►
From YouTube: 2021-10-06 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
C
D
E
E
Perhaps
the
other
way
around.
So
first
the
community
and
then
maintainers
can
make
the
decision
yeah.
So
I
think
we've
talked
in
privately
tyrion,
but
I
would
also
like
to
get
bogue
and
also
to
share
his
his
opinions.
B
B
B
B
All
right,
so
I
guess
the
next
one
is
from
me.
So
what
what
we
would
like
to
do
is
to
work
on
the
agent
management.
B
So
I
think
it's
useful
to
form
a
temporary
work
group
to
discuss
this
topic
and
I
posted
an
issue
there
in
the
community
repository
if
anybody
is
interested
in
being
part
of
that
work
group,
please
have
a
look
and
we
probably
will
need
to
set
up
a
time
to
discuss
the
details.
B
I
don't
think
we
need
to
make
any
decisions
here
right
now
in
this
call
it
is
related
to
a
collector,
but
I
think
it
will
warrant
its
own
discussion
and
it
will
probably
take
a
while
before
we
it's
not
going
to
be
a
single
call,
time,
guess
again
a
single
issue:
it's
probably
going
to
be
something
that
is
continuous,
so
probably
many
of
the
current
collector
contributors
will
be
part
of
it.
B
F
B
F
Yeah,
so
I
want
to
have
some
discussion
about
with
spokane
material
or
other
members
about
the
health
checks
new
design,
since
we
are
moving
our
new
function
to
the
existing
healthcare
extension
and
sony
post.
A
comment
in
the
link
that
we
want
to
have
some
modification
on
the
current,
like
the
handler
of
the
current
design,
current
extension,
so
yeah.
I
just
want
to
know
which
what
is
the
right
way
to
do
that.
F
I'm
sorry:
what
was
your
question?
I.
B
G
B
Okay,
so
I
guess
well,
unless
you
guys
want
to
discuss
that
life
this
this
can
be
done
offline
is
the.
F
Yeah
yeah,
we
can
talk
about
talk
about
the
vlog
and
yeah
feel
free
to
leave
a
comment.
There.
G
Yeah,
so
I
I,
if
I
understood
correctly,
the
comment
anurag
wants
to
make
it
a
bit
more
generic
and
allow
to
have
multiple
health
conditions
defined,
and
this
one
be
one
of
them.
So
then,
in
the
future
we
can
add
more
health
conditions.
If
we
want
to
the
extension,
it
may
sound
reasonable.
I
need
to
to
double
check
and
understand
better
what
it
is
there
and
what
we
can
do.
Okay,.
C
G
C
F
Yeah,
so
the
next
one
is
where
we
have
a
pr
for
the
multi-config,
but
actually
that
one
is
need
to
use
current
functions
named
computer
provider
in
the
splunk
specific
pr-
and
I
have
talked
with
apollo
to
you
to
see
if
we,
the
splunk,
I
can
move
that
one
to
the
main
ripple
so
that
we
can
get
it
unblocked
and
yeah
so
yeah.
It
is
just
just
a
comment
here
and
we
want
to
notify
other
members
about
it.
So,
by
the
way
I.
G
G
F
F
Yeah
because
I
think
I
changed
the
miracle,
but
I
didn't
find
the
computer
provider
functions
so
yeah
yeah.
I
can
double
check
it
at
a
confirmation
offline
about
it.
G
D
In
order
to
use
that,
though,
we
will
require
a
code
change
in
the
initialization
where
the
parser
provider
is
set
up
right.
I
think
what
cheongduo
was
looking
for
was
the
config
provider's
ability
to
specify
an
include
or
a
file
that
that
had
a
single
config
file
that
pointed
at
multiple
sources,
which
I
don't
think
is
quite
where
we're
at
yet
with
the
current
implementation.
G
No,
we
are
not
there,
but,
but
I
I
need,
because,
based
on
the
document
anthony
that
you
guys
shared
with
us,
you
are
looking
for
a
solution
to
have
a
list
of
files
in
the
flag,
and
that
is
possible.
