.
A
Oh
yeah,
that
would
work
too
yeah.
That's
a
fun
way
to
do
it,
although
it
would
be
nice
like
this
is
kind
of
what
the
thing
that
I
ran
into
was,
of
course,
on
that
exporter
stuff.
If
you
all
you
can
do
is
mutate.
The
builder,
you
can't
remove
span
processors
yeah,
so
you,
like
you,
you
there's
no
way
to
like
start
with
a.
B
A
A
B
B
B
B
B
A
Yeah
something
like
that
anyway,
I
mean
this
is
I
think
this
is
none
of
this
is
incredibly
urgent.
It's
just
you
know.
Having
going
through
trying
to
we
have
a
customer
who
wants
a
custom,
propagate
use,
a
custom
propagator,
so
I've
just
been
messing
around
with
trying
to
figure
out
like
the
right
way
to
do
that
easiest
way
to
do
it
so.
C
A
A
A
Yeah,
the
I
think
one
of
the
engineers
actually
put
a
dis
created
a
discussion
about
this
topic.
Well,
they
did
in
java
and
they
say
they
were
going
to
move
it
over
to
instrumentation.
I
don't
know
if
they
ended
up
doing
that.
A
A
A
So
I
posted
in
the
chat,
the
the
issue
we
got.
C
A
splunk
customer
because
he's
he's
posted
in
the
yeah
a
couple
of
issues
before
oh.
A
A
A
C
A
There's
a
there's
a
there
is
a.
I
can
dig
it
up,
there's
a
an
issue
already
that
nikita
logged
about
this
about
this.
It
is
issue
number
whoa,
okay,
issue,
number
1681.
B
B
E
A
B
A
But
I
didn't
have
a
ton
of
time
to
dig
in
so
I'm
not
sure.
B
A
C
One
thing
just
to
were
like
for
customers
using
this,
since
what
did
the
highlighting?
No
sorry
is
that
the
you
know
we're
changing
the
ap,
our
java
agent
apis
and
our
java
agent.
Api
won't
be
stable,
for
you
know
some
number
of
months
so.
A
Yeah
we
are,
we
are
definitely
messaging
very
carefully
to
the
customers
that
this
is
they
are,
they
know
they're,
they
know
very
well.
They
are
on
the
leading
edge
of
what's
going
on
here,
so
cool,
absolutely.
C
A
A
A
Yeah,
you
can
see,
I
guess
you
can
see
who
our
customer
is
by
looking
at
his
looking
at
his
profile.
C
Oh
yeah
and
john
was
bringing
up
this
idea
of
just
bare
bones,
supporting
only
with
span
and
potentially
context
propagation
executor,
class,
loader
instrumentation.
So
I
you
may
have
seen
his
issues.
C
A
C
A
No,
in
this
case,
I
what
I
was
what
I
really
wanted
is
I
wanted
to
use
the
bear
sdk
but
have
with
span
work,
but
I
didn't
want
to
go
through
jump
through
the
hoops
of
building,
dynamic,
proxies
and
all
sorts
of
yeah.
If.
B
A
I
mean
I
just
want
to.
I
want
to
be
as
lazy
as
possible,
yeah
nice
to
be
able
to
just
slap
an
annotation
on
there
and
not
have
to
worry
about
doing,
because
I
don't
want
to
like
the
thing
I
don't
want
to
have
to
worry
about.
Wrapping
around
every
one
of
my
methods
is:
is
the
con
putting
it
in
the
context
and
starting.
C
B
A
B
D
A
I
mean,
I
think,
it's
more
about
control
this,
and
this
is
purely
a
me
thing.
This
is
not
for
a
customer.
This
is
like
what
I
want
for
my
toy
projects.
Okay-
and
I
just
want
very-
I
want
fine
grain
control
over
what
spans
are
there
and
what
aren't.
E
I
think
there
are
customer
use
cases
for
this,
though,
and
it's
people
that
are
running
with
so
low
overhead,
like
sorry
so
low
headroom,
that
they
can't
afford
full-blown
instrumentation.
