►
From YouTube: Velero Community Meeting - April 27, 2022
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
All
right
guys
welcome
joining
the.
A
A
According
to
the
original
plan,
we
are
two
weeks
a
little
more
than
two
weeks
away
from
the
plan.
Fc
date,
so
first
party
status
update
recently
I've
been
working
on
the
refactor
of
the
backup
deletion
controller.
I
have
a
pr
and
also
we
fixing
I'm
fixing
some
issue
in
the
csi
plugin.
It's
mainly
about
how
you
think,
the
deletion
of
the
volume
snapshot,
content
and
I'm
also
reviewing
some
prs
from
shuban
and
other
guys,
so
we're
making
progress
to
the
fc
next
one.
B
A
So
the
output
eventually
will
be
a
pr
for
the
design
that
we,
how
we
gonna
replace
restricted
with
copyright.
B
C
A
Yeah
I
I
I
took
a
look
at
that
pr.
I
have
a
little
comment.
We
don't
have
any
test
written,
I
don't
see
you
made
any
tasks,
I
mean,
I
don't
see
you
make
any
change
to
the
test.
C
D
A
Yeah,
I
think
probably
we
should
do
some
basic
tests,
at
least
it's
a
surprise
that
it
doesn't.
We
don't
have
any
tests
for
the
rest
of
repo
controller.
E
Recently,
I'm
working
on
the
first
one
is
adding
insecure,
skip
tis
verify
flag
for
the
interpreted
rustic
the
faux
malero
worm
1.8,
the
integrated
rustic
doesn't
support
the
c
secure
flight
for
the
rustic,
so
I
updated
one
of
the
of
the
integrated
resting
and
enable
valero
to
pass
the
flag
to
the
rustic.
E
The
second
one
is
related
to
the
cells
I
plug
in
some
issues.
Some
issue
of
it
part
of
the
one
of
them,
is
about
the
remove
the
duplicate
snapshotting
of
the
csi
backup
and
another.
Is
that
a
metric
related
to
the
site,
the
snapshot
attached
and
the
other
one
is
where
to
remove
the
volume
snapshot
taking
during
the
csi
backup
process.
A
So
so
I
I
think,
as
for
the
csi,
we
plan,
the
only
one
big
issue
is,
I
I
don't
remember
the
number
but
the
one
that
we
don't
want
to
leave.
Any
volume
snapshot
resource
remain
after
the
backup
is
done
so
we're
gonna,
set
the
deletion
policy
to
retain
and
remove
them
right,
yeah.
Okay,
any
question
for
soon.
D
I'm
doing
some
very
effective
for
the
portable
restore
controller,
this
computer
and
how
submit
pr
the
pr
is
under
a
view,
and
the
second
thing
is
I'm
trying
to
fix
the
issue
that
the
rustic
controller
doesn't
pick
up
the
ingress
ingress
crs
after
restarting,
and
I
will
some
pr
to
introduce
a
short-term
fix.
The
fix
will
try
to
mark
all
the
in-progressions
as
field.
D
Maybe
maybe
we
need
a
long-term
long-term
fix,
but
the
fix
maybe
need
a
well-designed-
and
I
know
maybe
women
in
more
time
for
the
long-term
fix
and
this.
This
update
from
my
side.
A
D
Yeah
maybe,
but
I'm
not
sure
for
now.
A
A
F
So
I
I
don't
know
in
terms
of
historically
whether
you
know
where
this
fit
in
the
in
terms
of
future
goals.
I
I
think
we
all
from
what
I
remember
vaguely-
and
this
is
maybe
next
time
when
dave
is
on.
He
might
have
some
insight
here,
but
from
what
I
remember
when
this
has
come
up
before.
I
think
it
was
one
of
these
things
where
you
know
the
fact
that
we
never
pick
up
a
backup
that
was
in
progress
when
on
restart.
F
You
know
it's
kind
of
a.
We
recognize
that
the
code
doesn't
do
this
now
and
but
it's
something
that
probably
makes
sense
to
tackle
in
a
in
a
better
way
than
just
marking
it
as
failed,
especially
in
a
store.
