►
From YouTube: Ceph Crimson/SeaStore Meeting 2022-07-06
Description
Join us weekly for the Ceph Crimson meeting: https://ceph.io/en/community/meetups
Ceph website: https://ceph.io
Ceph blog: https://ceph.io/en/news/blog/
Contribute to Ceph: https://ceph.io/en/developers/contribute/
What is Ceph: https://ceph.io/en/discover/
A
All
right,
let's
get
started
so
first
off
an
announcement.
Yingjin
has
agreed
to
become
the
c-store
elite,
so
he
will
be
you'll,
be
doing
that
gonna,
the
big
leader
I,
think
in
this
door.
So
this
will
be.
This
will
be
good.
We
got
perhaps
more
more
review
throughput.
B
A
You
all
right
for
what
it's
worth
I'll
still
be
working
on
c-store
and
I'm.
Still
the
Crimson
late
I'm
not
going
anywhere.
Don't
worry
about
that
all
right
on
to
the
weekly
meeting.
Let's
see
this
week
has
been
mostly
reviews
and
multi-core
for
me
Fair
event,
how's
it
going.
A
All
right,
let's
come
back,
Kingston
how's
it
going.
B
Yeah
last
week,
I've
discussed
the
new
design
of
the
extent
indexing.
Since
we
got
an
agreement
about
about.
B
Okay
yeah,
since
we
can
can
go
in
the
agreement
of
the
design
and
I'm
I'm
also
reading
the
random
blog
manager
called
most
currently
most
in
this
circular.
Journal
implementation
and
I
just
sent
out
the
fix
there
and
I
will
continue
to
review
the
follow-up
dream
PR
this
week.
That's
all
okay,.
B
C
I
I
I'm
continuing
to
work
on
the
zns
part
of
it,
and
so
I
I
sent
a
a
a
PR
with
three
patches
last
week.
C
I
saw
that
make
check
failed
there,
but
I
I
think
it
is
failing
because
in
the
make
check
in
GitHub
whatever
it
is
running,
it
is
running
with
have
zbd,
but
in
the
make
file
it
is
like
have
like
when
I
compile
it
I
compile
it
with
like
with
the
zns
when
I
do
the
before
the
build
when
I
do
make
check
make
this
one
right
so
that
I
I
run
it
with
the
hyphen
with
DNS
and
I.
C
A
C
Right,
yeah
yeah,
so
so
yeah.
So
that
is
one
thing
the
yeah.
The
second
part
is
that,
like
I
updated
last
week,
I
I
saw
that
for
zns
zones
we
cannot
do
an
in
update
of
tail
segment
and
tail
segment
information
at
the
end
of
the
Zone,
because
right
pointer
would
be
somewhere
in
the
middle,
and
we
cannot
write
it
again.
C
So
my
idea
was
at
that
time
to
write
detail
information
wherever
the
right
point
was
and
then
continue
to
close
that
zone
and
then,
whenever
read
happens,
I'll
just
recalculate
it
from
the
current
right
pointer
and
then
go
back
whatever
that
size
of
the
time
and
then
read
it
from
there,
but
turns
out
that,
since
close
in
terms
of
crimson
segments
is,
is
essentially
a
finish
for
a
in
a
CNS
terminology,
because
if
we
keep
just
the
Zone
in
a
closed
state,
that
would
contribute
to
a
like
a
Max
active
zones
which
has
a
limit
on
actual
zns
hard
drives
and
the
drives
that
currently
I
have.
C
It
has
a
limit
of
like
14
maximum
active
active
zones
limit.
So
after
closing,
after
14
zones
are
in
closed
state,
it
won't
allow
15
Zone
to
be,
you
know,
opened
and
be
returned
to
so
so
what
happens
is
now
if
I
just
close
it
and
then,
if
I
just
close
it,
then
I
won't
be
able
to
open
it.
Next
Zone
after
the
max
number
of
zones
are
opened
or
in
closed
state.
C
So
if
I
do
a
finish,
then
the
problem
is
that
the
drive
moves,
the
right
pointer
to
the
end
of
the
zone
so
and
it
moves
the
Zone
straight
to
full
and
then
moves
the
right
pointer
to
the
end
of
the
Zone
on
a
as
per
the
spec.
Also
defines
it
that
way.
The.