Indeed,
there
is
need
of
a
change
for
initialization,
but
that
was
the
requirement
from
your
document.
D
Okay
yeah,
so
if,
if
it's
just
a
list
of
files
with
the
flag
that,
I
think
we
just
need
to
have
the
flag,
add
multiple
parser
providers
or
okay,
so
yeah
there's
a
small
change
there
that,
but
we
can,
I
think,
do
that
without
the
the
config
sources.
Does
that
meet
your
needs?
Just
specifying
multiple
flags
at
the
cli
or
multiple
configs
at
the
cli.
F
B
Is
very
related
to
this
is
very
related
to
the
configuration
remote
configuration
management
as
well,
and
I
think
it
would
be
great
if
we
all
probably
discussed
this
in
that
within
that
work
group
that
I
was
just
talking
about
right
program.
This
is
also
related
to
what
we
were
discussing
yesterday.
B
At
least
it
is
related,
or
maybe
it's
going
to
be
the
same
mechanism.
I
don't
know,
but
it
will
be
great
if
you
guys
can
also
maybe
be
in
that
meeting.
So
this
is
going
to
be
definitely
one
of
the
topics
for
for
that
work
group.
So
I
think
it
will
be
very
useful
for
you
also
to
be
there
present
in
that
discussion,
where
we
want
the
configuration
to
come
from
a
remote
location,
but
also
it's.
It
also
requires
a
different
parser
provider.
It
is
possible
to
implement
it
using
a
different
charset
provider
today.
B
The
question,
then,
is:
is
everybody
if
everybody
has
slightly
different
needs
for
the
behavior,
where
the
configuration
comes
from
and
how
it's
merged?
But
you
only
are
able
to
have
a
single
parser
provider
and
that's
a
that's
defined
in
the
code
and
not
in
somehow
user.
It
is
not
user
definable.
Then
it
doesn't
work
right.
Everybody
wants
something
slightly
different,
so
we
need
to
make
this
end:
user,
configurable
and
user
definable
somehow,
and
maybe
using
the
idea
that
you
have
done
with
with
the
urls
and
the
stuff.
B
G
Yeah,
I
think
that's
coming
from
from
from
amazon.com,
so
if
I
I'm
following
their
design
in
in
the
map
provider
yeah
and
that's
why
I'm
surprised
that
they
keep
asking
for
that,
because
they
propose
a
design
which
I'm
trying
to
follow
as
much
as
I
can,
and
it
seemed
that
they
need
other
things.
So
I
think
I
think,
anthony
probably
you
should
go
back
in
amazon
and
double
check
with
that
design,
doc
if
that
is
still
valid
and
that's
exactly
what
they
need
or
they
they
have
different
requirements.
Now.
D
Yeah
yeah,
so
I
think
there
may
be
a
couple
of
of
different
desires
happening
here,
but
chandra
can.
Can
you
set
up
a
meeting
with
me
and
whoever
else
internally
is
working
on
that,
then
we
can
make
sure
that
we're
all
on
the
same
page
before
we
come
back
to
the
group.
That
degree
is
setting
up.
B
Anthony
did
you
have
a
chance
to
read
the
issue
that
I
posted
it's
it's
an
old
issue
about
remote
configuration.
If
you
had
a
chance,
I,
if
you.
D
B
Okay,
can
you
please
have
a
look
at
that,
so
that
I
think
we
have
a
meeting
next
monday
right
so
that
we
see
what's
common
in
how
I
see
that
and
how
you
see
that
and
what's
the
differences,
and
that
will
probably
tell
us
in
what
direction
we
want
to
go
right.
What's
missing
in
the
api
of
the
collector
there,
okay
I'll
post,
a
link
here
in
the
chat.
B
Thank
you.
Okay.
Let's
move
to
the
next
one:
pull
request
for
stats
the
receiver.