But
they
care
about
these
few
little
areas
and
they're
willing
to
put
in
like
they're
willing
to
target
these
very
specific
areas.
But
they
can't
have
it
in
the
rest
of
their
app
because
their
headroom
is
too
low.
C
Hope
I
hope
this
makes
a
good
dent
in
our
head
room.
I'm
worried
that
all
the
infra,
like
that
all
the
infra
is
still
running.
We're
still
checking
all
these
classes
that
that
may
not.
B
A
Like
a
new
relic,
we
had,
there
was
a
quasi-secret
feature
called
light
mode
that
would
turn
off
instrumentation
but
leave
annotations
in
place
and
internally
it
was
used
by
our
ingest
pipeline,
where
they
they
we,
the
only
reason
they
used.
The
agent
was
that
they
were
required
by
our
architecture
team
to
use
the
agent,
otherwise
they
would
would
have
not
used
it
at
all.
Yes,
just
like
for
dog
footing,.
D
A
Everywhere,
yes,
we're
dog
fooding
we
just
knew
relic,
has
a
blanket
policy
that
every
service
must
be
instrumented
with
new
relic,
which
I
think
is
for
a
company
is
a
good
policy
to
have
right,
but
they
our
ingest
pipeline.
They
really
wanted
to
make
that
as
lean
as
possible,
and
they
did
not
want
the
agent,
but
they
were
required.
So
they
got
the
agent
team
to
introduce
this
light
mode.
That's
basically
what
I'm
proposing
here
interesting.
I.
E
Was
also
going
to
link
to
the
ultralight
tracer,
but
we
lost
all
the
history,
so
it
doesn't
have
your
name
on
it.
Oh.
C
It
was
a
fun
scene,
I'm
like
I'm
scratching
right
here.
I'm
pretty
sure.
I've
heard
this
exact
same
request
from
bogdan
like
over
a
year
ago,.
C
Sometime
a
long
time
back
that
yeah
this,
but
specifically
of
propagating
or
not
that
long
ago,
because
I
feel
like
it
was
with
nikita
also
maybe
when
nikita
was
digging
into
a
lot
of
the
context.
Propagation
local
context,
propagation
and
bogdan
was
interested
in
having
that.
Just
the
local
propagation.
Only
part
of
the
agent
which.
A
A
You
would
you
would
get
it
as
long
as
you're
thread
local
without
having
to
enable
anything
else
right
just
having
within
yeah.
It
would
get
it
into
the
context
and
it
would
still
be
on
the
thread
local
and
then,
if
you
needed
to
enable
some
executors
it'd,
be
pretty
easy.
To
just
add,
add
that
add
the
bait
you
wanted,
the
other
you
know.
Another
reason
for
this
is
that
at
least
I
know
it's
true.
It
was
true
it
near
relic,
and
it's
true,
I
think
it's
splunk
as
well.
A
It's
much
easier
to
get
people
who
are
scared
of
agents
on
board
with
something
that
is
very
lightweight
and
doesn't
do
anything
first.
They
think
it's
lightweight
right
and
it
feels
lightweight
to
them
right
because
it's
like.
Oh,
this,
isn't
doing
anything.
I
have
complete
control
and
then
you
can
kind
of
get
them
hooked
on
the
agent
and
then
they
can
add
instrumentation
in
as
they
want
so
like
I.
A
E
It's
also
like
having
a
flag
to
disable.
Basically,
all
of
instrumentation
is
kind
of
really
nice
too,
for
troubleshooting
when
stuff
is
just
falling
over
and
you
can
ask
a
customer
to
just
just
try
this
and
it
doesn't
still
work
and
if
they
say
no,
then
you
can
rule
out
that
it's
an
instrumentation
problem,
yep.
B
C
Cool
it's
partly
for
that.
You
could
also,
I
mean
I
don't
know
how
like
disabling
the
job,
the
agent
entirely.
This
is.
C
A
Anyway,
if
that
I'm
going
to
test
that
out
and
if
it
works,
then
cool,
we
just
add
some
documentation
and
then
we're
good
to
go.