I
mean,
I
talked
to
sean
about
this
a
little
bit
over
the
last
couple
of
weeks,
and
you
know
he
was
kind
of
the
opinion
that
we
really.
You
know
that
interim
fix
that
we
have
in
place.
That
way.
I.
G
F
Remember
whether
we
merged
it
yet
or
we're
going
to
merge
it,
but
where
we
just
market
has
failed,
is
certainly
an
improvement
over
leaving
it
in
progress.
F
But
the
issue
is
and
again
that's
that's
not
rustic,
that's
just
a
regular
funeral
era,
but
I
mean,
I
think,
they're
kind
of
related
in
the
sense
that
basically
right
now,
neither
rustic
nor
valero
handle
very
well
things
in
progress,
and
so
I
think
the
fix
to
market
has
failed
is
certainly
something
we
need
to
do
because
in
progress
confuses
users
because
they
say
hey.
G
D
F
Yeah,
so
I
think
marking
things
that
are
listed
as
in
progress
has
failed
because
we're
never
going
to
get
to
him
again
as
an
interim
solution,
but
I
think
ultimately,
valero
ought
to
be
able
to
recover
from
this,
because
one
of
the
cases
where
this
happens
is,
for
example,
with
an
automated.
You
know
overnight.
Backup
runs
out
of
memory,
the
out
of
memory.
You
know
bounces
the
pod
and
people.
F
Don't
necessarily
recognize
that
that
you
know
scheduled
backup
didn't
happen,
and
if
this
happened
several
days
in
a
row,
they
might
have
missed
several
backups
in
a
row
and
some
some
way
of
re
retrying
restarting,
might
make
sense.
I
mean
again,
this
is
where
I
was
talking
to
sean
last
week,
not
so
much
about
rustic,
but
just
in
general
and
for
backups
in
particular.
If
we,
you
know
the
thinking
would
be,
we
could
probably
do
something
that
wouldn't
be
that
difficult
to
say.
F
You
know
when,
when
rustic,
when
villarreal
restarts
the
the
code
right
now,
that
makes
it
marks
it
failed
if
we
could
determine,
for
example,
that
nothing
was
written
to
object
storage.
Yet
we
hadn't
gotten
to
that
point.
That
means
nothing
has
persisted
from
the
previous
attempt,
so
it
should
be
safe
to
just
start
over.
F
You
know
recover
from
a
halfway
written
to
disk
backup,
I
think
restore,
might
be
simpler
because
you
know
if
you
restore
and
it
fail-
and
you
don't
finish
it
and
then
on
restart,
if
you
restart
at
the
restore
for
those
resources
that
are
already
there
depending
on
this
new
function
that
we're
talking
about,
but
the
current
value
functionality
is
you're
not
going
to
touch
things
with
them
anyway,
they're
already
there.
F
So
if
you've
got
say,
50
out
of
100
resources
restored
and
then
on
restart
of
valero,
you
kind
of
kicked
it
again,
then
the
first
50
of
those
that
we've
had
before
just
wouldn't
get
written
and
the
next
50
would
get
written
and
then
and
and
we
and
this
could
be
a
configurable
behavior
too.
You
know
it's
one
of
these
things
where
you
know
there
could
be
a
server
setting.
You
know,
do
you
retry
on
restart
or
not?
F
I
don't
know,
that's
that's
something
we
would
need
to
discuss
and
figure
out,
but
it
you
know
the
fact
that
valero
can't
recover
at
all
from
something
you
know
a
middle
of
a
backup
or
a
bit
of
a
restore
pod,
but
because
the
conv,
the
valero,
doesn't
control
the
you
know
it's
pod
necessarily.
I
mean
there
are
various
reasons
why
the
deployment
or
the
cluster
could
bounce
the
pod,
and
then
we
just
lose
that
backup.
A
Yeah,
I
think
yeah,
but-
and
it
will
be
also
tricky
that
if
we
you
know
the
restore,
is
halfway
done
and
bellero
restarted.
We
market
failed,
but
there
are
some
resources
in
the
target
cluster
and
it
will
also
be
tricky
for
users
to
clean
that
up
if
market
failed,
but
it.