Okay,
is
the
author.
H
Yeah
I'm
here
there
might
not
be
much
to
discuss
here.
I
put
this
on
the
agenda
when
I
first
submitted
the
pr
just
checking
to
see
if
there
was
anything
else
not
being
considered
with
kind
of
you
know
we're
changing
the
calculation
of
the
receiver,
it
matches
the
proto
spec.
H
It
should
be
fine,
it
works
better
with
our
product
by
matching
the
spec,
but
I
just
wanted
to
double
check
that
there
was
nothing
else
that
we
missed
as
far
as
it
affecting
negatively
impacting
another
thing
or,
if
we're
just
good,
to
go.
D
Yeah,
so
I
think
we
discussed
the
concept
initially.
A
few
weeks
back,
I
created
a
design
doc
that
is
now
on
its.
I
think.
Second,
major
revision
after
we've
gone
through
some
discussion
with
with
tigran
and
bogdan
on
this,
and
I
just
want
to
get
this
out
in
in
front
of
the
community
so
that
before
we
commit
to
this
approach,
people
have
a
chance
to
review
the
design
dock,
the
the
implementation,
as
we
intend
to
put
it
forward
as
a
high
level
overview.
D
You
can
opt
into
beta
features
that
are
enabled
by
default,
that
you
can
opt
out
of,
and
then
the
feature
will
be
removed
when
it
becomes
generally
available,
and
the
feature
is
always
enabled
and
cannot
be
opted
out
of
from
a
code
user's
perspective
feature
gates
will
be
registered
in
with
a
global
registry
in
an
init
function,
and
then
the
global
registry
can
be
queried
at
runtime
to
get
the
the
status
of
the
gate,
as
the
user
has
configured
it.
D
Implementing
configuration
through
the
config
file
ends
up
being,
we've
decided
a
little
bit
too
late
to
accomplish
some
of
the
goals
we
have
particularly
around
being
able
to
control
configuration
itself
through
feature
gates,
which
I
think
may
come
into
play
with
some
of
the
work
that
the
the
configuration
work
group
that
tigran
is
trying
to
kick
off
would
have.
D
So
if
we
want
to
change
how
parser
providers
or
configuration
sources
work
using
a
feature
gate,
it
would
be
too
late
to
do
that
if
we
had
already
parsed
the
configuration,
so
the
initial
implementation
of
configuration
will
probably
come
from
a
cli
flag
or
an
environment
variable
that
lists
the
flags
that
should
be
enabled
or
disabled,
so
without
high
level
overview.
D
B
B
A
Awesome-
hey,
hey
everyone,
so
I
think
I
I
just
wanted
to
take
this
opportunity
to
kind
of
speak
more
about
the
fact
that
there's
been
some
great
progress.
So
we
did
see
the
collector
achieving
tracy
the
tracing
stability
milestone,
and
I
did
bring
this
up
during
the
maintainers
call
as
well.
But
I
I
am
doing
some
research
about
really
being
able
to
step
back
and
reflect
about
how
we
can
improve
the
getting
started.
A
Experience
for
the
age
and
the
collector
and
different
agents
and
and
the
collector
itself,
and
I
have
basically
kind
of
just
trying
to
get
more
visibility
into
what
the
concerns
are
being
voiced
by
the
community,
as
well
as
different
users,
who
are
actually
using
the
collector
itself
across
different
channels
of
communication
and
I'd
love
to
get
any
feedback
as
to.
If
there's
any
better
way,
that
I
can
collect
more
of
the
inputs
from
different
people
and
being
able
to
collect
more
feedback
from
different
sigs.
A
I
think
that's
what
I
was
focusing
on
doing
right
now,
but
yeah.
I
think
I
have
added
the
document
for
if
anybody
wants
to
really
be
able
to
contribute
and
add
in
it's
going
to
be
very
valuable
for
us
to
be
able
to
kind
of
step
back
and
understand
this
from
a
user
experience
perspective
as
well,
so
yep,
that's
all.