B
C
Okay
so
so
now
the
problem
becomes
that
I
don't
know
where
I
have
written
the
type
information
so
so
either
I
have
to
you
know,
cache
it
and
then
write
that
information
into
another
segment,
another
segment
or
a
zone
or
basically
I,
have
to
maintain
that
information,
and
that
would
you
know,
introduce
more
complexity
about
like
how
to
handle
it
in
a
powerful
kind
of
a
scenario.
Yeah.
A
C
No,
we
cannot
write
past
right
pointer.
You
have
to
always
write
that
right,
so
yeah,
so
one
solution,
I'm
considering
now,
is
that,
since
my
understanding
is
that
once
we
decide
to
close
a
Zone
only
then
we
write
the
title
information
and
also
once
we
close
the
Zone.
We
are
not
going
to
open
that
zone
for
writing
again.
That
is
my
understanding.
In
that
case,
we
can.
You
know.
C
C
See
essentially,
what
happens
is
even
within
the
drive
when
we
do
a
finish
operation,
if
the
let
us
say
it
zone
is
like
1
GB,
1
GB
size
and
if
the
right
pointer
is
at
512
MB.
If
we
issue
a
finish,
then
internally,
The
Zone,
the
firmware
writes
that
much
amount
of
data
to
the
500
MB
data
to
that
zone
and
then
only
that
zone
kind
of
goes
into
here.
Initiative.
C
Drive,
it
actually
writes
that
it
actually.
So
that's
why
we
are
the
the
finished.
Latencies
are
a
little
bit
high
for
zones
which
are
partially
filled,
so
if
you
I
mean
we
can
verify
that
if.
C
A
A
I
know,
let's
just
assume
we're
finishing
it
since
you
literally
have
to
in
order
to
keep
using
the
drive
like
in
the
in
the
context
we're
talking
about.
If
you
don't
finish
the
Zone
before
moving
on,
then
you
can't
keep
so,
let's
just
assume
we're
finishing
it.
So
if
you
only
write
half
the
zone
and
then
finish
it,
it
literally
writes
zeros
to
the
rest
of
the
zone.
A
C
We'll
have
to
keep
that
in
mind.
It's
a
bit
weird,
weird
Behavior,
but
at
least
the
the
the
the
the
drives
that
I
have
I
have
confirmed
with
you
know
the
firmware
Team
also,
and
that's
how
it
is
interesting.
A
A
C
I've
tested
it
and
verified
that
it
is
writing
at
the
end
of
The
Zone,
but
what
I
saw
was
that
at
least
the
initial
zones,
when,
when
we
do
a
mount
from
the
segment
cleaner
code,
it
essentially
closes
a
segment
which
is
almost
empty.
In
the
sense
it
has
written,
only
a
small
amount
of
data,
and
then
it
closes
that
zone,
which
kind
of
you
know,
makes
the
entire
you're.
A
C
Yeah
yeah
yeah,
so
yeah
so
yeah,
currently
I
am
I
was
not
able
to
get
the
mkfs
running
at
all
so
far
it
was
crashing
before
it
could.
You
know,
because
of
writing
the
trying
to
write
the
tail
at
the
end
and
then
trying
to
read
the
text
at
the
end
when
the
title
itself
was
not
written
and
not
a
host
of
these
issues.
But
now,
with
this
all
these
changes,
I'm
able
to
you
know,
run
the
Crimson
OSD
with
the
zns
drive
I'm
using
a
null
block
device.
C
As
of
now
to
you
know,
complete
the
mkfs,
and
you
know
it
doesn't
crash
so
far.
C
A
A
A
C
C
No
currently,
we
do
I
mean
in
the
zns
part,
so
whenever
the
segment
manager
does
this
segment
close
right,
so
in
the
block
part
of
the
code,
it
basically
just
moves
the
segment
information
into
closed.
B
C
A
Yeah
request
a
review
for
me:
okay,
so
if
you
click
on,
if,
if
I've
already
reviewed
it
you'll
have
to
click
it
again
to
re-request.
A
That
no
no
worries,
that's
how
I'm
finding
reviews
I
need
to
do
these.
A
Okay,
sorry
jinyu.
Maybe
that
is
that
right,
welcome,
I'm,
not
sure.
D
D
D
D
So
there
are
three
GC
paths
that
might
influence
our
education,
that
is
a
trim,
Journal
trim
back
Ray
tree
and
the
reclaimed
space
the
aerial
Behavior
might
occur
when
metadata
updates
or
extends
rewrite
it
in
this
GC
transaction.
