►
From YouTube: Ceph Performance Meeting 2021-09-23
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
Go
all
right,
so
I
don't
have
a
whole
lot
of
pull
request
updates
this
week.
There's
some,
but
not
not
too
many.
Okay,
so
closed
this
one
that
I
had
reviewed
for
kifu.
He
saw
that
our
changes
to
the
roxdb
lru
cache
are
our
own
implementation
of
it.
Essentially,
so
it's
really
just
their
lru
cache
with
some.
Some
modifications
did
not
function
with
a
new
version
of
roxdb.
They
changed
the
interface
a
little
bit,
so
he
fixed
it.
I
tested
it
it's
it
worked.
A
Well,
everything
looked
fairly
sane,
so
we
did
merge
that
the
only
thing
that
I
don't
like
is
that
the
reason
we
did
this
is
because
there
was
a
maintainer
for
some
distribution
that
wanted
to
compile
stuff
with
the
newest
version
of
rocks
to
be
rather
than
the
one
that
we've
tested,
and
I
kind
of
complained
about
this
in
the
pr
and
complained
about
it
on
the
mailing
list,
the
self
maintainers
mailing
list,
it's
a
little
scary
that
we
may
have
maintainers,
especially
compiling
stuff,
with
the
newest
version
of
roxdb.
A
It's
fine!
If
it's
just
experimentation
see
if
it
works,
seeing
if
it
improves
things
in
some
way,
we
really
really
don't
want
users
doing
that,
though,
we've
we've
gotten
burned
in
the
past
by
changes
to
rocks
db
that
needed
to
be
reverted
or
just
ones
that
changed
behavior
that
we
didn't
understand
at
the
time.
So
I
don't
know
what
the
status
is
for
fedora
or
other
kind
of
more
bleeding
edge
distributions.
A
The
ubuntu
guys
get
back
to
us
and
said
they
don't
do
this
and
the
oh,
who
was
it?
One
of
the
other
groups
also
said
that
box
mocks.
I
think,
maybe
so
that
they
don't
do
it,
but
it'd
be
good
to
know
if
any
of
the
other
distributions
are
are
using
kind
of
a
system
level
or
bleeding
edge
db
distribution,
and
probably
we
should
discourage
that
so
anyway.
A
That
was
that
pr
it
it's
fine
in
and
of
itself,
but
it
maybe
brings
up
questions
that
we
need
to
answer
updated
prs.
I
think
this
rgw
osd
compression
bypass
casey,
said
it
looked
good
last
week
looks
like
maybe
it's
been
rebased
again,
but
I
think
I
got
the
impression
that
that's
really
close
to
being
merged
and
should
be
in
pretty
good
shape.
A
B
I
don't
really
count
this
as
a
review.
It's
just
he
just
looked.
She
saw
something
commented
I
don't
think
he's
in
entirety
of
it.
That's
much
too
few
errors.
He
pointed
out
to.
A
B
A
Okay,
that
sounds
good
all
right.
What
else
do
we
get?
Oh
another,
one
from
you,
adam
incremental
update
mode
for
bluefest,
looks
like
that's
that
interesting
now,.
A
D
A
Yeah,
I
think
they
had
a
couple
of
questions
to
whether
or
not
this
was
useful
or
sufficient
for
for
what
might
be
helpful
on
your
end,
so,
okay,
yeah
I'll.
A
All
right,
let's
see
last
one
I
have
in
my
updated
list-
is
igor's
pg
removal,
optimization.
It
looks
like
there
was
some
additional
review
and
testing
there
adam.
I
think
you
had
maybe
noticed
like
a
small
thing
in
that
pr
that
need
to
be
changed.
A
And
there
was
some
discussion
there
about
whether
or
not
something
could
be
done
for
hard
drives,
rather
in
addition
to
ssds,
but
I
think
the
the
general
feeling
was
that
it's
not
really
good
to
encourage
people
to
run
rxdb
on
hard
drives
anyway.
So
it
may
not
be
worth
spending
a
lot
of
time.
Optimization
optimizing
for
that.
That
use
case.
B
A
I
see
I
may
have
misinterpreted
that
then
I
thought
that
the
the
thought
was
that
it
was
optimization
on
the
with
roxy
beyond
hard
drive,
but
maybe
I'm
wrong.
Okay
well,
in
an
even
some
a
little
bit
more
of
review
and
discussion
on
that.