B
G
For
anthony
anthony,
do
you
know
if
kubernetes
has
the
same
global
mechanism
for
gates
or
they
do
different
things.
I
G
G
Okay,
if
you
can
point
me
to
the
code,
will
be
good,
maybe
point
it
in
the
pr
that
anthony
has
just
to
have
the
last
time
he
checked,
but
I
think
you
should
be
good.
D
D
Yeah
and
I
implemented
that
without
looking
at
the
kubernetes
implementation,
only
looking
at
their
documentation
and
description
of
the
feature
concept.
So
if
it
does
end
up
being
identical
to
what
they
did
it's
coincidental,
but
if
it's
different,
that
might
also
be
good
to
know.
J
Thanks
steven,
I
added
one
line
to
the
agenda
very
late.
If
there's,
if
there's
room,
we
can
talk
about
that
yep,
we
have
time
okay.
So
I
wanted
to
ask:
how
are
people
feeling
about
the
state
of
the
collector
and
collector
contrib
ci,
so
the
builds
take
a
while
and
right
now
it
is
possible
for
flakes
and
unrelated
components.
First
of
all,
it
takes
time
because
you
know
we
will
keep
growing
the
number
of
things
in
contribute
and.
K
K
Yeah
now
we
can
hear
you
okay,
all
right.
Sorry,
let
me
try
again.
G
I
I
think
contrib
is,
you
know
needed
collector
collector
is
like
super
fast.
I
don't
you're.
G
Yeah
it
there
is
a
real
appetite
in
the
fix
the
flakiness.
Definitely,
I
would
really
appreciate
help
there.
B
Okay,
so
I
typically
when
I
see
failing
on
stable
tests,
I
open
issues
when
I'm
most
most
of
the
time.
I
do
that
there
is
a
few
recorded
issues
for
the
unstable
tests,
so
if
actually
anybody
could
help
fix
those,
it
would
be
great
like,
and
it's
going
to
get
only
worse
right.
It
would
let
that
sleep.
J
That's
what
I
mean
I
mean
that,
like
it's
going
to
get
worse
in
two
ways,
if
we're
running
all
the
tests,
all
if
we're
running
all
the
tests
for
every
change,
then
the
whole
test
will
take
longer
and
the
flakes
when
they
occur
clicks
will
happen
more
often
and
then
they
will
block
more
people
for
more
time.
So
it's
kind
of
like
a
quadratic
loss,
so
the
the.
J
But
even
leaving
aside
like
a
maybe
a
yes,
I
would
be
open
to
I'd
be
open
to
using
bazel.
That
could
be
interesting,
but
even
leaving
that
aside,
just
saying
hey
this
component
changed,
it
wasn't
a
shared
library,
let's
prune
the
set
of
tests.
B
J
B
G
There
are
a
bunch
of
things
that
we
need
to
do
so
punya,
jokes
aside,
I
think
bazel
would
be
great
and
would
be
great,
not
necessary
for
people
to
use
on
a
daily
basis
if
they
don't
want.
We
still
allow
them
to
use
the
make
command
to
to
use
the
normal
thing,
but
for
the
ci
to
have
bazel
enable
on
this,
it's
going
to
be
tremendous
help
in
terms
of
speed
up
that
test.
G
Then
then
there
is
another
problem
with
the
testbed,
where
we
have
a
mix
of
low
tests
correctness
test,
even
though
we
split
some
correctness
separately,
but
still
in
the
low
test.
There
are
some
correctness
tests
and
random
tests
there
that
we
need
to
probably
split
into
different
things.
G
It's
a
lot
of
work
in
front
of
us
if
we
want
to
do
it
nicely,
but,
as
I
said,
I
am
super
into
using
bazel
for
for
the
ci
test.
If,
if
that
will
will
help
and
speed
up,
because
I
feel
like
implementing
our
own
logic
of
not
building
some
of
the
tests,
if
code
did
not
change,
your
dependency
did
not
change.