A
F
Pretty
cool
yeah,
I
I
mean
I
mean
actually
getting
to
the
point
where
we
want
to
re,
restart
and
retry.
A
backup
restore,
has
got
some
edge
cases
we're
going
to
have
to
work
through,
but
the
thing
is
the
status
quo
right
now,
what
happens
before
this
rustic
issue
and
before
the
the
valero
one,
which
I
think
I
think
that
has
been
merged?
I
know
it's
been
at
least
well.
I.
C
F
I
acted,
but
I
don't
know
the
point
is
right
now,
if
you
bounce
valero,
meaning
in
one
eight,
for
example,
if
you
bounce
valero,
with
an
in
progress,
backup
or
restore
that
backup
or
restore,
will
never
be
completed,
but
if
you
do
a
valero
backup
or
restore
get,
it
will
show
up
as
in
progress
forever,
even
though
subsequent
backups
are
running
and
completing
you
know,
when
you
restart
the
server
the
pod,
that
backup
is
still
in
progress,
or
at
least
if
it's
listed
as
in
progress,
is
not
in
progress,
and
then
you
submit
a
new
backup
that
backup
is
actually
in
progress.
F
So
the
problem
there
is,
if
a
user
goes
to
look
at
backups,
and
I
you
know
they
want
to
look
at
the
what
backups
in
progress
they
see,
two
that
are
in
progress,
one
of
them
is
actually
running
the
other
one
is
just
an
old
backup
that
never
got
cleaned
up,
yeah
and,
and
so
the
intent
was
for
valero
and
rustic.
I
mean
this
is
the
kind
of
the
interim
fix
is
that
we
know
that
these
are
never
going
to
complete.
F
You
know
marking
them
has
failed
so
that
the
user
sees
that
it
didn't
finish
with
the
failure
message
to
that
effect
that
it
was
you
know
in
progress.
You
know
when
the
server
goes
killed
at
least
gives
the
user
a
more
accurate
view
of
what's
currently
happening
without
actually
changing
the
workflow.
We
still
aren't
continuing
backups
after
we
start
continuing
rustic
actions
after
we
start
that's
a
kind
of
longer
term
future
conversation
to
figure
out
how
best
to
do
that.
A
Yeah
yeah,
I
thank
scott
and
I
also
took
a
quick
look
at
the
milestone.
One
night.
It
seems
we
have
32
close
but
still
38
issues
open.
I
know.
A
Maybe
move
some
of
the
issues
out
of
one
nine
before
the
fc
and
we'll
keep
you
updated
in
the
slack
channel,
because
the
next
community
meeting
in
beijing
will
be,
I
think,
it's
a
main
around
maintenance
or
the
next
meeting.
When
we
have
the
community
meeting,
it
will
be
less
than
a
week
for
the
plan
every
day,
so
we'll
see
and
keep
you
updated.
F
Just
a
quick
comment
on
that,
when
you're
going
through
those,
I
think
it's
important
to
keep
it
make
a
distinction,
because
some
of
those
that
are
you
know
not
open
may
not
have
even
been
started
or
may
not
be.
You
know,
making
big
progress
on
and
so
clearly
those
are
the
ones
that
you'd
probably
end
up
having
to
push,
but
some
of
the
ones
in
progress
may
already
have
pr's
and
all
the
only
thing
holding
them
up
is
getting
reviewed
or
getting
corrections
or
whatever
you
know
fixes
in
there.
A
Yeah,
so
so
that's
the
next
thing
I
want
to
discuss
here
in
this
meeting,
so
it's
in
the
meeting,
yeah
yeah
yeah,
so
any
update
in
the
pr
regarding
the
resource
policy
and
the
selectors.
I
know
I
had
some
very
late
comment
regarding
the
naming
of
the
selectors,
but
I
think
we
really
need
to
distinguish
like
the
singular
and
plural
selector,
even
in
the
api
and
in
the
code.
Otherwise
it's
when
I
want
to
read
the
code.
It's
really
confusing.
H
Yeah,
I
agree
on
that
for
the
restore
policy
I'm
almost
done.
I
was
testing
some
scenarios
I'll
update
it
today
itself
and
for
selectors.
I
was
thinking.
