►
From YouTube: Velero Community Meeting/Open Discussion - Feb 11, 2020
Description
Feb 11, 2020
Agenda
Status Updates
---------------
[steve]
shipped v1.3.0-beta.1, working on blog post
put up draft PR with refactor of restore code related to CRDs
some testing
[carlisia]
community support last week
updated CLI revamp PR: https://github.com/vmware-tanzu/velero/pull/2202#issuecomment-577229282
POC for installing with kustomize
[nolan]
On community support this week
Moved last week, was out
Back to working on CRDs, troubleshooting #2249. Will go more in depth in discussion topics
Need to work with Ashish on CSI support
[ashish]
Update on memory leak #2069
Fixed with Steve’s change and v1.3.0-beta.1 has the fix. Please consider upgrading.
CSI update
Using CSI hostpath-plugin driver to validate Velero
Successfully setup an environment with hostpath-plugin driver using beta snapshot APIs
Blogpost?
[jonas]
Implemented an ADOPTERS.md file for public user statements
Working on a /plugins page
Discussion Topics
------------------
[carlisia] CRD/api/k8s versioning issues: https://github.com/vmware-tanzu/velero/labels/versioning
Visual summary of #2249
[dymurray] Add design doc for Custom CA support for S3 BSLs
Contributor Shoutouts
----------------------
@markrity - docs update to point to new Helm chart repo (https://github.com/vmware-tanzu/velero/pull/2209)
@jnaulty - add Restore API docs (https://github.com/vmware-tanzu/velero/pull/2240)
@mynktl - allow VolumeSnapshotter plugin to rename PV via SetVolumeID function (https://github.com/vmware-tanzu/velero/pull/2216)
@jenting - ongoing beta testing, bug-fixing
@betta1 - Thank you for providing information to narrow down memory leak
A
Hi
everyone
and
welcome
to
this
week's
episode
of
the
Valero
commuter
meeting,
slash
open
discussion.
We
get
a
bunch
of
things
are
usual
that
we
want
to
run
through.
If
you
have
any
thing
well,
any
items
that
you
would
like
to
add
to
the
meeting
notes
I'm
going
to
share
the
link
to
the
meeting
notes
in
the
chat
as
well
and
with
that
I'm
going
to
share
the
agenda
and
then
kick
it
off
with
some
status
updates.
B
B
At
the
beginning
of
this
week,
yesterday,
I
was
working
on
putting
up
a
draft
of
the
blog
post
for
the
1.3
release,
so
I'm,
just
working
on
refining
that,
let's
see
last
week,
I
also
worked
on
doing
some
refactoring
and
some
of
our
restore
code
to
make
it
a
little
bit
cleaner
to
be
able
to
restore
CR
DS,
followed
by
C
ours,
including
a
refresh
of
the
discovery,
API
information
that
falero
caches
I
think
I
mentioned
that
I
was
going
to
work
on
this
last
week.
So
I
have
a
draft
PR
up.
B
That
seems
to
be
working
as
expected
working
properly.
So
it's
it's
not
an
urgent
thing:
I'm
not
intending
to
get
it
put
in
to
1.3,
but
whenever,
whenever
folks
have
a
chance
to
take
a
look
at
it,
feedback
would
be
welcome
and
then
it
was
just
doing
a
little
bit
of
that
hoc
testing
of
the
the
beta
things
looked
looked
good
to
me,
based
on
the
scenarios,
I
was
tested,
I
put
up
one
one.
B
C
I
was
on
community
support
last
week
and
I
updated,
CLI,
revamped
PR,
basically,
I
reorganized
the
commands
based
on
feedback
and
also
based
on
more
insights.
There
are
others
too.
After
rethinking
the
whole
thing,
some
of
the
backup,
location
and
snapshot
location
commands
actually
are
not
going
to
change
much
with
minor
exceptions
and
in
the
mainly
mainly
the
the
change
there
is
that
those
are
not
going
to
change
the
little
config
that
I
had
originally
I
had
to
remove
that
and
I
explained
it
in
the
tickets.
C
I
am
putting
together
some
yellows
to
see
if
I
can
run
with
the
customized
commands,
so
hopefully
you'll
get
that
working,
I'm
thinking,
I'm
going
to
run
into
trouble
and
I
figured
it
out,
but
that's
what
I'm
doing
now.
If
anybody
he's
an
expert
in
the
customizing
wants
to
to
look
at
my
proof
of
concept
today,
just
being
my
place
and
that's
a.
