►
From YouTube: Ceph Crimson/SeaStore 2021-10-03
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
All
right,
let's
get
started,
let's
see
last
week,
I
spent
time
doing
reviews
the
journal
badging
pr
for
young
gem
and
c-star
metadata
pr
for
may
I'm
sort
of
working
on
the
cache
lru
as
well,
but
I
got
sidetracked
with
other
stuff
breath.
How's
it
going.
B
Yeah,
sam,
the
testing
is
blocked,
and
last
week
I
raised.
A
Can
you
paste
the
link
to
the
bug
name?
Where
are
you
interested
in
investigating
the
problem.
B
Yeah,
actually
it's
like
a
procedure
which
is
mentioned
in
the
talk,
is
I
just
followed
the
procedure
and
the
doctor
is
also
mentioned
in
that
bug,
so
I
cannot
able
to
proceed
further.
A
Right,
I'm
asking:
are
you
interested
in
investigating
it?
If
not
it'll
have
to
wait?
You'll
have
to
wait
a
bit.
I
think
personally.
B
I'm
trying
on
with
the
surf
edm
yeah,
that
is,
that
is
what
I
mean
in
the
previous
discussions.
What
we
discussed
is
like
it
is.
The
crimson
is
mainly
targeted
for
the
rook,
so
no
it'll
be
on
both.
A
C
Last
week,
I
spent
some
time
on
the
to
identify
why
the
build
option
influenced
that
performance
testing
result.
So
I
uneasy
got
the
different
results
on
my
machine:
the
the
build
option,
no
omega
frame
pointer
without
that
build
without
that
option
and
the
batching
er
seems
getting
worse
performance,
but
in
his
environment
it
didn't
bring
the
worst
performance.
C
C
C
C
A
D
C
And
then
the
second
one
is
that
uid
still
has
some
problem
provided
up
here
and
will
modify
the
pr.
According
to
the
comments,
the
third
one
is,
that
seems,
make
fs
works
and.
C
The
the
c-stall
startup
and
the
question
stopped,
but
when,
when
the
self
cluster,
for
example,
use
this
command
to
start
this
self
osd,
I
can't
see
the
safe
osc,
starter
startup.
So
currently,
self
osc
can
now
just
start
up,
and
I
still
take
on
later
why
the
command
doesn't
work.
A
Yeah,
I'm
not
sure
how
that
command
works.
Well,
let
me
know
if
you've
got
any
questions
for
the
fsid
thing.
By
the
way
we
don't
need
a
set.
Fsid
call
classic
osd.
Does
that
I'm
honestly
not
sure
why
it
could
just
pass
it
as
a
parameter
to
manifest,
but
it
doesn't
so.
You
should
be
able
to
just
check
for
xero
right
at
the
top
of
makefs
and
randomize
it
there.
That
should
be
all
you
need
to
do.
A
A
No,
I
mean-
or
I
guess
you
I
guess
you
could
do
it
there
either
way,
but
you
don't
need
a
set.
Fsid
call
like
if
you
look
at
well,
I
made
a
comment
on
the
on
the
bug
we
can
discuss
it.
There.
A
C
I
can't
understand
it
so
so
when
we
write
to
the
super
block,
the
h,
zero
currently.
C
Okay,
okay,
so
we
check
if
it
is
zero,
but
we
still
need
to
set
fsid
to
the
super
block.
A
A
A
All
right
how's
it
going.
E
Oh
last
week
I
intended
to
replace
spread
lfs
strategy
work
to
the
master
branch,
but
I
was
occupied
by
some
other
jobs
in
my
office,
so
I
didn't
make
much
progress,
but
this
week
I
I'll
be
back
on
track
and
continuing
that
work.
Sorry
that
that's
all
for
me.
D
A
Okay,
so
if
you're
not
seeing
a
performance
regression
ying-jin,
I'm
inclined
to
go
ahead
and
merge
it.
I
know
that
we're
seeing
a
problem
in
chad,
bay's
environment,
but
we
do
actually
need
the
ability
to
coalesce
records,
so
we're
going
to
have
to
just
figure
out
what's
causing
the
performance
for
russian
chinese
environment
later
yeah.
Does
that
seem?
Okay,
yes
cool!