►
From YouTube: 2019-04-16:: Crimson SeaStor OSD Weekly 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
B
Yep
I'm
I'm
going
through
failures
as
they
happen
and
rescheduling.
So
a
few
things
have
come
so
far.
I
probably
expect
a
few
more.
It
was
a
big
patch
series,
I'm
actually
already
working
on
the
integration
and,
if
I'm
do
well,
the
tests
are
running
right,
so
I
got
as
far
today
as
making
it
build
as
part
of
the
crimson
in
crimson,
cotton
or
whatever,
wasn't
too
bad,
the
only
real
did.
The
only
sort
of
hiccup
was
I'm.
B
C
Mostly
things
ready
to
performance,
let's
start
from
the
execution
stage
in
do
OSD
ops,
this,
the
comparing
this
very
simple
execution
stage,
simple.
Well,
the
comment
is,
is
few
lines,
but
it's
quite
promising
I
would
say
well,
we
are,
of
course
we
are
higher
latency
and
in
single
loud
generator
test.
C
C
Internally,
I
was
taking
a
look
on
how
on
how
those
execution
stages
are
implemented
internally
in
sister,
and
it
turns
out
that,
basically,
you
are.
You
have
a
couple
of
constraints.
First,
one
is
the
is
that
it's
the
data.
You
are
PI
your
function,
the
method
you
are
making
you
are
staging
to
have
very,
very
limited,
are
impaired.
My
parameter
set.
It's
because
the
it's
copied
it
goes
for
a
special
kill
it
in
a
special
kills
inside
inside
sister,
the
less
the
the
lesser
arguments,
the
shorter
arguments,
the
better
for
you
also.
C
There
is
each
call
to
execution
stage,
Krauss's
skid
schedules,
a
task,
and
in
sister
k2
that
task
and
also
instant,
instant
Yates
this
this
task
derivative
each
time
it
sucks
it's
such
it's
opportunity,
but
well
not
for
now,
I
think
second
topic
related
to
performance
is,
is
another
optimization
in
Utah
in
the
future
library
right,
a
comet.
It's
one
liner.
C
C
C
Within
lining
possibility,
you
are
not
going
the
actor
you
are
bypassing.
However,
it's
it
doesn't
work
when
you
have
unsatisfied
future,
then
in
in
such
case,
each
each
element
of
your
continuation
change
needs
to
go
for
reactor.
It
means
that
you
not
you
are
you
have
you
have
to
go
with
dynamic
these
parts
with
tables,
but
most
you
are
going
to
go
to
reactor.
Put
your
dust
inside
a
queue,
all
that's
stuff,
but
the
most
limiting
I.
Think
I.
C
C
C
Series
of
methods
responsible
for
doing
that
when
you
are
when
somebody
calls
but
then
or
then
wrap
something
like
that
on
future,
it
checks
whether
the
value
is
already
available.
If
it
is,
there
is
an
optimum
territory.
There
is
ultimate.
There
is
a
trick,
the
third,
but
if
not,
then
you
are
doing
a
continuation.
There
is
a
class
of
drift
from
completion
base.
C
You
are
scheduling
a
continuation
that
forwards
the
forwards,
the
value
you
get
from
this
Ingle
continuation
to
another
one,
but
if,
but
when
the,
when
the
K,
the
continuation
is
finally
scheduled,
it
wants
to
forward
its
value
to
another
element
of
in
this
inside
the
continuation
chain
and
each
it
should
be
possible
to
call
this
die
not
through
not
schedule.
Another
task.
B
C
C
C
A
C
B
On
board
with
that,
because
I
want
hearing
state
already,
has
them
so
I'd
actually
go
find.
D
B
A
C
C
B
C
B
A
A
A
D
I'm
working
though
last
week,
a
tip
home
finished
the
messenger
tested
across
the
server
the
remoter
locally.
So
on
the
local
test
case
and
most
cases,
crimson
messenger
is
better
than
a
single
messenger
and
the
photo
remote
means
crossed
not
working
he's
in
crossed
the
servers
and
maybe
50%
cream
semester
is
better
than
a
sink
50%,
a
it's
worse
than
I.
Think
so,
based
on
that
test,
without
I'm
trying
to
set
up
a
test
environment
I'm
not
used
to,
we
start
adjust
a
cross
the
server
to
try.
D
A
A
D
So
from
the
good
night
instructor
on
the
southwest
you
and
the
starter
book,
a
criminal
sths
gotta
admire
to
the
riddles
bench
and
accounted
the
message.
I'm
pester
here
there
is
Fault
you're,
accepting
battlenet
search
the
source
code,
I,
don't
know
who
and
now
this
log,
so
the
the
riddles
bank
cannot
connect
it
a
tutor
to
the
crewman
so
SD
in
my
situation.
So
the
currently
it
couldn't
decide
happen
so
that
you
mean
not
that
they're.
The
PG
problem
right,
PG
purine,
is
now
finished.
A
D
D
A
C
C
C
C
B
B
B
C
B
C
B
That's
the
thing
it's
actually
not
strictly
speaking
ugly.
We
actually
did
refactor
it
and
improve
yours.
Go
into
like
a
state
chart.
You
sort
of
deal
I
see.
Oh
yeah,.
B
Fairly
tolerable,
it
could
be
better,
but
we
would
get
to
that
point
by
refactoring
it
not
by
rewriting
it.
C
B
B
C
C
B
B
B
B
C
B
A
C
C
B
C
B
C
C
B
B
C
C
C
D
Not
Joey
because
the
Wi-Fi
networking
is
doesn't
a
table
and
has
a
question
and
proposal
here.
So
what's
our
next
plan
and
he
his
plan,
he
wants
to
implement
to
the
right,
I'll
pass
first
and
then
to
get
to
know
in
the
bottleneck
of
the
OST.
So
that
is
cut
open
question
or
the
next
next
plan,
though,
and
he's
open
it's
implemented
right,
I,
hope
us
and
our
Friday.
My
uncle
said:
hey,
we'll
e-mail,
you
or
and
contact
you
later.
B
D
D
C
D
C
B
D
B
B
Oh
I
see
okay,
so
what
that
does
is
as
soon
as
it
calls
wait
for
active.
It
immediately
sets
active.
So
it's
just
not
waiting
for
the
peering
machinery
to
become
active.
It's
just
doing
it,
which
is
fine.
You
you
get
in.
You
will
get
inconsistent
data,
but
it's
performance
testing,
so
whatever
exactly
yeah
that
that
seems
like
an
expedient
and
reasonable
solution
to
the
problem.
Cool.