►
From YouTube: 2020-04-21 :: Ceph Crimson 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).
A
We
can
we
can
we're
gonna
choose
one.
The
meeting
of
both
of
them
at
will
to
fit
your
calendar
needs.
Last
week.
I
was
doing
curve
you
and
new
people
at
rehabilitation
and
and
as
a
estimation
that
we
it's
a
it's.
It's
fine
to
spear
spear,
the
the
embedded
node
own
note
at
will.
So
we
don't
need
to
care
about
the
double
split
case.
Let's
simplify
the
a
lot,
but.
B
C
B
D
B
B
C
B
C
Depends
on
how
you
structure
at
the
interior
of
the
tree
space
you
could
make
it
so
that
the
individual
nodes
are,
for
instance,
chunked
or
you
can
make
it
so
that
every
node
is
assumed
to
be
up
to
some
size
and
every
and
anything
after
that
gets
shut
up
to
another
one
you
would
just
have
to.
You.
Just
have
to
be
clever
about
how
you
choose
the
internals
of
the
of
the
Onoda
notes
you.
C
C
You
could
do
other
things
like
if
you
know
that
some
parts
of
the
oh-
no
to
our
variable
size
to
another
parts
of
fixed
size,
you
could
factor
the
O
node
into
two
pieces.
The
fixed
size,
part
of
the
variable
size,
part
and
at
the
fixed
size
part
goes
in
one
portion
of
the
allocated
node
and
the
variable
size.
Port
parts
can
be
sort
of
filled
in
in
the
rest
of
the
space
that
you
don't
have
to
constantly
realloc
I
suggest
that
you
start
with
the
simplest
thing.
B
D
A
But,
but
to
me,
I
just
want
to
remind
that
what
what
but
anything
I
are
working
on
is
that
synchronized
version
of
the
prototyping?
So
you
can
you
can
if
you
can't,
you
want
to
design
based
on
it,
you
better,
you,
the
better
assume,
that's
the
our
finalize.
The
design
will
be
returning
future
when
it
comes
to
facing
cops
to.
E
E
F
E
A
C
The
route
certainly
will
be,
and
in
the
event
that
the
that
the
allocation
is
simple,
for
instance,
and
rgw
object
that
got
written
all
at
once.
Then
it'll
be
one
entry
deck,
in
which
case
there
shouldn't
be
any
children
notes.
It
should
be
fully
embedded
in
the
O
node,
so
just
ensure
that
your
that
your
root
node
can
be
arbitrarily
small
up
to
one
up
to
one
entry.
E
E
A
G
G
Okay,
but
the
best
one
test
at
least
another
one
is
a
showing
is
still
running
from
a
long
time
now
and
showing
some
JSON
dumps
error,
or
something
like
that.
I,
don't
know
what
it
is,
but
I
feel
I
probably
find
out
what
what
happened
there,
but
microspace.
So
at
least
the
basic
problems
were
solved,
and
so
thanks,
everyone
went
yesterday.
C
Okay,
so
I
have
the
unit
test
harness
up
to
a
point
where
it
can
make
FS
and
then
replay
in
mountain,
just
the
very
basic
writing
down
the
route
block
and
the
elbe,
a
route,
blocker
and
I
believe
I
have
all
the
code
required
to
do
insertions
and
splits
within
the
LBA
tree.
So
I'm
going
to
finish,
writing
those.
You
know
tests
and
then
I
think
I'm
going
to
write
some
documentation
and
then
push
a
PR.
We
won't
have
the
ability
to
remove
mappings.
C
A
C
I'd
literally
haven't
written
that
code,
yet
that's
that
that's
all
I've
written
like
60%
of
it,
but
the
top
bootstrapping
stuff
they
haven't
done
yet.
I
also
haven't
written
these
stuff
for
replaying
deltas
against
the
LDA
blocks.
It's
a
matter
of
writing
down
the
app
like
writing
encoded
versions,
the
operations
for
inserting
and
removing
entries
for
a
block,
but
that
also
is
just
working
in.
C
C
The
the
code
for
pulling
for
actually
literally
replaying
Delta's
out
of
the
journal
exists
and
all
of
the
code
required
for
formatting
and
writing
to
the
journal
exists.
But
the
part
after
that,
where
you,
when
you
pull
a
delta
off
disk
it
you
you
check
the
type
and
then
pull
the
relevant
replay
code
for
that
type.
That
replay
those
replay
operations
don't
exist
yet
for
underneath
the
block
types,
but
the
code
for
doing
it
does
so.
C
D
C
C
F
B
With
this
with
this
layout-
and
this
is
a
link
here-
I
have
jaw
the
diagram
illustrates
the
layout
and
my
plan
is
to
firstly
consider
the
space
efficient
rather
than
can
be
computationally
efficient,
because
most
of
the
structures
are
not
aligned
and
the
most
probably
will
firstly
implement
no
types,
0
and
3
for
simplicity
and
ignore
1,
&,
2
and
I
will
start
to
develop.
This
look
up
logic.