►
From YouTube: IETF-NFSV4-20230711-1900
Description
NFSV4 meeting session at IETF
2023/07/11 1900
https://datatracker.ietf.org/meeting//proceedings/
B
D
C
Sorry
well,
Dave,
there's
a
budget.
So
if
you
do
an
interim
schedule,
one
way
it
does
all
of
the
magic
behind
the
scenes
and
sends
it
out
and
if
you
do
it
from
another
way,
that's
also
recommended
on
the
website.
It's
broken
and
I
just
filed
that
bug
so
I'm
populating
the
materials
kind
of
as
we
speak.
C
F
D
C
D
D
D
C
F
C
C
C
A
B
A
D
A
E
C
D
C
A
C
Promise
so
we
restarted
the
all
right.
Brian's
I
can
stop
covering
his
chair
as
soon
as
he
he's
good
to
go.
So
I
will
cover
the
last
two
because
they're
super
easy
right.
So
the
last
call
on
Parallel
NFS
enough
response.
It
was
good
in
the
room
so
we'll
be
moving
that
forward
and
out
of
the
working
group
and
the
same
thing
with
the
RDMA.
C
Yesterday,
David
black,
this
is
really
just
for
you
got
a
new
revision
out
on
the
lease
list.
In
response
to
the
previous
last
call,
which
we
think
answers
everything
so
we're
restarting
another
working
group
last
call
for
it
two
weeks,
so
we
hope
by
the
time
we
get
to
San
Francisco.
We
can
say
that
that
one's
moving
forward,
that's
the
update
on
documents,
unless
there
are
questions
I
feel
like
I,
should
just
write
the
minutes
and
be
done
with
this
meeting
in
a
way.
A
F
A
I
see
it
all
right,
so
a
central
part
of
of
the
RFC
5661.
This
effort
is
this
document.
Rfc
5661
this
so
on
to
the
next
I.
Think
I
think
we
move
on
to
the
next
slide.
Okay
overview
is:
why
are
we
doing
this?
What's
in
it
and
some
attempt
to
get
discussion
started,
but
we
only
have
one
person
here
but
anyway,
any
discussion
that
you
have
David
will
be
appreciated.
But
let's,
let's
go
forward.
A
B
B
A
So
what
what
we
have
beyond
that?
We
would
have
had
a
ton
of
variety
of
reports
and
I
think
we've
got
most
of
those.
There
are
a
few
left
anyway.
We
also
want
to
have
a
solid
document
that
describes
NFC
4.1.
So
we
don't
have
we
had
a
number
of
additional
documents.
We
had
a
a
with
5661,
we
had
80
81,
and
then
we
also
had
Tom
haines's
document
8434
and
that
can
no
longer
be
a
separate
time.
A
A
A
A
A
Okay,
as
a
result
of
our
discussion
we
now
have
lidocs
internationalization
06.
is
is,
is
a
working
group
document
has
been
for
a
while.
We
don't
there's
not
much
discussion
on
the
list,
but
we're
not
having
an
art
art
review,
which
we
had
some
trouble
doing,
but
I've
been,
but
the
target
date
for
that
is
now
the
14th
and
we'll
see
what
happens
now.
A
In
fact,
one
part
of
that
is
that,
while
we're
doing
these
adoption
calls
Cycles
document
expired
and
I
think
I
have
to
at
least
within
the
next
week,
repost
the
the
current
individual
draft
and
then
we'll
use
that
as
a
basis
to
decide
on
what
we'll
do
about
an
adoption
call
another
document.
We
have
okay.
A
B
A
Been
a
big
issue:
it
I,
don't
I
I,
think
that
will
just
go
through
once
we
do
the
appropriate
scheduling,
because
it's
like
it's
like
10
lines
different
and
then
we
have
RFC
8178.
That
is
already
an
RFC,
but
we
need
our
new
document
to
clearly
indicate
that
is
the
definitive
definition
of
extension
for
all
minor
versions
and
RFC
8434
I've
looked
at
that
and
I
think
we
have
to
figure
out
over
the
next
few
drafts.
A
Now
in
the
process
of
doing
this,
I
had
to
have
the
task
of
reading
documents
and
they
actually
found
some
some
things
that
had
to
be
done.
Lots
of
uses
of
RFC
2119
terms,
not
in
court
with
the
definition,
despite
the
fact
that
every
document
has
this
thing.
That
says
it
does
that,
but
very
few
actually
do
it
and
there's
a
lot
of
confusion,
confused
terminology
to
clean
up
I
found
out
in
this
last
draft
use.
Client,
there's
one
place
where
it
uses
serum
client
under
to
include
the
verifier
one
time
it
doesn't
also.
A
There's
the
definition
of
the
word
verified
describes,
one
of
like
six
different
things
that
are
different
and
that's
been
changed
and
fixed
up.
There
are
a
number
of
we
Implement
director.
We
we
Define
directory
delegation
and
no
one
in
in
implemented
it,
and
we
had
at
a
previous
meeting,
we
discussed
G
what's
wrong.
Why
are
we
so
so
slow
handling
and
I?
