►
From YouTube: 2022-02-22 Rook Community 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
A
You
know
last
patch
release
was
still
before
the
last
community
meeting,
so
I
don't
think
there's
anything
to
catch
up
on
for
1.7
still
and
interrupt
me
if
I'm
wrong
on
that.
One
though,
but
nothing
planned
and
nothing
that
has
happened
since
the
last
community
meeting.
B
Yep
agreed
there's
when
blaine
was
looking
at
a
a
security.
What
do
you
call
a
security
issue
yesterday,
which
would
affect
going
1.16
right
blade
and
so
there's?
I
guess
if
we
decide
to
update
well
the
next,
so
the
next
build
that
we
release
should
just
pick
up
the
latest
going
with
the
fix.
C
B
A
Wait
were
you
saying
that
you
in
order
to
fix
it
you
would?
We
would
need
to
update
the
version
of
go
that
we
build
with
or
what
was
the
scope
of
what
we
would
have
to
change.
B
C
Effectively,
we
would
want
to
update
the
latest
galang
version
either
one
at
16
or
1.17
both
have
the
fix.
It's
also,
I
think,
pretty
unlikely
that
it
affects
us,
given
that
it
has
to
like.
It
is
a
like
math.big
overflow
issue
that
I,
I
don't
think
any
of
our
crd
types
are
directly
connected
to
math.big,
so
I
it
is
listed
as
like
a
medium
priority
or
a
medium
severity
anyway,
and
I
think
I
think
we're
pretty
unlikely
to
be
affected
by
it.
C
I'm
going
to
add
a
recovery
20
seconds,
I'm.
C
B
A
Oh,
that
sounds
good
thing.
Thanks
for
clarifying
on
that
one
blaine
okay,
then
sweet.
So
then
we
can
talk
about
1.8.
That
is
more
relevant
and
ongoing
right
now,
so
we
did
do
the
patch
release
1.8.5.
A
Since
the
last
community
meeting
on
february
8th,
we
did
the
patch
on
the
10th
and
then
we'll
we're
planning
a
release,
then
for
later
this
week
as
well
for
1.8.6.
So
let's
take
a
look,
then
at
the
board
and
get
an
update
here
on
anything.
That's
blocking
or
you
know,
loose
ends
for
the
upcoming
patch
on
thursday.
B
Yeah,
first
of
all,
with
1.8.5
that
was
only
one
week
after
the
previous
release.
We
we
did
that
one
sooner
because
there
was
an
operator
crash
affecting
some
clusters,
so
we
just
got
out
that
quick,
which
was
aggression
a
regression
from
198.4
anyway.
So
that's
why
we
kind
of
we're
shorter
on
that
side
right
here,
travis
yeah,
that
first
one.
A
B
And
basically,
it
was
affecting,
let's
see,
clusters
that
updated
the
csi
config
map,
I
think,
which
isn't
very
often
or
maybe
it
was
when
that's
right,
mon
failover.
B
A
Okay,
just
back
to
cadence,
nothing
like
that's
particularly
driving
this
release
right
is
there
any.
I
don't
see
anything
in
the
blocking
release
category.
So
there's
nothing
here.
That's
in
progress
that
we
anticipate
you
know
holding
the
release
for
just
just
putting
it
out
the
door
on
thursday
either
way.
That's
right!.
B
A
Cool
all
right
how's,
it
did
for
1.9.
We
were
discussing
last
time
in
the
community
meeting
like
the
time
frame
for
quincy
any.
Is
that
like
becoming
more
clear
about
when
quincy
would
be
available
and
when
we
could
adopt
it
and
and
get
our
1.9
out.
B
A
It's
a
no
no
big
update
there
of,
like
oh
quincy's
delayed
for
a
month
or
anything
like
that.
B
B
But
yeah,
I
think
after
there
is
a
an
rc
candidate
as
well.
I
don't
know:
we've
already
got
the
daily
ci
test
running
against,
so
I
don't
know
that
we
that's
real
urgent
to
get
something
in
sooner
just
for
the
basic
ci
test,
so
we'll
see
maybe
1.1
is
out.
We
should
just
update
the
smoke
suite
to
run
against
quincy
and
master,
and
just
move
to
that.
I
think
that'll
be
good.