Then
the
methods
get
extend
and
a
complete
commit,
so
the
choose-
a
transaction
judgment
could
be
added
to
the
above
logic.
A
I
agree
completely:
I
have
the
following
cautions:
it
does
need
to
go
into
the
cache
for
correctness
reasons,
but
it
does
not
need
to
go
into
the
lru
list.
I
believe
you
have
pointed
that
out
correctly.
It's
touch
extent,
I.
Think
I.
Think
you
found
the
right.
The
right
call,
that's
the
right
one
not
to
do.
A
Yeah
I
I
agree.
Yes,
like
I
said
it
needs
to
be.
It
needs
to
remain
in
the
cache
until
the
until
the
transaction
commits,
but
you're
correct.
It
does
not
need
to
go
into
the
lru
list
that
way
when
the
GC
transaction
completes.
If
Unless,
somehow
it's
gained
a
reference
from
somewhere
else,
it'll
drop
out
of
cash,
which
is
what
we
want.
Yep,
that's
right.
D
Okay,
you
open
this
link,
I'm,
not
sure
I
said
it
right.
A
Okay,
thank
you,
cool
hello,
how's.
It
going.
B
Because
I
was
trying
to
implement
the
fixed.
A
Kbb
tree
optimization,
we
discussed
last
week.
That's.
A
All
right
sounds
good,
well,
welcome,
again,
Juan,
looking
forward
to
seeing
you
in
the
future.
Anyone
have
anything
else,
did
I
miss
anyone.
C
A
For
now
for
zns
drive,
if
you
can,
if
you
can
get
it
running
on
vsert,
that's
good
enough.
This
is
going
to
be
a
little
bit
tricky
because
we
don't
have
a
good
way
of
testing,
specifically
the
code
you're
working
on
in
any
scenario
in
in
any
environment
other
than
your
machine
and
the
one
machine
we
have
CNS
Verizon,
so
yeah
any
code.
You
change
outside
of
zns,
though
there
are
a
lot
of
unit
tests,
but
I'd
like
to
encourage
you
to
add,
add
tests
any
bugs.
A
C
Right,
okay,
yeah
so
I
understand
about
that
in
the
unit
tests.
But
I
was
like
more
asking
about
running
some
workloads.
Now,
I
I
have
the
you
know:
voicities
I,
think
I
have
the
all
the
MDS
and
all
the
other
demons
up
and
running
and
whatever
either
drive
and
then,
if
I
could
write
something
into
that,
then
that
I
could
validate
whether
rights
are
going
properly
and
that
kind
of
thing
positions.
A
Well
again
for
validating
rights
that
would
be
well
running,
one
of
the
White
staff
test
rados
or
something.
But
if
you
want
to
test
performance
or
a
workload
of
that
nature,
the
most
popular
strategy
is
to
use
something
like
fio
with
RVD
or
yingsen.
Do
you
want
to?
Can
you
link
the
the
the
performance
test
repository
you
guys
have
I
think
it's
in
CBT,
it's
a
little
tough
to
use,
but.
B
Yeah
so
yeah,
mostly
I,
think
rado's
bench
or
rbdf
file
should
be
good
enough.
Okay,
that's
just
two!
Let
me
see
we
need
to
cut.
B
B
A
The
easiest
thing
for
you
to
do
might
be,
if
you
aren't
already
familiar
with
stuff,
would
be
to
start
up
a
perfectly
normal
classic
OSD.
If
you
start
cluster
and
then
make
sure
you
understand
how
to
create
an
RPD
image
and
mount
it
and
run
fio
against
it
once
you
have
that
down,
you
can
then
use
a
chromosome
cluster.
Instead,
that'll
help
you
to
get
used
to
the
tool
chain
before
you
have
to
throw
in
the
trying
to
get
Crimson
to
work.
Part.
B
You
yeah,
so
anything
you
can
also
reach
me.
C
I
was
asking
some
questions
on
how
to
run
how
to
test
our
right
rights
on
the
zns.
You
know
part
of
it,
so
everyone
was
like
kind
of
giving
me
suggestions.
B
A
All
right,
let's
get
started
so
first
off
an
announcement.
Yingjin
has
agreed
to
become
the
c-store
elite,
so
he
will
be
he'll,
be
doing
that
that
the
big
leader
I
think
in
the
store.
