►
From YouTube: GMT 2018-07-12 Apache Mesos Dev Sync
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
C
D
Ben
and
Ben
wanted
to
chat
briefly
about
on
some
Metro
stuff
has
been
on
yeah,
oh
sweet,
hey
Ben
should
I
kick
it
off
Ben.
Do
you
want
them
I.
D
Let
me
start
without
and
then
I
can
go
grab
my
laptop
if
that
makes
sense.
Okay,
so
so
I'll
start.
So
the
context
was
we
want
to
add
more
metrics
around
live
process
and
I.
Don't
know
great
man's
on,
but
that'd
be
good,
because
because
I
know
he's
also
gotten
a
bunch
of
framework,
specific
metrics
I
know,
there's
other
metrics
changes
that
are
being
considered
and
thought
about
right
now,
as
well,
especially
around
exporting.
D
So
the
note
you
might
have
a
bit
more
context
on
on
that
one,
and
so
when
Ben
and
I
were
chatting
about
adding
met,
there
were
specifically
for
the
processes
for
the
actors
running
and
live
process.
There
were
a
couple
things
that
came
up
that
we
wanted
to
run
by
the
dead
sinking.
Get
gets
keep
so
the
two
big
ones
that
I
wanted
to
throw
out
there.
D
F
D
It's
actually
very
easy
to
add
the
rig
stuff,
but
when
Ben
and
I
were
chatting
about
it,
there
might
be
some
concerns
with
reg
X
being
properly
supported
across
all
compilers
and
standard
C++
libraries
there's
anybody
I
actually
know
about
that.
Does
anybody
know
if,
because
Bennett
suggested
that
actually
we
tried
to
use
some
reg
stuff
in
the
past
and
we
ran
into
some
issues?
Does
anybody
on
the
call
remember
that
and
and
can
speak
to
that
I.
D
A
D
D
D
D
Yeah
so
so
I
mean
the
idea
would
be
in
the
MVP
that
we
would
go.
We
would
as
part
of
the
documentation
we
share
with
people
hey
like.
If
you
want
to
collect
the
metrics
for
the
master,
these
would
be
the
afternoons
you're
gonna
write
down,
you
want
to
collect,
metrics
or
the
you
know
whatever
it
is,
these
would
be
the
actor
names
you'd
write
down,
and
so
we
would
just
give
examples
to
start
or.
J
D
D
J
How
to
get
that
list
it'll
be
nice
to
have
a
description
for
each
actor
right
now,
I'd,
just
like
a
name
who
knows
what
that
means,
there's
some
very
weird
name
for
actors
that
we
have
any
nice
point
resolved
that
you
have
a
name
of
the
actor
and
I
also
have
a
description
like
the
human
readable
description.
What
that
is
well.
E
Okay,
it
it's
not
free
to
do
that
right,
like
how
many
bytes
per
act
are
you
gonna
need?
If
you
want
to
have
a
lot
of
actors,
it's
it
could
add
up.
I
guess
the
question
is
like.
Why,
like
in
my
mind,
if
you're
gonna
monitor
an
actor,
you
have
to
have
an
understanding
of
like
why
you're
doing
that,
so
you.
J
E
J
C
D
J
J
D
D
We
can
put
something
out
there
where
organizations
like
the
folks
that
are
on
this
call
right
now
just
know
which
actors
they
probably
want
to
collect,
metrics
for,
and
they
can
start
doing
that,
and
then
we
can
follow
up
and
add
something
which
gives
people
that
have
no
idea
which
actors
they
want
might
want
to
follow.
You
know,
okay,.
H
E
D
Very
first
first
MVP
that
Ben,
M
and
I've
been
talking
about
her
just
for
the
just
for
the
actors
and
just
for
the
HTTP
endpoints.
That's
it
so
so
that
was
going
to
be
the
MVP
James,
just
those
two,
but
again
we
can
follow.
We
can
follow
pretty
quickly
with
other
other
the
process.
Metrics
I
just
think
this
is
actually
a
good
venue
to
talk
about
that
and
get
feedback
from
people
about
which
metrics
we
think
we
want
to
add.
I
will
say
one
thing,
which
is
the
second
question
I
wanted
to
ask.
D
So
the
first
question
was:
how
do
we
feel
about
the
the
instrument?
Basically
deciding
which
actors
we
want
to
instrument
which
is
again
with
like
akka
and
a
bunch
of
these
other
things
do
and
sounds
like
we've
got.
We
that's
what
Ben
and
I
kind
of
kind
of
landed
on
and
sounds
like
there's
there's
good
consensus.
People
feel
good
about
that.
D
We
just
need
to
make
sure
that
we
either
document
it
well
or
give
people
a
way
to
find
out
which
actors
they
might
want
to
get
get
metrics
on,
but
the
second
question
and
then
Ben
I,
think
you
had
something
to
say
in
response
to
James.
But
the
second
question
that
I
wanted
to
bring
to
this
forum
was,
what's
the
best
way
to
name
the
metrics
and
and
so
anyway.
So
I
do
want
to
get
to
that
one,
because
I
want
to
get
feedback
from
folks.