Let's
go
with
the
or
label
selector
the
last.
F
F
H
A
Any
strong
disagreement
regarding
the
design
of
the
existing
resource
policy
right,
a
resource
policy.
D
F
Early
on,
we
had
some
some,
you
know
comments,
and
there
were
some
back
and
forth.
There
were
some
changes
made,
but
I
think
at
this
point
things
have
pretty
much
converged.
A
A
A
Yeah
yeah,
you
please
just
add
one
note
and
we
can
remove
this
approach
two
and
three
either
way.
It's
fine
just.
F
Section
saying
we
considered
these,
but
they're
they're
they're
out
of
scope
now,
but
that
way,
because
I
think
the
approach
three
is
something
we
might
consider
in
the
future.
If
we
need
that
additional,
so
it
would
be
good
to
have
this
documented,
but
but
we
need
to
be
clear
that
this
is
not
being
implemented
for
this
milestone.
This
is
something
we've
thought
about.
It
might
be
done
in
the
future,
but
the
scope
for
this.
You
know
this
release
is
just
the
approach.
One.
H
A
Yes,
I
think
I
think
the
first
thing,
maybe
hopefully,
by
the
end
of
this
week,
we
have
more
of
this
design
prs
and
I
I
think
you
asked
me
to
review
the
the
selectors,
the
coaching
selectors
yeah.
D
A
You
know
monitoring
those
ones
and,
as
for
the
existing
resource
policy
code,
pr,
I'm
not
sure
who
has
been
reviewing
that.
Please.
F
Yeah,
I
have
an
issue,
I'm
not
sure
if
there's
a
second
one
yet,
but
obviously
we
need
the
two
acts
and
I'll
I'll
re-review
it
once
he
makes
that
other
change.
That.
F
I
requested,
but
but
yeah
if,
if
there's
not
a
second
reviewer
yeah
at
night,
I
don't
remember
offside,
my
head.
If
you
know
we
do,
we
need
someone
else
and
just
in
case
there's
some
other
things
that
I
missed.
You
know
interviewing
that
that
are
needed.
B
F
Be
fixed
or
changed
yeah.
A
We're
going
to
get
that
pretty
totally
since
you've
been
reviewing
the
design.
Also,
will
you
take
some
time
to
reveal
the
code
change
for
the
existing
resource
policy?
I.
B
A
B
Had
time
to
work
on
that,
I
may
be
able
to
work
on
that
late
of
this
week
on
that
or
early,
you
know,
after
the
vacation.
A
Yeah,
okay,
so
that's
a
step!
So
scott!
You
have
our
phone.
You
have
any
update
regarding
the
plugin
versioning
refactor.
I
think
that
will
be.
A
I
Been
working
on
it
in
the
background,
but
I
haven't
I
kind
of
got
stuck
on
it
a
few
weeks
ago
and
then
I
switched
to
another
project,
so
I
I
now
I
think
I
will
try
to
get
some
time
on
it
this
week,
so
we
can
get
at
least
some
ignition
initial
pr
for
you
to
to
look
at
I'm
not.
G
A
Yeah,
so
so,
actually
the
my
thought
is:
if
we
can
make
sure
that
or
we
don't
need
to
really
rush
to.
You
know
try
to
make
a
lot
of
things
before
the
1
9
fc.
It
will
be
a
lot
of
pressure
on
the
pr
review
and
it
will
some
pressure
for
the
quality.
If
that
happens,
I
would
rather
push
the
fc
dave
or
move
things
out,
so
I
think
it
would
actually
be
better
if
you
can
make
decision
that
this
one.
We
are
not
going
to
merge
in
one
line
so
from
you.
A
F
G
F
Yeah,
I
I
I
would
suggest
in
that
case,
then
we
still
would
would
like
to
get
it
merged,
basically,
as
early
as
possible.
In
the
you
know,
post
one
nine,
in
other
words
the
for
one
of
the
first
things
we
want
to
merge
during
that
110
developments,
because
that
way,
that'll
give
us
time
during
that
110
time
frame
to
you
know,
start
putting
together
the
plan,
for
you
know
what
the
v2s
are
going
to
look
like.