So
this
will
be.
This
will
be
good.
We
got
perhaps
more
more
review
throughput.
B
Yeah
Sam,
thank
you
for
all
this
announcement.
A
You,
but
for
what
it's
worth,
I'll
still
be
working
on,
C
store
and
I'm.
Still
the
Crimson
late
I'm
not
going
anywhere.
Don't
worry
about
that
all
right
on
to
the
weekly
meeting.
Let's
see
this
week
has
been
mostly
reviews
and
multi-core
for
me,
Irvine
how's
it
going.
A
Ervin,
can
you
hear
us
all
right?
Let's
come
back
Kingston
how's
it
going.
B
Yeah
last
week,
I've
discussed
the
new
design
of
the
extent
indexing.
Since
we
got
an
agreement
about
about.
B
Okay,
yeah
things
that
we
can
can
go
in
the
agreement
of
the
design
and
I'm
I'm.
Also
reading
the
random
blog
manager
called
most
currently
most
in
this
circular.
Journal
implementation
and
I
just
sent
out
the
fix
there
and
I
will
continue
to
review
the
follow-up
dream
PR
this
week.
That's
all
okay,.
A
Cool
aravind
I
think
that
was
you
just
remembered,
Maybe.
B
C
I
I
I'm
continuing
to
work
on
the
zns
part
of
it,
and
so
I
I
sent
a
a
PR
with
three
patches
last
week.
I
saw
that
make
check
failed
there,
but
I
I
think
it
is
failing
because
in
the
make
check
in
Gita
whatever
it
is
running,
it
is
running
with
have
zbd,
but
in
the
make
file
it
is
like
have
like
when
I
compile
it
accompan
it
with
like
with
zns
when
I
do
the.
C
Before
the
build,
when
I
do
make
check
make
this
one
right
so
that
I
I
run
it
with
the
hyphen
with
DNS
and
I.
Don't
get
that
issue
whatever
that
make
check
is
complaining
for
the
pr
that
I
said
so.
I
am
I
will
send
a
I
think
I
I
can
set
a
patch
to
fix
that
make
file.
Yeah.
A
C
Right,
yeah
yeah,
so
so
yeah.
So
that
is
one
thing
the
yeah.
The
second
part
is
that,
like
I
updated
last
week,
I
I
saw
that
for
zns
zones
we
cannot
do
an
in
update
of
tail
segment
and
tail
segment
information
at
the
end
of
the
Zone,
because
right
pointer
would
be
somewhere
in
the
middle,
and
we
cannot
write
it
again.
C
So
my
idea
was
at
that
time
to
write
detail
information
wherever
the
right
point
was
and
then
continue
to
close
that
zone
and
then,
whenever
read
happens,
I'll
just
recalculate
it
from
the
current
right
pointer
and
then
go
back
whatever
that
size
of
the
time
and
then
read
it
from
there,
but
turns
out
that,
since
close
in
terms
of
crimson
segments
is,
is
essentially
a
finish
for
a
in
a
gns
terminology,
because.
A
C
We
keep
just
the
Zone
in
a
closed
state
that
would
contribute
to
a
like
a
Max
active
zones
which
has
a
limit
on
actual
zns
hard
drives
and
the
drives
that
currently
I
have.
It
has
a
limit
of
like
14
maximum
active
active
zones
limit.
So
after
closing,
after
14
zones
are
in
closed
state,
it
won't
allow
15
Zone
to
be,
you
know,
opened
and
be
returned
to
so
so
what
happens
is
now
if
I
just
close
it
and
then,
if
I
just
close
it,
then
I
won't
be
able
to
open
it.
C
So
if
I
do
a
finish,
then
the
problem
is
that
the
drive
moves,
the
right
pointer
to
the
end
of
the
zone
so
and
it
moves
the
Zone
straight
to
full
and
then
moves
the
right
pointer
to
the
end
of
the
Zone
on
a
as
per
the
spec.
Also
defines
it
that
way.
The.
B
C
So
so
now
the
problem
becomes
that
I
don't
know
where
I
have
written
the
type
information
so
so
either
I
have
to
you
know,
cache
it
and
then
write
that
information
into
another
segment,
another
segment
or
a
zone
or
basically
I,
have
to
maintain
that
information,
and
that
would
you
know,
introduce
more
complexity
about
like
how
to
handle
it
in
a
powerful
kind
of
a
scenario.