C
Alright,
so
the
thing
to
pay
attention
is
so
I
I
reorganized
the
commands
and
another
big
thing
that
I
reorganized.
That's
new
is
I,
moved
everything
from
the
server
to
the
villa
in
needs,
because
I
thought
those
to
me.
They
seems
to
belong
in
the
same
place.
So
if
you
look
at
that-
and
you
think
not
really
it
in
a
rethink
things,
otherwise,
I
am
so
I'm
going
to
work
on
this
proof
of
concepts
for
customized
based
on.
What's
there
now,
but
okay
I
think
he's
going
in
the
right
direction.
Okay,.
D
D
Troubleshooting
number
two,
two:
four:
nine,
which
I'll
go
more
into
depth
on
and
discussion
topics.
I,
don't
want
to
take
the
whole
update
session
to
talk
about
that,
but
basically
that
with
some
of
the
changes
we
made
that
went
into
the
one
point
three
point:
zero:
the
kind
of
surfaced,
some
issues
with
the
solution
we
had
implemented
for
the
beta,
so
I'll
talk
about
what
the
the
problem
is
and
I
think
I
need
to
do
some
more
research
today
to
make
sure
I'm
on
the
right
path.
A
E
E
Other
thing
is
an
update
on
the
CSI
stuff.
Yesterday,
I
was
able
to
set
up
an
environment
using
the
meter
snapshot
if
CSI
environment
to
use
the
beta
snapshot,
fbi's
and
I
was
able
to
perform
like
snapshots
and
restores
so
I.
Have
that
going
I
wasn't
sure
I
know
we
spoke
about
having
a
blog
post
about
that
I'd
I
wasn't
sure
if
the
blog
post
was
once
we
have
it's
working
with
velaro,
or
was
it
just
how
to
use
CSI
how
to
use
the
beta
IB
is.
E
Firstly,
si
snapshot
so,
based
on
what
we
decide,
maybe
I
will
put
together
a
graph
for
a
clock
first
and
like
circulate
internally
and
then
kind
of
continue
working
on
the
CSI
stuff
and
see
how
if
our
CSI
plug-in
works
as
is
or
whatever
changes
we
need
to
make,
we
will
I
will
have
to
start
working
on
that.
So
that's
that's
all
the
updates
coming.
A
B
A
All
right
so
quick,
one
from
me,
Association
I'm,
going
to
yeah
perfect.
Thank
you,
quick
one
from
me.
We've
implemented
and
a
doctor's
MD
file
for
public
user
statements.
So
we
have
a
few
in
here.
Some
of
you
are
on
its
call
as
well.
So
thank
you
so
much
for
for
adding
your
information
in
here.
So
we've
got
statements
from
bit
cone
or
mana
cloud
kheema
project,
Red,
Hat,
Dell,
EMC,
bug,
snug
and
Banzai
cloud.
A
If
you
are
an
adopter
of
velaro-
and
you
want
to
add
yourself
in
here
as
a
success
story
or
maybe
a
solution
built
with
velaro,
please
do
so
please
out
of
PR
I
know
we
have
some
updates
coming
in
for
this
adopters
file
as
well
from
Michael
and
kaliesha,
so
just
small
language
updates,
mostly
so
yeah.
Please
add
yourself
there.
If
you
are
a
bolero
adopter
and
the
other
thing
that
I'm
working
on
is
a
plugins
page
for
Valero.
A
D
A
A
F
A
D
C
Looks
awesome
by
the
way
I
hadn't
seen
that
page
I
said
that
PR,
but
I
hadn't
seen
it
fit
together,
like
that,
I
was
actually
hoping
that
we
would
only
keep
one
of
those
and
I
preferred
this
one.
So
is
because
and
I'm
going
to
also
mention
this
on
my
PR.
We
did
talk
about
this
before
we.
We
have
this
concept
of
plugins
and
also
the
concepts
concept
of
provider
and
can
be
confusing.
C
C
It
makes
more
sense
and
another
one
I
think
no
Lee,
no
I
still
agree
with
this
concept,
because
he
doesn't
make
a
comment
on
the
when
I
you
wax
CLI,
you
wax
epic
before,
if,
if
so
just
trying
to
make
people
remember
nobody
talked
about
this
before
so
right
now
we
had
a
lot
of
things
as
I
go.
What
provider
do
you
need?
So
if
you
need
this
provider
go
get
this
plugin
and
that's
how
that
that
original
documentation
is
laid
out.
I
think
it
should
be
the
opposite.