A
B
Agent,
allergies:
I
know
we
get
a
lot
of
getter
nowadays,
like
how
do
I
see
my
mongodb
traces
and
that
sort
of
thing
that
has
been
giving
me
a
bit
more
motivation
to
want
to
split
out
more
library
instrumentation,
because
that's
the
easiest
way
to
answer
those
questions.
It's
like!
Oh,
if
you
want
to
use
mongodb
here,
check
out
our
library
instrumentation,
because
again
I
mean
or
use
our
agent
that's
another
way
we
can
answer,
but
that's.
B
C
At
least
start,
I
would
at
least
suggest
that,
starting
with
that,
I
know
that
the
the
other
vendors
that's
sort
of
the
our
starting
pitch
is
hey.
Here's,
the
java
agent,
you
know
just
because
it's
the
lowest
overhead
from
a
support
perspective.
That's
true.
C
C
B
Yeah,
so
I
guess
yeah,
I
didn't
think,
like
these
questions
tend
to
come
with
open
terms
you
java
channel
and
when
they're
talking
about
using
the
sdk,
and
how
do
I
get
my
or
something
the
natural
answer
was
always
to
try
out
in
library
instrumentation,
but
pointing
them
at
the
agent
actually
sounds
like
a
good
answer.
Also
just
didn't
think
of
that.
E
A
but
good
to
know
if
there's
some
logging
compatibility.
C
Oh
yeah,
that's
2017.,.
D
C
Keep
can
people
keep
hammering
me
on
glow
root
about
java
15
like
because
I
haven't
updated
the
the
asm
asm.
C
Long
time
so,
actually
it
broke
in
java
14,
and
so
I
started
getting
java
14
bugs
and
now
java
15
bugs
sort
of
like.
B
B
Like
I've
always
wondered,
why
asm,
I
guess
it's
just
in
case
the
new
byte
code
has
some
feature
it
doesn't
support,
like
almost
all
asm
bumps
are
just
adding
something
to
that
enum.
It
seems
so
boring
that
that
tends
to
be
the
blocker
for
supporting
a
new
java.
Like
I
know,
gradle
always
has
that,
and
I
always
wished
that
there
was
a
way
to
just
force
it
to
work.
C
Oh
just
the
because
the
the
problems
I've
that
have
been
reported
for
me
are
more
real
like
when
they
like
it'll
work,
unless
you're
actually
using
like
a
java,
11
feature.
Okay,.
C
Otherwise,
it
has
worked,
maybe
the
I
mean
gradle
or
some
other
projects
may
be
more
conservative
and
maybe
say
that
we
won't
even
try
to
instrument
you
see,
which
would
seem
reasonable.
Also,
yeah
approach
does
glory,
use
it
or
asm
asm
yeah
yeah.
It
has
its
own
its
own
bite,
buddy-ish,
infra,
annotation
driven.
You
know,
method.
C
I
imagine
yeah,
I
think,
there's
lots
of
lots
of
people
did
the
similar.
It
was
nice
that
bite
buddy.
Finally,
like
kind
of
brought
everybody
together.
C
I
know
I
had
a
lot
better
success
at
performance
tuning,
the
glow
root,
startup
overhead
because
it
was
lower
level,
and
so
once
I
switched
us
over
to
the
open,
telemetry
stuff,
the
the
startup
overhead
did
go
up.
E
C
It's
not
bad.
I
have
gotten
the
the
pushback
I've
gotten
internally
is
from
the
functions.
D
E
D
B
C
I
I
have
them
the
the
the
path
forward
is
put
the
java
agent
on
the
the
whole
pool
of
pre-warmed
instances.
A
B
B
A
So
that
does
work
to
disable
all
instrumentation.
I
will.
I
can't
confirm
that
turns
everything
off.
B
So
for
the
sdk,
I
think
we're
in
good
shape
not
that
the
auto
configure
got
merged.
There's
my
one
pr
vlogging,
the
warning.