Yeah.
A
C
We
cannot
write
past
right
pointer.
You
have
to
always
write
that
right,
so
yeah,
so
one
solution,
I'm
considering
now,
is
that,
since
my
understanding
is
that
once
we
decide
to
close
a
Zone
only
then
we
write
the
title
information
and
also
once
we
close
the
Zone.
We
are
not
going
to
open
that
zone
for
writing
again.
That
is
my
understanding.
In
that
case,
we
can,
you
know,
move
the
right
pointer
till
the
end
of
the
Zone
by
writing
some
adding
information
and
then
once
it
reaches
there.
C
C
See
essentially,
what
happens
is
even
within
the
drive
when
we
do
a
finish
operation,
if
the
let
us
say
zone
is
like
1
GB,
1
GB
size
and
if
the
right
pointer
is
at
512
MB.
If
we
issue
a
finish,
then
internally,
The
Zone,
the
firmware
writes
that
much
amount
of
data
to
the
500
MB
data
to
that
zone
and
then
only
that
zone
kind
of
goes
into
a
initiate.
C
That
it
actually
writes
that
it
actually.
So
that's
why
we
are
the
the
finished.
Latencies
are
a
little
bit
high
for
zones
which
are
partially
filled.
So
if
you
I
mean
we
can
verify
that
if
wait.
C
A
A
I
know,
let's
just
assume
we're
finishing
it
since
you
literally
have
to
in
order
to
keep
using
the
drive
like
in
the
in
the
context
we're
talking
about.
If
you
don't
finish
the
Zone
before
moving
on,
then
you
can't
keep
so,
let's
just
assume
we're
finishing
it.
So
if
you
only
write
half
the
zone
and
then
finish
it,
it
literally
writes
zeros
to
the
rest
of
the
zone.
A
C
We'll
have
to
keep
that
in
mind.
It's
a
bit
weird,
weird
Behavior,
but
at
least
the
the
the
the
the
drives
that
I
have
I
have
confirmed
with
you
know
the
firmware
Team
also,
and
that's
how
it
interesting.
A
A
C
I've
tested
it
and
verified
that
it
is
writing
at
the
end
of
The
Zone,
but
what
I
saw
was
that
at
least
the
initial
zones,
when,
when
we
do
a
mount
from
the
segment
cleaner
code,
it
essentially
closes
a
segment
which
is
almost
empty.
In
the
sense
it
has
written,
only
a
small
amount
of
data,
and
then
it
closes
that
zone,
which
kind
of
you
know,
makes
the
entire
you're.
A
C
Yeah
yeah
yeah,
so
yeah
so
yeah,
currently
I
am
I
was
not
able
to
get
the
mkfs
running
at
all
so
far
it
was
crashing
before
it
could.
You
know,
because
of
writing
the
trying
to
write
the
tail
at
the
end
and
then
trying
to
read
the
text
at
the
end
when
the
title
itself
was
not
written
and
not
a
host
of
these
issues.
But
now,
with
this
all
these
changes,
I'm
able
to
you
know,
run
the
Crimson
OSD
with
the
zns
drive
I'm
using
a
null
block
device.
C
As
of
now
to
you
know,
complete
the
mkfs,
and
you
know
it
doesn't
crash
so
far.
So
that's
where
I'm
currently
at
but
I
think
I
need
to
yeah.
Do
some
core,
cleanup
and
then
I'll
say
and
send
the
changes
as
a
set
of
patches.
A
C
A
A
C
No
currently,
we
do
I
mean
in
the
zns
part,
so
whenever
the
segment
manager
does
this
segment
close
right.
So
in
the
block
part
of
the
code,
it
basically
just
moves
the
segment
information
into
closed
State
and
in
the
zns
implementation
it.
It
goes
and
issues
the
block
close
Zone
ioctl,
which
basically
tells
the
you
know
drive
to
close
the
Zone,
but
that.
C
B
A
A
That
no
no
worries,
that's
how
I'm
finding
reviews
I
need
to
do
these.
A
Okay,
sorry
jinyu.
Maybe
that
is
that
right
welcome
right,
not
sure.
D
D
D
Gc
transaction
is
not
sourced
by
user
behaviors,
so
the
exchange
rate
operations,
franchise
transaction,
don't
satisfy
the
time
locality
principle.
This
extent
should
not
be
added
to
Aerial
case.