C
It
should
be.
Okay,
here
are
all
the
plugins
for
whatever
provider
so
just
pick
which
plugin
you
need.
Instead
of
you
know
what
I
mean.
So,
instead
of
leading
with
provider
deal
with
plug-in,
there
is
a
flag
on
the
OE
style.
Cli,
that's
named
provider,
I'm,
going
to
suggest
that
we
rename
that
to
plug-in,
and
then
I
take
this
ties
in
with
this
modest
change
of
May
that
they
could
find
a
way
I
mean
making
this
page.
C
If
everybody
approves
a
list
of
that,
this
change
from
from
provider
to
again
this
page
will
have
to
be
reformatted
and
I.
Think
maybe
this
is
something
we
can
talk
about,
can
think
about,
and
maybe
reformulates
in
integrate
tears
into
the
new
page
you
have.
So
when
I
was
asking
you
that
question
I
was
hoping
you
would
say.
Yes,.
C
But
if
and
if
we
free,
don't
I
think
it's,
it
might
be
confusing
to
have
this
page
in
in
their
other
page.
So
I
do
have
this
opinion
that
no
matter
why
it
could
be
confusing,
because
what
is
the
difference,
if
I,
if
I'm
new
to
the
ER
and
I'm,
trying
to
understand
what
the
components
are
and
I
see
these
two
things
they
seem
similar.
So
like
I'll,
be
thinking.
Why
do
you
have
these
two
things?
If
the
things
that
seem
to
be
different,
but
they
seem
the
same
anyway?
That's
my
my
many
cents.
D
Strictly
in
terms
of
the
the
docs
and
the
website,
I
think
it's
duplicating
the
information,
that's
a
little
confusing
because,
like
the
in
terms
of
the
website
structure,
this
lives
in
the
documentation
section
and
so
far,
we've
kind
of
kept
the
documentation
segment.
It's
separate
from
the
higher
higher
level
website
and
if
we
like,
merge
that
content
and
move
it
up
to
the
plugins
part
we're
starting
to
blur
that
line
kind
of
without
intentionally
planning
for
it.
D
D
D
A
Yeah
definitely
so
the
the
plugins
page
was
similar
to
the
doctor's
MD,
an
easy
way
for
plugin
creators
to
add
themselves
without
having
to
go
through
a
bunch
of
documentation,
changes
and
things
like
that.
So
that's
the
genesis
of
it
is
actually
one
thing
that
Michael
talked
about
last
week
in
a
meeting
was
to
add
in
some
sort
of
badges
to
the
to
the
little
squares
here.
So
you
would
have
the
supports
object,
store.
A
G
And
the
basis
for
that
was,
some
providers
have
different
capabilities
and
others
right
so,
for
example,
local
storage.
You
know
like
port
works,
that
supports
it
unless
you
actually
go
through
every
provider,
one
by
one
and
read
through
that,
it's
hard
for
you
to
figure
out
what
each
one
of
them
supports
so
for
the
customers
are
coming,
say:
hey
I
care
about
AWS.
How
do
you
figure
out
if
iwz
support
or
not.
D
C
G
C
C
My
wish
would
be
to
take
the
page
that
Jonas
just
put
together,
and
maybe
we
can
have
a
link
from
the
docs
to
the
same
leg.
Instead
of
having
that
support,
provider's
link
on
the
left,
we'll
just
have
supported,
plugins
or
like
plugins
in
that
page,
that
just
others
just
created
I
would
title
like
sub
sub
sections
for
each
plug-in
type
and
add
links
to
the
documentation
from
each
square
for
the
documentation
that
we
know
of
and
just
to
indicate
which
one
of
those
are.
G
G
G
I
I
I
Basically,
our
proof
of
concept
had
to
do
some
really
hacky
changes
to
get
the
rustic
component
working
here.
But
if
we
take
a
step
back
and
think
about
what
we
want
to
support
customs
case
for
Valero,
there's
really
two
ways
that
me,
as
a
user
of
Valero,
can
declare
these
custom
CAS.
So
one
is
at
install
time
and
one
is
per
object:
storage
on
a
backup,
storage
location,
the
install
time.
I
But
the
real
goal
of
what
I
was
trying
to
solve
and
my
team
is
trying
to
solve
with
a
proof-of-concept
was
well
when
I
created,
backup
storage
location.
How
can
I
actually
supply
a
secret
and
inside
that
secret,
storm
I,
see
a
bundle
then
actually
get
allow
that
to
be
used
through
the
entire
backup
and
restore
process.