F
What
do
we
need
to
add,
because
that's
one,
this
difference
with
this
work
versus
the
other
two
we
just
talked
about
is
because
the
the
other
two
things
that
shiban
are
working
on
are
actual
user,
visible
features
that
you
know.
We
have
people,
at
least
from
the
red
hat
side
that
are
interested
in
right
now,
and
the
phones
work
with
the
plug-in
version,
and
we
also
have
people
that
need
that,
but
it
by
itself
it
doesn't
produce.
You
know
a
user
visible
feature.
It
sets
the
groundwork
for
a
future
user
visible
feature.
F
A
Yeah
and
another
big
thing
is
the
upload
progress
monitor
that's
been
working
by
dave.
I
I
used
to
review
that
pr
left
a
lot
of
comments,
but
didn't
have
time
to
work
through
them
again.
So
my
feeling
is
that
as
a
whole,
it's
still
at
risk.
A
Maybe
we
can
merge
some
pr
by
day,
but
after
that
there
are
something
left
I'll
I'll
text
on
since
dave
is
not
in
this
meeting
I'll
check
with
him
regarding
the
status,
I'm
curious
in
addition
to
the
you
know,
vmware
engineers
that
used
to
work
with
they
want
the
upload
progress,
monitor
scott
or
phone.
Do
you
know
anyone
from
you
know?
External
maintainer
have
revealed
that
design.
F
A
Okay,
I'll
I'll
think
with
them.
I
generally
understand
how
it's
gonna
work,
but
the
tricky
thing
is
it
try
to
in
introduce
a
new
type
of
plugin,
which
I
think
the
vsphere
plugin
will
be
refactored
to
the
new
type
of
plugin.
But
since
now
I
hear
some
new
plan
regarding
the
vco
plugin.
So
first
thing
is:
you
know,
he's
trying
to
make
some
change
to
the
framework
to
introduce
a
new
type
plugin
without
implementing
any
of
this
new
type
of
plugin.
A
So
it's
tricky
to
reveal
like
how
things
may
work
together
and
if
and
I'll,
also
think
with
that
like
what,
in
his
mind,
should
implement
this
plug-in
and
what's
left
if
the
apr
is
merged.
But
my
feeling
is
that
one
may
be
partially
merged
in
one
line
and
there
will
be
some
additional
work,
I'm
not
sure
will
they
are
still
interested
to
work
on
them,
but
yeah
I'll
also
include
him
and
keep
you
updated.
Yeah.
F
A
Yeah
and
I
I
also
want
to
suggest
that
for
v1,
we
should
either
move
everything
to
the
new
constructor
or
leave
it
as
is
now
right.
We
don't
want
to
have
things
half
done.
F
A
Okay,
okay,
so
any
thing
you
guys
want
to
talk
about
regarding
this
topic
for
the
issues
proposed
by
red
hat
and
the
phone
has
been
working-
okay,
no
one!
I
I
think
do
we
want
to
take
some
time
to
discuss
this
data
mover
design
by
shubham,
or
we
can
have
a
separate
meeting.
What
do
you
guys
prefer?
I
know
sean
and
dylan
have
some
has
been
participating
in
the
design
discussion.
H
A
I'm
asking:
are
you
targeting
it?
I
I'm
I'm,
I
I
think
that's
a
flexible
one
for
one
night
so,
but
but
I
think
you
guys,
I
have
some
email
communication
with
dinner
and
it
seems
he
has
some.
H
B
H
Something
which
we
agreed
upon
the
use
cases,
the
user
workflow
and
the
requirements.
So
this
was
like
a
phase
one
design
sort
of
thing.
So
that's.
H
Yeah,
we
would
want
to
merge
this
pr
first,
if
we
agree
upon
all
the
things
in
this
pr
which
are
not
interface,
design
related,
but
requirements
and
user
workflow
related
and.
A
A
Okay,
if,
if
that's
the
case,
I
think,
but
I
still
have
some
comment:
have
you
ever
had
a
chance
to
take
a
look.
A
Okay,
okay,
so
I
my
feeling
is
that
at
least
to
clarify
something
if
needed,
we
can
have
some
zoom
meeting
if
needed.