A
Think
part
of
the
answer
was
that
nobody
had
implemented
directory,
delegate
direct
delegation
and
notification,
and
that
has
to
be
written
people
who
have
have
like
including
Tron
and
say
well,
it
doesn't
work
because
it
doesn't
do
this.
That
and
the
other
thing
and
I
have
to
be
clear
of
what
it
does
do
and
what
it
does
do,
and
that
is
not
that
is
discussed
in
the
appendix.
But
it's
not
there's
no
text
for
that
and
I
think
that
might
be
something
I'll
do
in
the
O2.
A
A
Diff
with
between
RFC,
881
and
01
is
kind
of
bad
because
it's
artifacts
in
the
sections
move
around,
but
but
but
you
do,
that
is
explained
independent.
All
those
changes
are
explained
dependencies
B1
through
B3,
and
that
is
helpful
and
then
once
you
get
to
that
point,
it's
pretty
easy
to
compare
to
Darcy
diff
of
the
dash
zero
draft
and
the
oh
one
draft
all.
A
A
A
Protocol
extension
been
in
its
own
document,
which
is
already
done
and
Security
Buildings
on
that
we
had
to
address
a
lot
of
variety
of
reports
like
tons
of
them.
There
are
about
half
a
dozen
left
and
we
got
around
the
thing
where
way
back
when
someone
decided
was
that
the
optional
actors
were
recommended
and
that
doesn't
fit
and
in
fact
RFC
7530
says
well.
This
document
uses
during
recommended,
except
as
it's
defined
in
RFC
2119,
except
for
this
case
and
I.
Don't
think
we
want
to
do
that
anymore.
A
A
E
A
A
D
E
A
A
So
really
the
really
meaning
of
must
in
in
RFC
881
is.
It
would
be
really
nice
if
things
worked
out
work
this
way,
but
they
don't
work
this
way
they
can't
work.
This
way,
no
implementations
do
it
so
that
can't
go
on
all
right.
Another
thing
is
the
discussion
of
end
of
of
of
of
reach
of
of
retry,
and
because
we
had
exactly
one
semantics,
retry
is
unnecessary,
but
there's
no
reason
and
it's
not
a
good
idea,
but
they
have
must
not
in
in
the
existing
documents,
but.
A
D
A
Scribes
client
owner
in
one
case
would
use
uses
it
to
meet
the
the
string
plus
the
Vera
IFR
and
one
thing
in
one
place.
It
uses
to
mean
just
the
string
so
I've
distinguished
recline
owner,
which
includes
both
and
client
owner
ID,
which
is
the
string
and
I
think
there
are
about
500
verifiers
there's
the
the
right
verifier
is
the
reader.
Verifier
is
the
verifier
on
on
client
instance.
A
It's
called
verified,
there's
a
whole
bunch
of
them
and
there's
one
definitely
in
RFC.
881
is
one
definition
and
it's
just
confusing
all
right.
I
found
the
pnfs
terminology
need
to
be
thing
because
you
know
we
have
these
in
some
places
refers
to
storage
protocol,
a
data
Pro
protocol
and
in
some
cases,
data
server,
which
is
appropriate
for
file
servers
and
storage
device,
which
is
appropriate
for
for
Block
device.
A
So
I
created
the
generic
term
file
data
provider
and
do
it
and
data
protocol
rather
than
storage
protocol
and
the
discussion
of
control
protocol,
because
it's
in
in
in
Tom's
flexible
file.
He
says
he
doesn't
have
a
control
protocol,
but
the
text
in
the
current
RFC
881
says
every
yeah.
Anything
you
do
to
accomplish.
The
need
is,
can
is
control,
considered
control
protocol,
so
I
now
have
clarified
that
so
distinguish
between
those
that
don't
have
a
separate
control
protocol.
That.
A
A
A
Now
I've
tried
to
read
it
reorganize
the
discussion
of
pnfs
security,
but
I'm
not
done
what
is
needed
is
the
thread,
analysis
and
dividing
pnfs
layer
types
into
three.
One
is
the
the
one
the
files
layout
and
the
flexible
files
layout
in
what
he
call
in
what
Tom
calls
the
tide,
coupling
mode
that
one?
You
basically
have
enough
material,
so
so
you're
not
requiring
the
client.
A
A
All
right
so
I
now
have
to
consider
what's
in
draft
two,
so
I
think
I
have
two
suggestions
from
David
they're
clarifications,
and
then
there
are
two
items
in
appendix
C2:
a
clarifying
director
delegation
and
need
to
work
required
persistence.
Now,
if
you
read
the
description
of
persist
of
persistent
sessions,
you'll
find
I
think
that
what
this,
what
is
described
is
simply
depending
on
how
you
interpret
it,
but
that
needs
to
clarify
what
it
really
requires
and
what
it
doesn't
require.
A
A
Though
it
doesn't
use
those
words,
but
if
you
tried
to
do
it,
you
couldn't
do
it
so
I
think
I
need
to
make
two
things
clear.