So
from
the
Valera
side,
this
isn't
too
bad,
mainly
because
we're
using
the
AWS
SDK.
I
So
we
can
just
apply
the
see
a
bundle
to
a
AWS
session
object
where
the
question
that
I
really
wanted
to
bring
up
here
in
the
community
meeting
and
I
encourage
you
guys
to
read
this
all
in
your
own
time,
because
I
don't
want
to
like
run
through
the
entire
thing
right
now.
But
since
rustic
requires
this
cert
to
be
inside
the
container
for
lack
of
a
better
term
on
disk.
The
only
way
that
once
I
create
a
backup
storage
location
with
a
secret.
I
Because
I
don't
think
this
would
ever
be
accepted,
we
added
in
an
it
container
to
rustic
and
in
the
anit
container.
It
essentially
sets
up
a
sim
link
between
a
mounted
secret
and
the
Resik
Damon
set,
and
what
that
does
is
that
every
time
the
secret
is
updated
because
of
the
sim
link,
the
actual
folder
on
disk
and
rustic
gets
updated
with
the
proper
certs.
I
That
was
the
only
way
you
could
really.
We
could
get
this
working
in
like
a
quick
one
week,
proof
of
concept
here
and
I'm.
Basically,
whenever
a
question
is
like
from
the
Bolero
team,
is
it
a
non-starter
that
bolero
would
go
and
modify
the
rustic
daemon
set
every
time
it
BSL
is
created,
because
that
seems
like
it's
not
a
good
idea
to
me.
B
B
H
B
I
J
H
J
I
I
So
mainly
what
I
wanted
to
just
bring
up
was
mentioned,
and
that's
kind
of
an
open
issue
here.
I
would
appreciate
you
guys
just
take
a
look
and
leave
some
feedback.
Maybe
I'll
scope,
this
back
as
a
first
pass
to
install
for
the
install
case,
but
my
team
is
definitely
still
interested
in
solving
this
BSL
thing,
but
for
our
like
proven
concept,
we
are
moving
down
the
path
of
the
same.
Like
thing
it's
just
you
know
it's
not
definite,
not
a
deal.
B
D
D
D
J
I
Yeah
yeah.
Actually
there
is
a
workaround
today
with
manual
intervention.
If
the
user
goes
in
and
modifies
the
deployment
for
the
valero
and
modifies
the
rustic
Damon
said
you
can
get
this
working,
no
problem
yeah,
but
we
also
had
to
build
our
own
custom
velaro
image
to
pass
on
the
ca
surf
flag
to
rustic,
that's
kind
of
like
the
only
missing
component
in
valero.
A
C
D
We
just
kind
of
highlighted
it,
because
CRTs
were
an
exception
that
we
fixed
so
with
the
exception
of
Sierra
DS
falero
has
always
used
the
preferred
version
of
an
api
group
and
with
1
v
1
point
3
beta
1.
We
are
making
sierra
DS
respect
that
this
user
is
requesting
some
way
to
specify
something
different,
basically
saying
hey.
I
know
this
is
the
preferred
group,
but
I
want
this
one
and
so
I
need
to
come.
I
need
to
circle
back
to
this
and
basically
label
this
as
a
feature
request.
D
D
D
But
basically
we
ran
into
this
issue
which
actually
velaro
CR
DS
in
the
helm.
Chart
repo
have
an
issue
with,
and
so
basically,
if
a
CR
D
is
defined
without
a
schema,
it
can
be
submitted
to
a
kubernetes
116
cluster,
as
if
v1
beta,
1,
C
Rd,
just
fine,
so
kubernetes
116
will
take
it
it'll
be
readable
via
the
v1
CRT
end
point
without
the
schema,
so
you
request
it
from
kubernetes
using
kubernetes
ETL
get
you'll
come
back.
D
D
It's
backing
it
up
as
v1
and
if
you
go
to
restore
into
a
116
cluster,
restore
it
as
v1
CRT
without
the
schema
and
the
restore
will
fail,
because
a
v1
scaredy
must
have
a
schema.
Otherwise
it's
not
valid.
So
if
we
go
to
the
next
slide
in
in
in
visual
terms,
submitting
it
with
helm,
helm
is
submitting
it
to
the
v1
beta
one
endpoint.
D
So
kubernetes
says:
oh
yeah,
that's
a
v1
beta
1,
that's
completely
valid
and
if
you
say
hey,
give
me
that
that's
already
back
from
the
API
server,
it's
gonna
give
you
the
same
C
or
D,
but
as
v1
and
then,
if
you
go
to
the
next
slide
and
you
back
it
up
with
Valero,
Valero
is
gonna
request.