I
know
you
guys
prefer
not
having
a
meeting
in
the
evening.
It's
okay!
We
have
the
meeting
in
the
evening
in
beijing,
time
zone,
okay,
yeah
that
works
yeah,
so
so
yeah.
So
we
can
continue
the
async,
a
communication
for
this
design.
H
Comments
which
you
left
and
again
bring
you
on
slack
then.
H
A
Okay,
any
comment
regarding
the
data
more
design
right
now,
but
I
just
want
to
check
that
they
block
any
plans
by
red
hat
to
deliver
the
feature.
H
A
I
mean
I
heard
from
dylan
that
he
wants
something
to
work
for
the
oat
oadp
end
user
in
the
money
or
during
time
frame.
A
A
A
Well,
I
think
we
have
some
internal
reviewing
and
also,
we
also
feel
it
seems
that
the
current
pr
is
a
half
a
user
scenario
and
a
half
a
little
tag,
detail
yeah.
So
that
is
a
little
bit
confusing
to
us.
Please
check
this
comment
here
and
let
us
know
what
you
think:
yeah.
H
Yeah
sure
thing
do
the
first
thing
tomorrow:
okay,.
A
Okay,
I
think
that's
all
from
our
site
anything
else.
Anyone
want
to
discuss.
G
A
G
Hi,
I'm
I
have
a
question,
I'm
the
first
time
y'all
joining
the
call
hi
this
may
see.
So
let
me
introduce
myself
give
a
little
background.
I'm
john
sand.
I
work
for
data
stacks
and
I
work
on
a
maintainer
of
some
a
series
of
operator
projects
under
the
kate
sander
dot
io
umbrella
and
for
primarily
focused
on
orchestrating
maintaining
cassandra
clusters,
and
so
the
kate
sander
operator,
which
we
released
a
1.0
release
earlier
this
year,
provides
multi-cluster
support.
G
Cassandra
is
a
distributed
if
you're
not
familiar
distributed
out
of
the
box
with
and
so
multi-region
is
kind
of.
It's
it's
a
primary
use
case,
and
so
you
know
to
facilitate
that
multi-click
implemented
multi-cluster
support.
So
we
have
part
of
part
of
the
operator.
We
have
backup
restore
functionality
with
the
open
of
another
tool
that
we've
integrated.
That's
specifically
focused
on
cassandra
is
actually
open
source
through
spotify
and
but
I
I
had
recently
been
started
doing
some
prototyping
with
valero.
G
It's
there's
some
features
that
we're
trying
to
implement
that
valero
already
has
like
syncing
the
the
remote
storage
with
objects
in
an
ncd
in
terms
of
the
source
of
truth
schedule
backups.
So
there
are
some
things
that
you
know:
there's
not
total
fee,
there's
some
things
that
our
backup
tool
does
but
then
performance
wise.
G
It's
it's
a
dramatic
difference,
so
I
I'm
very
interested
I've
done
a
little
poc
for
for
our
projects,
and
so
you
know
something
I'm
going
to
continue
with
so
my
use
case
here
is
a
little
bit
odd
in
that
right
now.
So
cassandra
there
are
still
a
lot.
There's
a
large
number
of
cassandra
users
with
very
large
install
bases
not
running
in
kubernetes,
and
some
of
these
users,
for
various
reasons
cannot
or
won't
at
any
time
your
time
soon
migrate
to
kubernetes.
G
So
we're
are
exploring
building
out
a
basically
is
not
more
than
just
an
operator,
but
a
management
control
plane
for
that
would
run
in
cluster
within
kubernetes,
but
managing
cassandra
clusters
that
run
outside
of
kubernetes
and
now
the
under
the
primary
api
for
any
clients
would
be
through
kubernetes.
So
everything
would
be
crd
based.
G
So,
even
though
the
the
cassandra
nodes
will
exist
and
run
out
of
cluster,
we,
you
know
there
will
be
custom
resources
to
represent
those
and
one
of
the
one
of
the
priorities
that
we
need
to
address
for
this,
so
is
backup
restore,
and
while
we
have
an
existing
backup,
resource
solution,
which
was
actually
developed,
is
a
command
line
batch
processing
tool
that
runs
it's
a
python
tool
that
runs,
you
know,
doesn't
wasn't
defined
for
designed
for
kubernetes.