G
It's
actually
re-implementing
part
of
the
main
reason
why
people
exist,
so
I
would
not
reinvent
the
wheel
if
we
already
have
a
solution
for
for
that
problem,
unless
you
know
a
three-line
solution
for
for
that,
I
would
probably
choose
to
to
use
bazel
for
for
this
comments.
E
Yeah,
so
I
was
going
to
say
that
contrib
is
in
a
way
split
into
modules
right,
so
it's
not
like
you
can
change
one
component
and
then
in
the
same
pr
change
another
component
and
expect.
I
don't
know
things
through.
C
E
So
perhaps
the
test
can
be
triggered
in
database
on
path
right,
so
if
it
touches
these
this
module
here,
this
go
module.
Then
this
test
here
is
triggered
or
the
test
for
the
module
are
triggered.
So
I
think
it's
not
that
complicated
to
implement
something
that
you
know
only
tests,
whatever
is
touched.
J
So
I
think
I
will
write
a
doc
that
says
like
here's,
the
problem,
here's
the
solution
that
tries
to
use
go
tooling
solution
that
tries
to
adopt
bazel.
What
are
the
benefits
of
each?
What
is
the
cost,
and
then
we
can.
We
can
circulate
that.
I
think
the
knowledge
is
in
this
room.
I
just
want
to
have
it
on
paper.
E
The
other
thing
that
I
would
just
leave
in
the
air
as
well
is
we
don't
actually
have
to
run
all
the
load,
tests
and
correctness
tests
for
all
the
pr's
right.
So
I
think
it
is
quite
acceptable
to
run
them
on
a
daily
basis
like
on
a
night
rebuild.
B
B
E
E
Bit
more
of
work
than
what
we
have
today,
but
the
the
the
plus
on
the
plus
side.
We
we
don't
run
those
just
for
every
pr
which
you
know.
B
B
E
Yeah,
so
I
guess
the
one
problem
that
I
have
with
running
the
low
chess
all
the
time
is
that
sometimes
I
have
a
a
quick
fix
and
I
I'm
guilty
of
not
running
the
test
locally
and
and
then
I
see
that
they
failed
on
the
unit
test
part,
and
I
only
see
that
two
hours
after
and
the
load
tests
were
executed
already
for
that.
So
they
didn't
actually
have
to
be.
B
You
know
yeah,
I
think,
by
the
way,
for
the
law
tests.
Sorry
for
the
load,
tester
is
actually
a
different
thing.
That
we
wanted
to
do
is
to
run
them
on
a
stable
hardware.
They
are
currently
running
on
on
unstable
hardware,
and
they
that's
why,
as
expected,
sometimes
they
do
fail.
We
put
some
significant
margin
there
to
avoid
often
failures
happening,
but
it's
still
sometimes
it
can
be
significantly
slower
the
hardware
right
and
then
they
do
fail.
B
J
Just
to
control
scope
here,
here's
how
here's
the
goal
that
I
am
I
would
like
to
set
for
myself.
If
the
community
is
on
board,
let's
not
change
whether
or
not
we
change
the
set
of
tests.
I
would
like
that
for
someone
who
makes
a
component
change,
they
should
get
fast
feedback
on
unit
test
failures
in
their
component.
J
B
J
Much
faster
than
local
dev,
I
think
local
dev
is
solved.
I
think
within
ci,
for
example,
there
is
a
there
is
a
matrix
of
running
my
tests
on
windows
and
linux.
That's
not
something
people
do
in
their
local
dev,
like
local
iteration
loop.
I
would
like,
like
them
to
get
basically
component
level
tests
that
are
not
since,
like
across
all
components.
I
should
get
them
back
fast,
even
if
I
did
not
run
them
locally
for
whatever.
Maybe
I'm
running
them
on
one
package,
and
this
is
getting
on
the
whole
component,
yeah
yeah.