The
preferred
version
and
kubernetes
116
will
return
v1,
which
Valero
will
stuff
in
the
back
up.
Go
to
the
next
slide.
Kubernetes
then
submits
the
v1
CRD
to
the
API
server.
D
If
you
go
to
the
next
slide,
get
an
error,
saying:
hey:
you
need
to
have
a
schema,
because
Valero
basically
got
essentially
a
v1
beta1
document
but
updated
the
version
say
v1
so
slide
number
seven.
The
last
one
current
thinking
is
we're
thinking
about
making
a
backup
item
action
that,
as
we're
gonna
get
back
up.
If
we
see
a
v1
CRT
with
no
schemas
at
all,
like
that
fields,
nil,
we're
gonna
back
it
up
as
V
1
beta
1,
and
when
we
restore
it,
we
will
restore
it
as
V
1,
beta
1.
D
So
this
is
actually
we
saw
this
in
the
beginning
of
this
whole
saga,
and
we
thought
that
was
a
bug,
but
this
actually
turns
out
be
the
correct
behavior,
because
we
can't
reconstruct
a
paint
schema
for
you.
We
have
no
idea
what
about
it.
Schema
is,
but
I'm
going
to
do
some
more
research
to
make
sure
this
is
our
best
path
and
maybe
put
out
a
demo
of
this
I've
got
a
very
basic
plugin.
D
That
just
says:
hey
yes,
I
found
I
found
v1
CRTs
with
no
schemas
I
need
to
do
do
some
more
work
on
that
to
make
sure
that
this
will
actually
work
and
you
can
double-check.
My
findings
with
kubernetes
is
API
conventions,
Justin,
B
and
compatibility
guarantees
just
to
make
sure
this
is
correct.
So
we
don't
delay
1.3
inning.
Any
longer
than
we
need
to
and
we're
not
running
into
other
special
cases.
D
That's
that's
the
update
that,
on
what
I've
been
working
on
and
what
I've
seen
Genting
original
report
was,
there
was
a
nil
pointer,
dereference
and
that
no
pointer
was
there
was
no
schema,
so
I
did
the
nil
check
and
then
it
led
to
the
API
server
complaining
so
yeah,
that's
that's
the
current
state
of
things.
So,
given
that
summary,
any
questions.
C
Just
one
or
two
other
comments
now
that
I
got
the
picture
of
this
whole
entire
issue.
I
think
that
upgrading
CRT
to
version
one
in
making
invalids
is
a
failures.
I
would
think
that
we
definitely
should
not
go
in
that
direction.
So
if
a
CRT
comes
in,
that's
a
beta
and
doesn't
have
a
schema,
then
we
just
make
it
better.
C
D
D
But
I'm
not
exactly
sure
eggs
and
I'm
not
sure
exactly
what
manipulation
I'm
gonna
have
to
do.
Whether
I
don't
know
if
all
just
re
query
for
like
generate
a
new
client
for
view,
1
beta,
1,
CR,
DS
and
get
it
or
just
change
the
struct
or
just
add
beta
1
to
the
unstructured
field,
or
something
like
that.
But
yeah,
that's
that's.
My
hope
is
I
can
just
change
that
I
inspect
it
and
say
hey!
This
I
was
given
this
it's
v1,
it
doesn't
have
a
structures.
D
1,
so
that
seems
like
the
correct
thing
that
correct
thing
for
Valero
to
do
in
terms
of
long
term
support
or
not
gonna
be
able
to
do
that
forever.
I
think
the
one
beta
1
is
only
going
to
be
around
for
three
or
four
more
releases.
If
that
I
can't
remember
exactly,
we
went,
but
not
a
whole
lot
longer
make
a
year
so
folks
are
gonna
have
to
upgrade,
including
us.
The
home
there's
already
an
issue
to
upgrade
our
helmet
chart
with
this.
The
structural
schema
in
but
yeah.
D
Right
it
becomes,
it
becomes
a
larger
kubernetes
issue,
I'm
sure
we'll
get
bug
reports
for
it,
because
people
will
be
like.
Oh,
my
old
backup,
I
can't
pull
forward,
and
maybe
when
we
get
closer
to
end
of
life
cycle,
we
can
start
like
pushing
out
warnings
to
say
hey
just
so
you
know
this
does
backup
won't
be
valid
on
the
next
version
of
kubernetes
or
something
like
that.