I
don't
know
if
the
test
failures
ever
were
resolved
or
not.
I
think
that's
the
maybe
the
the
next.
D
A
D
The
last
time
it
was
tested,
it
was
failing
test,
so
I
think
it's
time
for
ego
to
review
those.
A
All
right
lots
of
stuff
in
the
no
movement
column,
some
of
the
new
mds
stuff,
already
needs
a
rebase
unfortunate
from
so
we'll
just
have
to
see
if
he
can,
if
he
can
keep
up
on
those
cash
binning
rebase,
I'm
working
on
that
now
in
a
separate
branch
I
was
able
to
rebase
on
top
of
master
without
a
whole
ton
of
work.
A
I
had
done
a
rebase
like
six
or
eight
months
ago
or
something
so
it
hasn't
been
too
bad
to
to
rebase
that
again,
I'm
going
through
I'm
trying
to
go
through
some
more
extensive
testing.
Now
on
that,
since
now
I
have
a
way
to
do
much
more
aggressively
test
omap
I
can.
I
can
kind
of
alternate
between
heavy,
oh
no
testing
and
heavy
omap
testing,
which
is,
I
think,
where
we
will
see
the
biggest
benefit
from
that
pr.
A
If
we
see
a
benefit,
let's
see
what
else
not
a
whole
lot
of
other
things
again.
The
subtree
map
from
journal
update
for
mds
needs
to
rebase.
Again,
I
think
the
rgw
team
are
working
on
this
shared
object,
cache
pr,
including.
A
Of
other
fixes,
so
that's
just
kind
of
sitting
there
right
now.
I
don't
think
I
need
to
do
anything
on
that.
What
else.
A
A
Yeah,
that's
kind
of
about
it
for
right
now,
all
right,
I
don't
really
have
a
whole
lot
of
discussion
topics
this
week,
I'm
personally
just
working
on
this
age
bidding
pr
which
I'll
have
more
to
talk
about
at
some
point,
but
not
right
now,
so
maybe
I'll
open
it
up.
Are
there
any
pull
requests
or
any
projects
that
people
would
like
to
discuss
this
week
at
the
meeting.
A
If
not
that's
okay,
too,
then,
the
only
other
thing
I'll
remind
folks
is
that
next
week
we
are
going
to
discuss
the
paper
that
josh
found
for
us
about
re-architecting
the
linux
storage
stack.
A
Questions,
oh
not
for
me
I'll,
be
I'll
I'll,
be
waiting
for
josh's
analysis
and
I'll
I'll
go
to
the
paper
and
then
and
then
be
excited
to
learn
from
from
other
people
on
this.
So
no
no
quiz
just
do
your
best.
That's
the
way
it
is
with
all
these
papers
usually
and
then
we'll
we'll
see.
E
Okay,
okay,
I'll
I'll,
try
to
make
sure
sam
can
see
you
sam
can
join
too
she
likes
to
enjoy
these
discussions.
A
Yes,
yes,
and-
and
don't
don't
be
worried
about
impressing
me-
be
worried
about
impressing
stan
because
he's
the
one
that
will
that
will
task
all
of
us
with
with
thinking
thinking
more
deeply
about
about
these
kinds
of
things.
So.
A
All
right
well,
then,
have
an
excellent
week,
everyone
and
be
ready
for
the
paper
next
week
and
and
we'll
see
you
then.
F
Thank
you.
Thank
you
guys
come
to
the
meeting
with
like
three
takeaways
from
the
paper
or
something
for
for
discussions
sake.
It
might
be
good
to
come
to
the
meeting
with
like
three
takeaways
from
the
paper
or
not.
You
know
just
to
sort
of
like
these
are
three
things
I
I
I
found
from
the
paper
that
were
interesting.
I
don't
know,
might
be
good
to
know
your
discussion.
That
way.
A
Yeah,
that's
a
that's
a
reasonable
take
in
the
past.
I
I've
done
that
kind
of
thing
and
it's
led
me
to
to
usually
be
pessimistic,
which
I
don't
know
if
it's
good
or
bad,
but.
A
All
right!
Well,
then,
yes,
that's
a
good
suggestion,
laura,
let's,
let's
try
it
we'll
see,
we'll
see
what
people
come
back
with
optional,
of
course,
but
but
if
you
you
can
then
yeah
think
about
think
about
what
your
your
takeaways
from
the
paper
are.
All
right
have
a
great
week.
Everyone
see
you
next
week.