So
there
are
three
GC
paths
that
might
influence
our
education,
that
is
a
trim,
Journal
trim
back
Ray
tree
and
the
reclaimed
space
the
aerial
Behavior
might
occur
when
metadata
updates
or
extends
related
in
this
GC
transaction.
Then
the
method
get
extend
and
a
complete
commit,
so
the
GC
transaction
judgment
could
be
added
to
the
above
logic.
A
I
agree
completely:
I
have
the
following
cautions:
it
does
need
to
go
into
the
cache
for
correctness
reasons,
but
it
does
not
need
to
go
into
the
lru
list.
I
believe
you
have
pointed
that
out
correctly.
It's
touch
extent,
I.
Think
I.
Think
you
found
the
right.
The
right
call,
that's
the
right
one
not
to
do.
A
Yeah
I
I
agree.
Yes,
like
I
said
it
needs
to
be.
It
needs
to
remain
in
the
cache
until
the
until
the
transaction
commits,
but
you're
correct.
It
does
not
need
to
go
into
the
lru
list
that
way
when
the
GC
transaction
completes.
If
Unless,
somehow
it's
gained
a
reference
from
somewhere
else,
it'll
drop
out
of
cash,
which
is
what
we
want.
Yep,
that's
right.
D
Okay,
you
open
this
link,
I'm,
not
sure
I
said
it
right.
A
A
Fixed
KBB
tree
optimization,
we
discussed
last
week.
That's.
A
All
right
sounds
good,
well,
welcome,
again,
Juan,
looking
forward
to
seeing
you
in
the
future.
Anyone
have
anything
else,
did
I
miss
anyone.
C
A
For
now
for
ziadas
drive,
if
you
can,
if
you
can
get
it
running
on
vsert,
that's
good
enough.
This
is
going
to
be
a
little
bit
tricky
because
we
don't
have
a
good
way
of
testing,
specifically
the
code
you're
working
on
in
any
scenario
in
in
any
environment
other
than
your
machine
and
the
one
machine
we
have
CNS
Verizon,
so
yeah
any
code.
You
change
outside
of
zns,
though
there
are
a
lot
of
unit
tests,
but
I'd
like
to
encourage
you
to
add,
add
tests
any
bugs.
A
C
Right,
okay,
yeah
so
I
understand
about
that
in
the
unit
tests.
But
I
was
like
more
asking
about
running
some
workloads.
Now,
I
I
have
the
you
know,
voice
I,
think
I
have
the
all
the
MDS
and
all
the
other
demons
up
and
running
and
either
drive
and
then,
if
I
could
write
something
into
that,
then
that
I
could
validate
whether
rights
are
going
properly
and
that
kind
of
thing
positions.
A
Well
again
for
validating
rights
that
would
be
well
running,
one
of
the
like
staff
test,
rados
or
something,
but
if
you
want
to
test
performance
or
workload
of
that
nature,
the
most
popular
strategy
is
to
use
something
like
fio
with
RBD
or
yingsen.
Do
you
want
to?
Can
you
link
the
the
the
performance
test
repository
you
guys
have
I
think
it's
in
CBT,
it's
a
little
tough
to
use,
but.
B
Yeah
so
yeah,
mostly
I,
think
riddles
bench
or
rbdf
file
should
be
good
enough.
Okay,
that's
just
true!
Let
me
see
we
need
to
out.
B
This
can
also
add
some
generate
workloads,
but
basically
it
is
behind
behind
scenes.
It
is
still
the
radius
bench
and
the
file,
but
it
adds
some
ping.
B
A
The
easiest
thing
for
you
to
do
might
be,
if
you
aren't
already
familiar
with
stuff,
would
be
to
start
up
a
perfectly
normal
classic,
OSD
v-start
cluster
and
then
make
sure
you
understand
how
to
create
an
RBD
image
and
mount
it
and
run
fio
against
it.
Once
you
have
that
down,
you
can
then
use
a
chromosome
cluster.
Instead,
that'll
help
you
to
get
used
to
the
tool
chain
before
you
have
to
throw
in
the
trying
to
get
Crimson
to
work.
Part.
B
Thank
you
yeah,
so
anything
you
can
also
reach
me.
C
I
was
asking
some
questions
on
how
to
run
how
to
test
our
right
rights
on
the
zns.
You
know
part
of
it,
so
everyone
was
like
kind
of
giving
me
suggestions.