One
is
what
is
the
real
animistic
requirement
and
the
second
thing
is:
there's
some
discussion
of
a
server
restarting
in
the
middle
of
compound
and
how
the
second
instance
will,
in
instance,
of
the
of
the
of
the
server
would
complete
the
compound
which
doesn't
seem
to
me
how
I
don't
see
how
anyone
could
do
that,
and
it
doesn't
give
you
an
option
to
do
anything
else.
F
C
D
A
Well,
I
think
I'm
going
to
draft
what
I
can
draft
and
then
Reach
Out,
hopefully
out
from
the
security
here,
but
I,
don't
I,
don't
think
we're
in
a
position
to
say.
Okay,
we
have
to
write
a
thread
analysis.
Please
do
it
they're
not
going
to
do
that
one
through
that,
it's
not
right
to
ask
them
so
I'll
have
to
draft
something
which
I'll
probably
will
take
probably
a
month
after
IDF
117.
But
after
that
I
think
we
will
be
interested
in
both
review
comments
or
help.
A
Okay,
okay,
well,
okay,
so
get
back
to
the
current
slide.
Well,
we
do
have
a
few
remaining
Rod
of
reports
and
we
have
to
do
those
and
I've
been
putting
them
off,
but
I
have
to
do
it
and
back
to
the
lowercase
recommended
version
versus
lowercase
recommended.
It
doesn't
seem
like
that's
right,
although
officially,
according
to
the
current
spec
every
option,
every
attribute
which
is
not
mad
terrorist
is
lowercase
recommended,
although
some
of
them
it's
recommended,
and
you
really
have
to
do
it.
A
Slash
lowercase
recommended
attributes,
whereas
the
same
is
true
of
some
new
attribute
that
is
putting
but
I
I,
think
there's
a
there's
levels
of
recommendation
and
I
think
we'll
have
to
be
clear
on
what
we
really
recommend
and
and
get
away
from
that
formula
which
says.
Well,
we
recommend
you
do
all
the
actions
that
you
you
think
you
can
do,
which
doesn't
mean
anything
but
I
think
there
are
a
few
that
are
really
essentially
optional.
E
Well,
hang
on
a
minute
so
from
an
abstract,
plain
point
of
reference
and
I'm
going
to
put
that
in
put
that
in
here
as
the
disclaimer,
because
we're
going
to
get
lost
fast
if
situations
as
you
describe
it,
which
is
that
Mo
things
like
mode
and
group
will
cause
clients
not
to
work
with
the
server.
Those
aren't
just
recommended.
Those
are
required
in
that.
If
you
don't
do
it,
you
know
our
ability
isn't
going
to
happen.
E
A
C
A
A
A
All
right
things
mentioned
no
one,
but
not
yet
done
director
delegation
I,
think
I'm
gonna
start
drafting
that
persistence
issues.
I
mentioned
the
Persistence
of
also
there's
some
ambiguity
about
whether
when
a
server
restarts,
he
can
have
some
way
to
communicate,
to
keep
locks
persistent
and
avoid
the
grace
period
and
I.
Think
that
might
be
controversial,
but
I
think
my
what
I'm
going
to
do
is
is
include.
D
A
A
D
A
B
A
C
C
C
B
C
Was
hoping
that
there'd
be
more
people
who
want
to
bash
this?
We
should
be
able
to
say
you
know
that
we're
shepherding
some
of
these
documents
as
we
get
there.
I
don't
know
if
there
will
be
any
final
conversation
on
bells
did
at
the
meeting
the
layout
I
think
should
be.
You
know
that
should
be.
We
should
be
able
to
say
where
we
are
outside
of
working
group
at
that
point.
C
Rdma
extensions,
so
hopefully
we'll
get
something
from
talpy
we'll
do
a
Reach,
Out
I,
think
I
think
we're
gonna
have
to
talk
about
Security
in
more
broad
terms
and
our
approach
to
it.
I
appreciate
Dave's
willingness
to
put
a
big
bullseye
on
his
back
and
attempt
to
write
the
threat,
analysis
and,
and
the
outcome
of
that
I
think
we
should
have
a
broader
conversation
about
how
we
want
to
approach
that
and
what
it
might
mean
and
how
we
think
about
kind
of
protocol
structure
for
that
topic.
Going
forward.
C
I
think
that
should
be
a
conversation
there
internationalization
efforts.
Hopefully
we
have
some
interaction.
Well,
so
hopefully
we
have
since
I
guess
technical
interaction
to
to
discuss
at
117.
With
regards
to
that
and
John,
if
you
want
to
add
anything
on
that
Feel
Free
at
this
point,
and
then
I
mean
it,
we
obviously
participation
for
this
interim
is
is
really.
B
C
I
will
admit:
I
I
found
the
bug
in
in
ietf
web
space.
I
have
reported
it
earlier
this
afternoon,
which
is
also
how
I
managed
to
get
a
interim
meeting
approved
and
announced
in
like
10
minutes
before
an
actual
interim
meeting,
but
I
mean
if,
if
this
is
because
the
main
topic
of
conversation
was
5661,
this
we're
gonna
have
a
problem,
because
we
will.