A
That
sounds
like
a
good
approach
to
me
to
you
know:
make
that
cut
over
and
start
getting
that
bake
time
and
validation.
When
you
can,
I
mean.
B
A
B
B
Okay,
it's
pretty
lightweight
board,
though
we
need
to
really
put
stuff
in
there
and
move
stuff
start
cleaning
up
1.8.
A
There's
a
ton
of
stuff
right
exactly
in
1.8,
still
like
in
the
you
know,
to
do
column.
That's
not
going
to
happen
there.
So.
A
A
That's
about
it
all
right,
so
we
we
already
talked
about
the
golang
update
around
the
math.big
library.
So
then
blaine
was
just
capturing
that
what
we
said
out
loud
into
the
meeting
agenda
here.
So
thanks
for
doing
that
blaine.
So
so
yeah
multis
csi
integration,
hiccups.
I
don't
I'm
not
super
up
to
speed
on
this.
One
wanna
go
ahead
and
catch
us
up
on
that
plane.
C
Yeah,
I
guess
I
I
can
kind
of
talk
about
this,
a
little
more
generally
as
well.
So
we
we
we've
been
talking
about
this,
I
mean
said
travis
and
I
and
some
people
who
work
on
the
csi
component,
sort
of
specifically
from
a
technical
standpoint,
but
there's
a
sort
of
generalized
problem
that
exists
with
any
kind
of
overlay
network,
whether
it's
the
pod
network
or
malta,.
C
We've
been
kind
of
focused
on
multis,
but
it
is
broadly
applicable
where,
because
we
use
kernel
mounts
for
which
are
which
are
more
performant
for
likes,
fs
and
rpd
for
getting
user
pods
access
to
to
like
stuff
storage,
if
that
csi
container,
that
makes
those
mounts
gets
restarted
like
for
an
update,
for
example,
that's
like
the
big
and
very
common
case.
C
Then
we
lose
the
initial
network
name
space
that
the
mount
was
created
in
and
existing
mounts
no
longer
like
can
use
that
kernel
connection
and
because
the
kernel
is
slow,
we
don't
really
have
a
way
to
like
kill
a
connection
and
restart
it.
C
We
are
our
best
mitigation
today
is
to
have
everyone
use
host
networking
for
the
csi
driver
that
does
still
break
like
the
network,
isolation
that
kubernetes
ideally
would
give
us,
and
we
we
had
a.
C
C
A
Yeah,
I'm
glad
you
recognized
that
blaine
I
was
going
to
say
like
yeah.
It
makes
enough
sense
to
to
me
in
terms
of
like
not,
you
know,
being
involved
in
the
details
and
not
having
a
lot
of
a
good
good
background
training
in
those
areas.
Anyways,
but
yeah
definitely
definitely
useful
to
understand
that
a
little
bit
better
in
the
direction
it's
going.
C
A
B
Here
we
are.
We
architecting
the
multi
solution
every
couple
months
yeah,
but
it's
it
finally
feels
like
it's
starting
to
settle.
So
that's
good.
A
Is
there
what's
the?
Is
there
a
particular
like
maturity
or
stability?
You
know
like
like
an
alpha
or
beta
beta
designation
to
that
particular
support
right
now.
Is
it
still
an
alpha.
A
Yeah
the
multis
aspect
of
the
csi,
you
know
driver.
A
A
As
a
whole
has
just
progressed
and
matured,
and
I
you
know
that
one's
fairly
stable
but
like
multicell,
I
was
wondering,
if
that's
you
know,
has
a
different
designation.
C
Before
I
think
it
can
be
like
really
truly
supported,
so
I
think
we're
all
sub
no
is
definitely
more
about,
like
multis
as
a
whole
yeah.
So
I
think
they're
just
things
we're
waiting,
even
for
like
the
maltest
kind
of
community
itself
to
to
kind
of
fill
in
some
gaps
as
well.
A
Cool
all
right
that
sounds
good
anything
else
to
add
on
this
topic
here
before
we
move
on.
A
B
Yep,
so
it's
all
deprecated
yesterday,
I
was
just
realizing
that
the
main
website
of
website
is
still
referred
to
cassandra
as
one
of
the
main
operators,
so
I
cleaned
that
up
removed
it
from
there.