D
But
then
why
don't
you
say
your
thing
and
then
that
portion
is
nicely
naming
I
was.
D
H
It
makes
sense
well,
I,
guess
the
the
model
I
had
but
I
think
I
would
probably
do.
Is
that
I
would
probably
look
at
aggregate
metrics
for
something
right.
I'm,
looking
at
some
kind
of
aggregate
metrics
of
message,
processing,
message
to
report
or
something
and
and
I
would
go.
Oh
that's
interesting
and
then
I'd,
probably,
and
then
that
would
cause
me
to
want
to
drill
down
more
into
to
find
out
what
specific
actors
are
doing.
So
that's
the
sort
of
analytic
framework
that
I
was
I
had
it
bud,
got.
K
H
H
H
D
D
There's
the
actor
level
metrics
and
it
seems
like
that's
what
most
people
are
going
after
is
they
actually
want
to
be
able
to
get
metrics
for
what's
happening
within
within
the
individual
actors
and
and
so
IIIi
think
we
can
look
at
adding
a
couple
of
them
like
I,
just
pulled
up
the
the
akka
stuff
here
and
there's
I.
Think
four.
There's
number
of
active
actors,
number
of
process
messages,
number
of
unhandled
messages
and
and
number
of
dead
letters
which
doesn't
exactly
attract
to
us.
D
So
I
think
we
could,
if
we
wanted
to
add
a
couple
of
just
system-wide,
metrics
but
I.
Think
a
lot
of
people
are
going
to
be
caring
about
what's
happening
within
the
individual
actors
pretty
soon,
which
is,
which
is
why
we
focused
on
the
metrics
for
the
actors
and
and
and
then
I
think
what
James
and
Ben
Raj
is
talking
about
is.
Do
you
want
to
have
a
gauge
for
something
like
the
number
of
of
messages
that
are
currently
in
in
the
the
actors
mailbox?
D
E
H
D
D
You
know
other
prefix
slash
the
metric
name,
and
so
what
we
could
do
that
we
could.
You
know
for
the
specific
actors
we
could
put
the
actor
name,
slash
the
metric
name,
but
Ben
and
I,
don't
like
this
conflicting
with
people's
own
metrics.
They
might
be
creating
once
the
ideas
is
that
we
put
all
these
metrics.
That
live
process
is
creating
namespace
under
metrics
slash
and
then
any
other
potential
prefix
or
code,
followed
by
the
metric
name.
D
I
G
D
I
H
D
I
E
That
might
be
fine
for
the
actor
related
once,
but
for
routes
I.
Personally,
don't
really
like
the
idea
of
coupling
routes
to
the
actors
that
currently
are
responsible
for
those
routes
like
if
we
refactor
things
in
the
future.
This
happened
once
in
the
agent
I
think
we
had
a
thing
called,
monitor
and
there's
a
route
for
that
and
we
refactored
it
so
that
actor
doesn't
even
exist
anymore,
but
we
have
to
preserve
that
route
name
because
it's
an
API
now
there's
no
actor
for
that
route.
E
So
I
I
don't
know
if
processes
need
to
be
namespace
in
the
same
way,
but
I
guess
you
could
use
the
same
idea
with
like
master
metrics,
like
maybe
master,
becomes
less
a
less
of
a
clear
actor
in
the
future.
Maybe
it's
like
a
whole
bunch
actors
sharing
the
work
and
there's
not
one
master
actor
anymore,
but
those
metrics
to
have
to
stay
the
same
names
right.
D
D
So
one
of
the
reasons
why
I
wanted
to
bring
this
up
is
because
I
know
for
whoever's
thinking
about
doing
the
exporters,
the
metrics
exporters
for
other
for
other
formats
like
when
it
comes
to
naming
what
they've
learned
and
what
makes
sense
from
from
a
naming
perspective.
I
can't
be
learned
something
specific
about
how
we
should
be
thinking
about
naming
the
metrics.
Do
we
have
any
other
insight
there?
That
would
be
helpful,
I.
A
Think
the
biggest
thing
that
I've
heard
was
we
should
not
put
like
dynamic
names
into
metric
release.
Yeah
like
a
lot
of
systems,
do
not
work
well
in
that
model,
part
of
metric
systems.
Okay,
can
you
elaborate
what
you
mean
by
dynamic
names
like,
for
example,
for
our
perfo
mathematics
that
we
recently
added?
A
We,
for
example,
put
in
the
framework
name
and
framework
ID
in
there
in
the
metric
key
name,
which
is
really
hard
to
dashboard
I
think
most
when
systems
expect
the
names
to
be
more
or
less
well-known,
the
metal
key
name,
and
then
this
dynamic
thing
should
be
tags
kind
of
thing
and
I
think
we
do
not
have
that
kind
of
support
in
our
metrics
infrastructure
or
something
that
has
that's
probably
necessary.
If
you
have
to
really
support
different
metric
systems,
I
think
collection
systems
or
we
have
to
write
all
this
intermediaries.
Do
the
transformation
yeah.