If
people
restore
it
in
that
time,
maybe
they
won't
but
yeah
at
some
point
it's
out
of
our
hands.
We
can.
B
B
Allow
the
user
to
say:
hey
I,
want
to
back
up
the
view,
one
beta
one
version
of
CR
DS:
either
you
know
for
this
backup
or
have
that
as
a
server
flag,
but
that
you
know
that
could
be
one
approach
that
basically
puts
the
control
in
the
ends
and
the
users
I,
don't
know
if
it's
preferable
or
not,
but
just
wanted
to
throw
that
out
there
and
then
the
other
thing
is
that
you
know
this.
This
may
be
a
general
problem
that
we
just
haven't
encountered
before
like.
B
If
this
is
you
know,
if
this
scenario
where
it
can
be
converted,
but
then
it
can't
be
gotten
and
reapplied.
If
that's
a
general
scenario
that
can
happen
across
any
kind
of
kubernetes
type
and
version
upgrade,
and
we
may
need
to
try
and
think
about
a
more
general
solution
to
this
problem
down
the
road
we
we've
never
seen
this
before
with
any
of
the
other
version
upgrades
for
the
kubernetes
types,
but
that
doesn't
mean
it
couldn't
happen
more
in
the
future.
D
Yeah
and
I
think
my
suspicion
is
crts
just
happened
to
be
the
first
one
to
hit
it
right
because
they
they're
graduating
such
that
my
understanding
is
Sierra.
Ds
are
coming
with,
with
more
and
more
stuff
kind
of
being
nudged
away
from
kubernetes
core
Sierra
DS
were
pushed
to
graduate
to
be
GA,
so
they
were
pulled
out
of
beta
such
that
they
could
be
relied
upon
for
things
like
CSI
snapshot.
D
J
D
D
D
So
yeah,
unless
anybody
has
any
more
thoughts
on
that,
that's
all
I
had
I'll
keep
an
eye
out
for
issues
as
people
are
testing
in
all
tests.
I
will
retest
with
I.
Part
of
part
of
the
issue
was
I
completely
missed
testing
with
the
helm
chart.
So
apologies
are
there
but
I
know
Jen
things
not
on
call
Thank,
You,
Jen
team
for
testing
at
and
catching
that
with
the
beta
really
appreciate
that.
A
B
Yeah
I've
got
a
handful
of
contributions
in
the
last
week,
so
we
got
a
couple
of
Doc's
pr's
in
the
first.
One
is
from
mark
ready,
updating
our
Doc's
to
point
to
the
new
help
chart
repo
so
appreciate
that
update
and
then
Jay
Nolte
I
know.
Jonah
was
on
the
call
earlier.
I,
don't
see
him
anymore,
he
may
have
dropped,
but
added
notice
that
we
were
missing
kind
of
API
documentation
or
yeah
Mel
documentation
for
the
restore
API
type.
So
he
added
a
doc
there.
B
So
much
appreciated,
and
then
my
uncle
I
think
is
on
the
call.
We
got
the
pr
merge
to
allow
volume
snapshot
or
plugins
to
rename
tv's
during
her
stores
through
the
set
volume
ID
function,
so
that's
gonna
be
helpful
for
their
use
cases
around
cloning
Peavey's.
I
believe
so.
We
should
be
able
to
include
that
in
1.3.
So
thanks
for
FPR
and
then
yeah,
just
a
general
call-out
to
genting
who's
been
doing
a
lot
of
testing
digging
into
these
CRD
issues,
getting
a
few
different
PRS
and
so
I.
Don't
he
didn't.
J
Yeah
sure
so,
Antony
I'm
not
sure.
If
he's
on
the
call.
He
was
also
super
helpful
in
like
testing
out
various
versions
and
like
providing
like
a
very
crisp
test
case
or
a
way
to
reproduce
the
memory
leak,
which
kind
of
shortened
the
time
to
discover
and
fix
or
identify
the
fix.
So
that
was
awesome.
Thank
You,
Anthony,
yeah.
A
Awesome
yeah.
Thank
you,
everyone
I!
Thank
you,
everyone
for
for
joining
today.
If
you
have
any
questions
or
follow-ups,
please
get
those
up
on
the
velaro
slack.
It's
like
channeling,
the
kubernetes
fact
there
we
go
and
with
that,
thank
you
all
for
joining
and
have
a
fantastic
rest
of
the
week.
Thank
you
guys,
bye.
Everyone
bye.