If
the
sdk
is
on
not
initialized.
A
B
A
B
C
A
B
A
A
Why
is
it
only
11
because
it
uses
jfr
in
some
way
that
only
actually
it
might
work
with
very
recent
versions
of
java
8.
C
B
A
B
A
In
there
and
when
they,
when
they
open
sourced,
actually
I
think
this
was
probably
a
lot
of
it
was
driven
by
marcus,
hurt
who's.
A
data
dog
was
datadog,
really
really
wanted
jfr
in
their
in
their
profiling
solution.
A
A
B
A
E
A
C
Is
it
just
so
that
you
can
correlate,
then
your
profiles
with
I
mean
because
if,
if
you're
emitting
events
for
spans
I
mean
is,
that
is
that
I
would
assume
that's
what's
needed
on
the
back
end
to
then
correlate
your
profiles
to
your
traces.
A
C
A
E
A
Have
a
user
land
consumer
of
jfr
events?
You
can
get
all
of
your
data
all
of
your
deep
jfr
profiling
data
into
user
space
and
then
export
it
as
telemetry
to
whatever
telemetry
system
you
want.
I've
been
thinking
for
a
while
that
you
could
implement.
You
could
ditch
the
agent
altogether
and
base
it
all
on
jfr
events,
and
you
could
actually
build
a
really
good
tracing
profiling
combined
solution
based
on
using
jfr
as
your
low-level
instrumentation
system,
but
java,
14
plus,
of
course,
super
limits.
A
D
A
Yeah,
but
I
think,
there's
some
interesting,
there's
some
interesting
experimental
work
that
could
be
done
with
using
jfr
as
your
primary
source
of
telemetry
once
you've
got
streaming
apis
where
you
can
bring
that
data
back
into
user
space.
B
C
Yeah
I'll
keep
you
posted,
we
we're
doing.
Martin
veraberg's,
the
j
clarity,
folks
that
we
acquired
microsoft
acquired
are
looking
at
doing
some
integration
where,
because
we
don't
have,
our
our.net
version
does
have
a
profiling
integration
and
so
they're,
looking
at
doing
something
similar
with
java
and
so
they're.
The
approach
they're
taking
is
to
just
write
to
disk
because
of
the
streaming,
and
I
think
they
had
some
kirk
had
a
bunch
of
a
bunch
of
other
objections
to
the
streaming.
C
C
Stuff,
looking
at
this,
like
that,
I
trust
to
give
good
good
advice
and
to
put
together
something
good.
So
I'm
hoping
that
you
know
this
is
step
one
and
hopefully,
at
some
point,
I'd
love
to
know
when
splunk
starts
thinking
about
profiling,
and
you
know
if
we
can
combine
up
you
know,
our
goal
is
to
upstream
everything
also.
A
Well,
new
relic
has
some
jfr
based
profiling,
something
something
that
jason
knows
lots
about.
I
don't
know
how
much
we
that's
all
been.
Has
that
been
released
as
alpha
or
anything?
I
don't
even
know
what
this
is.
E
A
B
A
C
E
And
it
really
does
want,
I
mean
it
feels
like
and
like
many
things
in
java,
it
feels
like
they
were
kind
of
playing
catch
up
with
some
of
the
stuff
in
the
clr
right,
like
the
clr
has
tons
of
built-in
instrumentation,
like
maybe
not
instrumentation,
but
hook,
points
for
like
and
I
don't
know
anything
about.net.
But
my
understanding
is
that
it
has
a
lot
of
hook.
Points
for
building
this
kind
of
instrumentation
like
it's
built
into
the
vm.
C
C
B
E
B
C
A
A
B
A
B
A
A
No,
the
reason
why
are
are
hand-written
and
they're
not
up-to-date,
like
they
actually
have
things
that
are
wrong
in
them.
So
that's
anyway,
there's
an
issue
that
we
need
to
that's.
C
C
Now,
what
about
oh
h2?
That's!
We
didn't
have
that
in
here
anyway,
right,
but
we
did
have
h2,
but
we
weren't.