L
H
L
D
D
Have
we've
done
anything
where
we've
tried
to
basically
take
some
of
our
metrics
and
convert
them
to
metric
name,
Plus
tags
yeah,
there's
missus
explore
there
that
does
stuff
for
that's
not
for
meters,
okay
and
and
did
we
come
up
with
a
standardized
like
mapping
that
we
can
like
apply
back
here.
Some
learning
that
we
can
apply
back
here,
there's
an
ongoing
threading.
L
D
That'll
likely
have
to
have
some
some
influence
here:
okay,
anyway,
I
gotta
jump.
This
is
what
I
wanted
to
use
that
okay
first
one
I
saw
so
thanks
for
thanks
for
the
questions
and
feedback
hooks
as.
C
M
M
We
looked
at
the
venue
in
Brooklyn
to
kind
of
keep
prices
down,
but
it's
a
really
nice
hotel,
a
nice
thing
to
see
Marion
Brooklyn,
which
has
enough
space
and
also
seems
just
from
like
the
high
screen,
show
or
stains
they're
unreasonable,
and
we
have
to
finish
up
some
more
details
and
but
expect
like
a
notice.
Shortly.
We
reached
out
to
a
number
of
you
to
just
have
like
a
broad
kind
of
like
committee,
at
least
or
content
around
so
people.
M
A
lot
of
you
should
have
gotten
emails
from
either
myself
or
Gaston
over
the
last
week's,
so
I
said
to
goal.
We
talked
about
that
a
few
months
back
already
is
to
kind
of
make
it
again
more
missus
con
to
focus
more
users
community,
in
the
hope.
Having
this
broad,
diverse
content
committee
MC
helps
what
the
direction.
F
M
That,
as
soon
as
we
have
kind
of
like
the
final
confirmation
on
the
venue
and
the
other
thing,
if
you're
on
the
Apache
missus
PMC
you'll,
probably
get
an
email
shortly
from
myself,
because
we
actually
need
your
approval
that
we
can
use
the
name
Apache
missus
calm.
So
once
we
have
said
sorted
will
send
out
see
if
Furyk
will
first
send
an
email,
just
informing
asking
hey
everyone
agrees
and
that
will
include
also
Steve
P
deadlines
is
good.
M
A
M
M
C
L
L
All
right,
so
we
did
some
tests
yesterday
and
we
had
many
marathons
running
in
that
classroom.
We
call
them
anymore
m-mom
Marxism
marathon,
and
we
enable
the
new
pour
phenol
metrics
I'm,
not
a
master
yet,
but
we
use
em
anyway
for
the
for
this
experiment
and
I
wanted
to
show
you
what
kind
of
things
it's
possible
to
see
with
these
metrics,
it's
possible
to
see
how
many
API
calls
each
framework
did
same
for
for
events.
L
That's
like
an
overview
for
all
the
all
the
frameworks
and
then
I
will
choose
one
of
these,
but
there
are
moments
and
show
what
they
break
up
to
break
down.
Post-Reagan
is
again
I
haven't
prepared
any
of
these
I
just
so
you
would
be
a
nice
chance
because
I
have
this.
That's
what's
open,
so
feel
free
to
ask
anything
you're
in
doing
this,
I
will
just
go
over
the
dashboard,
hey.
L
Just
American
framework
you
can't
forget
about
mom
just
needs
marathon
and
that
there
are
moody
ones
in
this
cluster,
so
it's
possible
to
see
the
offer
office
right,
I'm,
good,
if
I
break
down
by
sand,
except
that
the
client
understand
it.
So
if
someone
is
complaining
about
not
getting
or
all
tasks
not
being
launched,
we
can
see
if
different
start
or
not.
A
I
L
L
I
F
N
K
B
C
L
N
C
C
F
C
L
Well,
you
can
see
every
since
they're,
basically
the
one
that
you
see
below,
that
is
a
suppressed
gauge,
so
you
can
see
where
that
framework
is
suppressing
or
not,
which
can
have
performance
influence
or
in
this
case
marathon
doesn't
suppress
anymore.
That's
something
that
would
be
people
working
on
on
on
meses
atmosphere
learned
recently.
It
used
to
suppress
before
an
nad
exchange
and
I
will
try
to
get
them,
but
that
to
suppress
again
and
our
interest.
L
The
one
on
the
right
so
you're,
very
good
teacher
nice
fighting,
keel
or
anything
like
that,
and
then
you
kind
of
try
to
figure
out
what's
going
on
with
what
a
family
is
doing
so
when
I
favor
accepts
an
offer,
they
can
pass
mini
operations
there.
So
we
have
another
metric
or
our
metrics,
showing
how
many
operations
of
each
type
were
sent
by
the
framework.
That's
the
operations
rate
that
you
see
below,
and
we
have
a
program
for
events.
L
The
subjects
sent
to
the
framework,
how
many
tasks
are
in
each
state
and
that's
basically,
basically
so
there's
an
open
chain
right
now
with
these
metrics
Greg
is
working
on
them
and
I
just
wanted
to
give
you
an
overview
of
how
of
what
they
are,
what
we
want
to
expose
and
why
we
think
they
are
useful
or
you
could
use
them.
Are
there
any
questions.