The
cassandra
docks
are
still
there,
of
course,
so
that
they
just
have
the
deportation
notice
and
they
will
remain
indefinitely,
but
the
the
last
thing
I
think
action
item
is
to
go
ahead
and
archive
the
repo.
I
think
we're
ready
to
do
that.
Is
there
any
reason
not
to
do
that
now.
Anybody.
A
I
don't
think
so
because
so
links
when
a
repo
is
archived.
If
somebody
has
like
a
like
this
documentation
or
a
blog
post
or
whatever
a
link
to
that
repo
people
click
it
they're
still
taken
to
that
repo
right
and
it's
just
like
the
at
the
top
there's
some
type
of
banner.
That
says
this
repo
has
been
archived
but
you're
still,
it
doesn't
like
result
in
a
404
or
not
found
type
of
thing
right
right.
That's.
A
A
Yeah
that
sounds
reasonable
to
me
that
you
know
it's
not
like
it's
completely
wiped
off
the
face
of
the
earth
and
people
wouldn't
even
know
if
it's
deprecated
or
whatever,
because
the
notice
isn't
even
there
anymore,
but
you
know
not
not
having
it
actively.
You
know
active
contributions
and
collaboration
around
the
repository
is
is
a
fine
result
of
that.
B
A
A
B
A
It's
all
right
sounds
good,
so
cesaro's,
not
here
I
yeah.
I
I
put
that.
B
Some
work
he
did,
I
didn't
want
to
take
credit
for
it,
but
because
he
did
it
all
so
there's
this
osd
raw
mode
which
we
had
enabled
last
summer.
But
then
we
had
all
these
issues
with
atari
partitions
and
blaine
worked
through
that
pane
and
but
now,
with
the
latest
step
release,
we
can
go
ahead
and
enable
raw
mode
and
the
atari
partition
is
user.
B
Fixed
and
it's
been
fully
verified
by
certain
looks,
looks
like
we're
good,
so
only
newer
set
or
the
latest
separately
1627
will
be
enabled
for
raw
mode,
but
I
think
ready
and
we're
we're
backporting.
It
had
some
merge
complex,
so
terra
was
going
to
merge
or
to
resolve,
but
I
think
we're
good
to
go
and
just
get
into
the
next
patch
release.
A
A
Okay
and
then
this
last
issue
here
travis
around
encryption
on
the
wire.
B
Yep,
just
another
fyi,
really
the
so
messenger
2
is,
you
know
the
set
protocol
has
been
around,
for
I
don't
know
two
or
three
years
already
and
it
it
supports
encryption
on
the
wire,
no,
since
at
least
specific,
but
it
does
require
a
new
kernel.
So
I've
had
this
pr
open
to
get
it
working
with
rook
for
since
november
and
finally
realized.
Oh,
it's
just
my
kernel
version.
B
That's
why
it
won't
work
for
me,
and
so
I
was
working
with
radic
from
the
stuff
team
on
why
it
was
failing
and
there's
a
workaround
to
use
the
rbd
mbd
or
fuse
drivers
with
the
csi
driver,
but
those
those
things
still
suffer
from
csi
padres
starts
causing
an
unmount
of
the
volume
or
detachment
or
whatever.
So
it's
just
not
really
feasible
to
use
it
in
production.
B
So
I'm
going
to
go
ahead
and
merge
the
feature
and
just
point
out:
you
need
the
newest
kernel
and
hopefully
someday
well.
Madhu
from
csi
team
was
saying
that
rbd
mbd
is
getting
close
to
a
solution,
so
it
won't
have
that
issue
with
the
csi
restart
during
upgrades
and
all
that
so
hoping
that
will
be
the
case
soon.
A
Will
this
be
when
you
merge
this
will
will
when
you
merge
this,
will
this
option
be
enabled
by
default
no
it'll
be.
A
B
A
All
right
that
makes
sense,
then
thanks
for
updating
us
on
that
one
travis,
okay,
so
any
other
community
topics
then
to
bring
up
today.
A
I'm
going
to
go
ahead
and
do
this
action
item
right
now,
travis
and
the
the
30
minutes
we
have
left
in
this
hour
here,
I'm
just
going
to
do
it
right
now,
so
that
we
can
stop
stop
having
an
action
item
from
jared
that
goes
across
meetings
right
all
right.