►
From YouTube: GMT 2018-07-10 API WG
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
A
A
A
A
A
D
A
A
A
D
D
I
guess
in
our
environment
the
we
don't
really
have
a
very
good
metrics
collection
story.
Right
now
we
have
a
bunch
of
legacy
stuff,
but
we're
moving
towards
Prometheus
using
the
base
exporter.
So
we,
if
we
event,
probably
in
the
next
week
or
so
hopefully
we'll
be
pushing
that
out,
in
which
case,
if
you're,
looking
at
your
last
bench,
marks
after
optimizations
with
2000
frameworks,
you
know
the
metric
get
master,
get
metrics
response
is
22
seconds.
That
seems
for
other,
that
sort
of
seems
under
playable
right,
yeah.
A
D
D
D
A
So
I
do
know
from
our
own
experience.
The
maysa
exporter
has
just
a
simple
fetch
time
out
for
getting
the
metrics
from
ASOS
and
on
the
exporter
end
you
just
have
to
be
concerned
that
the
fetch
time
doesn't
go
above
the
time
out
that
you've
set,
but
I
guess
that's
also
related
to
the
interval
of
fetching.
You
know
if
you
want
to
fetch
metrics
on
some
interval,
that's
much
shorter
than
the
fetch
time
that
maybe
becomes
an
issue.
D
I
think
I
fetched
is
likely
to
be
a
minute
like.
A
A
C
That
was
one
of
the
things
that
was
pretty
costly
was
that
there
are
so
many
futures
involved
and
for
synchronous,
metrics.
We
don't
actually
need
a
future,
so
we
could
split
that
interface
so
that
synchronous
metrics
expose
directly
a
double
instead
of
the
future
novo
uh-huh.
That
was
one
of
the
things
that
shows
up
in
perf.
D
C
Thought
is
like
there's
some
additional,
pretty
easy
improvements
we
can
make
here
so
I
guess
we
can
see
how
low
we
can
get
it
as
long
as
the
efforts
reasonable
and
then,
if
that,
like
I'm
kind
of
including
not
doing
poll
grades
transition
in
there
and
then,
if
that's
still
bad,
we
could
okay,
also
whether
those
pull
gauges
are
thinking
with
that
I'm
or
not
I,
don't
actually
know
off
the
top.
My
head
yeah,
so
one
there's.
A
C
A
D
That
makes
sense
I
guess
it's
been
a
few
years
since
I've
had
since
I've
had
a
matrix
collection
with
infrastructure
which
had
the
allocation
time
but
yeah
that
that
totally
makes
sense.
One
thing
about
one
thing
about
our
environments,
which
might
be
different
from
from
the
environments
you're
testing
in,
is
that
almost
all
of
the
frameworks
that
are
running
suppressed.
D
A
A
Great
yeah
and
I
will
say
also
on
the
Mesa
sphere
side.
We're
gonna
continue
doing
some
metrics
work
in
the
coming
months,
so
I
would
be
interested
in
in
meeting
with
you
as
time
goes
on
to
figure
out.
If
you
know,
if
we
can
align
our
our
interest
somewhat,
and
maybe
you
know
I
think
in
OSS
mesas,
it
would
be
nice
to
have,
as
we've
discussed
before
some
some
richer
metric
format
that
can
be
consumed.