►
From YouTube: SIG - Storage 2023-08-14
Description
Meeting Notes:
https://docs.google.com/document/d/1mqJMjzT1biCpImEvi76DCMZxv-DwxGYLiPRLcR6CWpE/edit#
A
A
B
So
can
you
hear
me
yeah
cool
so.
A
B
Introduced
the
wait
for
first
consumer
Lane
a
couple
of
weeks
ago
and
I
started
fixing
the
tests
up,
so
we
don't
get
any
failures,
so
the
lane
can
be
a
required
one
and
during
testing
I
noticed
that
the
runtime
for
this
new
Lane
is
significantly
longer
than
that,
just
regular
cephlane.
B
B
I
didn't
have
ideas
about
what
could
cause
it
like
running
locally.
It
seemed
pretty
much
the
same.
Maybe
there's
some
some
ideas
in
this
form,
but
anyway,
that's
the
gist
of
it
is
that
the
runtime
was
really
longer
than
than
the
immediate
mining
line.
A
Yeah
I
feel
like
this
underscores
the
desire
to
be
able
to
kind
of
profile
our
components
to
see
where
they're
spending
the
time
we
kind
of
have
I
guess
Michael's
working
on
another
performance
related
issue,
so
might
be
useful
there
too,
as
well.
Anyone
else
have
any
thoughts
or
ideas
things
to
try.
A
All
right,
so
it
seems,
like
we've,
got
a
a
good
one.
Here,
it's
probably
going
to
take
a
little
time
for
us
to
dig
into
over
time
oops
When
I
close
that.
A
Let's
pop
back
up
and
go
to
the
next
one,
so
DB,
smart
clone
is
stuck
when
a
pot
is
using
the
old
PVC.
A
A
B
C
C
C
With
this
with
our
cfrbd,
it's
you
know
not
as
bad
as
with
the
dumb
clone,
but
yeah
I
mean
maybe
with
if
we're
doing
a
smart
clone.
We
can
allow
the
source
to
be
in
use,
but.
A
Okay,
I
see
Alexander
jumped
in
and
mentioned
that
we
won't
do
the
Clone,
but
he
could
use
a
snapshot.
A
A
A
C
Yeah,
it's
true
that
there
are
no
guarantees,
but
also,
if
you
want
to
be
real,
like
you
know,
it's
totally
possible
that
we
check
that
the
source
isn't
in
use,
start
the
operation
and
then
it
becomes
in
use.
Then
you
know
so
yeah
they're
really!
No,
even
even
though
we're
attempting
to
do
something.
The
fact
is,
we
really
can't
keep
okay
guarantee
any
consistency.
Yep.
C
A
I'm
just
going
to
do
this
for
now
and
I,
don't
think,
there's
much
else
that
we
can
do
all
right.
Let's
go
up
to
500
error
during
DV
import
on
a
gcp
environment.
B
So
there's
two
very
similar
issues.
One
of
them
is
related
to
to
describe
change
from
a
couple
of
weeks
ago.
Last
meeting
I
believe
this
is
the
one
and
I'm
not
sure
on
the
CDI
side.
If
we
can
do
much
more,
we
we
have
talks
about
about
using
block
PVC,
that's
not
root,
and
we
have
communicated
this
of
the
issue.
B
So
I
think
this
is
just
about
the
reporter
still
not
getting
a
new
openshift
build
or
kubernetes
whatever
they're
using.
A
Is
there
something
we
could
do
to
at
least
clearly
articulate
what
the
problem
is
and
then
like
link
to
information
about
it
like
I,
think
well,
I,
don't
know
is
an
error.
Five.
Maybe
error.
500
is
appropriate
in
this
case
because
it's
like
legitimately
an
error
in
the
server
instead
of
I.
Just
wonder.
B
Yeah
upload
code
has
has
the
permission
denied
errors
in
there?
Okay,
so
that's
how
we
derived
because.
B
A
To
keep
this
open
until
we
feel
like
until
we
feel,
like
things,
have
merged.
B
Yeah
I
think
so
I
think
we
can
close
it,
but
no
I
don't
want
to
discourage
the
issue.
Author.
A
A
A
B
No,
this
is
a
different
one
and
there's
few
chats
about
this,
not
sure.
What's
the
cause
here,
but
essentially
they're,
just
importing
from
registry
and
just
to
get
an
unexpected
file,
there
writing
the
data,
so
I've
tried
a
couple
of
suggestions
for
one.
Their
image
was
looking
real,
weird
to
me,
like
usually
these
container
disk
images
they're
just
a
directory
called
disk
and
an
image
file
inside
of
it.
But
in
this
case
it
was
a
lot
of
garbage
in
there.
Okay.
C
A
B
Yeah,
if
it's
created
the
same
way
as
the
docker
file
here,
there
should
not
be
a
boot
folder.
A
We
should
just
see
if
they,
if
they
encounter
this
issue
with
one
of
our
other
container
disks
yeah
like
the
official
ones
and
I,
don't
have
the
specific
link
to
that.
But
maybe
if
someone
can
help
me
find
that
so
I
can
add.
It.
B
This
is
the
official
ones
we
maintain
under
the
cube.org.
A
A
Okay,
I
think
maybe,
with
the
sharing
screen
it's
compacted
or
something
yeah,
it
is
possible.
B
A
Okay,
all
right
yeah
if
it
works
with
the
other
ones,
then
there's
probably
some
sort
of
strange
issue
like
we
don't
know
if
they're
using
like
which,
which
tool
they're
using
like
to
build
the
images
if
they're,
using
builda
or
Docker
or
anything
either
so,
okay.
A
C
Yeah,
this
was
something
that
came
up
as
he
was
adding
support
to
the
PVC
Source
I.
Don't
think
it's
a
big
deal
now,
because
s
aren't
doing
snapshot
sources
by
default.
I,
don't
think.
A
They
sh
well
and
likes.
Eventually
they
will
be
right.
C
A
C
A
Okay
sounds
good
all
right,
so
I'm
gonna
pop
back
out
of
here
we've
gone
through
all
the
issues.
I
guess
this
is
the
last
call
to
see
if
anyone
has
any
agenda
items
or
open
floor
topics.