Well,
that
could
be
used.
G
I'd
really
really
want
to
explore
valero.
So
I
asked
I'm
slacking
somebody
mentioned
joining
the
meeting
tonight
and
now
by
and
I
realized
valero
was
designed
for
kubernetes.
I've
only
looked
at
the
code,
just
a
little
bit
at
the
code
base.
To
answer
some
of
my
questions.
While
I
was
working
on
my
prototype,
so
you
know
if
there
is
a
if
there
were
plugins
or
or
hooks
that
could
implement,
you
know
things
that
you
know
I'm
certainly
open
to
extending.
G
You
know
various
subsystems,
if
you
will,
if,
if
that,
would
open
up
the
possibility,
but
I'm
not
familiar
enough
with
the
project
to
know,
even
if
that
is
the
possibility.
A
Yeah
there
is
actually
a
valero
supports.
Some
kind
of
plugins,
like
bada
backup
item
background.
First,
I
assume
you
have
a
basic
idea
like
how
the
arrow
works
right,
just
walk
through
this
resource
by
calling
the
apis
or
to
serialize
them
and
take
snapshots
for
the
tvs
and
bundle
them
together,
upload
to
the
object
store
as
a
backup,
and
they
do
it
reversely
in
the
resource
and
in
that
process
you
can
write
some
plugin
like
the
backup
item
action
plugin.
So
to
trigger
your
code.
A
When
you
see
some
particular
resource,
I
think
that
one
may
be
useful
for
you,
and
also
there
is
some
hooks
on
the
pod
basis.
You
can
add
you
know
to
the
annotation
of
the
pod
and
so
that
you
can
trigger
some
run
some
command
within
the
pod
before
or
after
the
backup
restore.
I
think
those
are
some
way
to
extend
valero's
workflow.
That
may
be
helpful
for
you.
You
can
check
the
valero's
official
website
and
there
is.
I
think
the
dog
should
be
clear
enough
to
help.
A
G
A
G
A
Well,
the
example
is
that,
when
valero
serializing
the
resources
when
he
sees
the
party,
try
to
you
know,
add
the
pvc
and
take
snapshot
for
the
pv,
and
that
is
the
source
mounted
to
the
part.
So
that's
one
you
know
usage
of
this
plugin
just
to
check
the
code
or
see
the
dock,
but
writing
this
plugin
is
a
little
tricky
because
it's
a
grpc,
plugin
and
based
on
the
go
plugin
framework.
G
I'll,
take
you
to
your
word,
but
with
with
our,
I
think,
with
what
we're
gonna
be
doing
is
probably
a
lot
of
grpc
and
with
our
existing
backup
restore
solution.
The
calls
are
made
the
the
it's.
It's
not
there's
some
inefficiencies
we
have
to
run.
We
run
a
sidecar
container
that
for
the
backup
service-
and
you
know
it
makes
talks
to
our
operator
over
grpc,
and
you
know
it's
specific,
so
I'm
only
interested
in
backing
up
the
pvs
and
pvcs,
but
I
did
some
like.
G
I
said
I've
got
that
prototype
work
with
the
enough
of
the
solution
working
with
valero
and
and
with
a
multi-cluster
prototype
so
to
make
it
makes
it
even
better.
So
when
I
get
a
chance
to
maybe
I'll
write
up
a
blog
post
of
doing
multi-cluster
backups
for
cassandra
with
valero,
so
you
can
keep
an
eye
out
for
that.
Yeah.
G
Be
cool,
keep
us
updated,
yeah,
so,
okay,
I'll
I'll
I'll,
be
probably
you
know.
Hopefully
you
know
in
the
coming
days
weeks
lurking
some
more
and
on
the
slack
channel
and
maybe
with
some
more
questions
just
to
see
how
feasible
this
is.
So
I
appreciate
you
taking
the
time
to
to
explain
that.
Thank
you
all
right.
Thank
you.
Bye.
A
All
right,
so,
okay,
in
addition
to
that
anything
else,
anyone
want
to
discuss.