►
From YouTube: Velero Community Meeting/Open Discussion - May 12, 2020
Description
May 12, 2020
The v1.4.0-beta.1 release is out! Please try it out in your non-critical environments and give us feedback.
Status Updates
[carlisia]
Lotsa PR reviews
Moving our BSL CRD to a kubebuilder CRD + controller (Vince Prignano has been offering guidance)
Next:
move BSL logic out of the controller
add handlers to create/update
[ashish]
You can now supply feature flags to Velero using the Velero CLI. Read more about it here
Getting CSI integration with Velero ready to ship. Take a look at the CSI page on our website
Backup syncing: include CSI API objects
Backup deletion: cleaning up storage snapshots on backup deletion.
Set up Github actions to build and push container images for velero-csi-plugin. Hoping to do this for all our repos.
steve
finalizing things for v1.4.0-beta.1
starting to look at 1.5 planning
nrb
Writing the v1.4 blog post
Catching up reviews that I pushed out in order to get the beta out; particularly want to get a look at the restore hooks document this week
Will be testing v1.4.0-beta.1 against a few different platforms and versions of Kubernetes to verify the overall package
Some docs PRs to get in for the GA release
Discussion Topics
stephanie
Review community feature reqeust for backing up restic volumes by default
Detailed product requirements are available and we will be providing a review of the three main use cases during this meeting.
You can find the product requirements here: https://drive.google.com/file/d/1i5IQfuOtsHhGuYiiTrCZ6zjVhtKWLRqV/view?usp=sharing
And, sample annotations to illustrate the main product use cases are here (these are also included in the UI/UX section of the requirements doc): https://gist.github.com/skriss/dccc494aea67a47473e184c18aac9af6
vitta
Backup/Restore progress for operations performed by velero plugins
In similar lines, if plugin have their own CRs (similar to PVB, PVR), how to get those CRs backed up from source cluster and get them synced at destination cluster regularly?
Finalizers on Backup CRs to protect data from getting deleted
A
B
So
what
I
want
to
do
is
make
sure
that
the
allo
that
can
be
instantiated
loaded
to
the
server
and
you
know
alongside
the
other
crzs
and
I
paint
vince
real
quick
and
he
said
it
should
be
possible,
and
so
what,
if
I
get
that
working,
then
I'm
going
to
move
the
the
logic
from
the
current
control?
The
controller
excuse
me
that
I
have
into
this
new
one
and
except
the
logic,
the
logic
I
have
to
find
another
place
for
it,
and
so
that's.
B
C
B
B
B
D
For
me,
I
was
doing
kind
of
getting
ready
with
prs
and
getting
things
merged
for
the
1.4
release.
One
of
the
things
that
you
can
do
with
the
1.4
release
is
pass
feature
flags
when
you
install
valero,
so
whether
all
ads,
when
we
add
new
features
in
beta
stages,
we
added
behind
a
feature
flag
and
we
didn't
have
a
way
to.
Though
we
had
support
for
feature
flags.
We
didn't
have
a
way
for
it
to
be
expressed
on
the
cli.
D
So
I
added
some
support
to
do
that
and
then
other
the
work
on
getting
csi
integration
ready
to
go
was
something
that
I
did
last
week
and
then
I
also
set
up
the
set
up
a
github
actions
job
to
build
and
push
our
images
for
the
csi
plugin,
so
we
were
using
trav
or
we
use
travis
to
do
this
for
the
other
repos.
Now
we
because
travis
is
probably
end
of
life
or
end
of
support.
We
want
to
move
to
github
actions.
We
got.
I
got
this
done
for
the
csi
plugin.
D
D
E
F
E
D
E
I
think
the
aws
plug-in
recently
merged
thing
to
build
arm
as
well,
that
one
may
be
the
only
one,
that's
the
oddball,
but
we
can.
We
can
talk
about
that
separately,
but
yeah
for
the
the
one
five
time
frame.
I'd
I'd
love
to
help
you
tackle
this,
but
it
sounds
like
since
we've
got
since
the
valero
csi
plug-in
didn't
really
have
images
being
generated
anyway.
It
was
a
good
one
to
start
with.
C
Yeah
we've
sort
of
alluded
to
it
already,
but
just
to
make
sure
everyone's
clear,
so
we
released
the
1.4
beta
the
first
beta
on
friday,
so
that
is
out
there.
The
container
images
are
published.
So
if
you
go
to
the
github
release
that
jonas
says
up
here,
it'll
have
all
the
information
on
where
to
get
the
downloads,
how
to
upgrade
and
and
all
the
changes
that
are
in
there.
So
we're
definitely
super
excited
to
get
this
out
and
into
the
hands
of
everybody.
It's
got.
C
The
csi
snapshot
support
in
beta
form
that
we've
been
talking
about
a
lot
and
that's
kind
of
the
the
big
feature
that
we
have
in
here,
but
we
have
a
whole
bunch
of
other
good
stuff.
We
have
custom
ca
support
thanks
to
dylan
and
sam
from
red
hat
and
backup
progress,
reporting
and
a
bunch
of
other
features
in
here
as
well.
So
definitely
check
it
out.
G
E
Nolan
yeah
so
last
week
was
mostly
trying
to
get
the
1.4
beta
out,
so
I
was
on
speedy
support,
but
as
a
shisha
were
too,
I
was
a
little
bit
behind
on
the
github
issues,
so
I'm
going
to
try
to
help
him
get
those
answered,
but
this
week
is
going
to
be
a
lot
of
blog
and
docs.
Writing.
There's
a
few
issues
that
are
for
writing
some
docs
that
we
want
to
get
out
with
the
1.4
release,
but
they
weren't
really
blocking
the
beta.
E
I'm
also
going
to
catch
up
on
some
reviews.
I'd
really
like
to
take
a
look
at
the
restore
hooks
document
that
mark
campbell's
put
up
and
also
I'll,
be
testing
out
the
complete
1
1
4
beta
package,
just
because
this
will
be
the
first
time
we've
gotten
to
sit
down
and
test
it
as
a
whole.
I
know
steve
has
done
some
testing.
E
I
saw
an
issue
this
morning
around
some
formatting
of
the
valero
backup
describe
command,
but
definitely
invite
folks
to
play
around
with
that.
If
they
can,
the
csi
feature
will
be
between
behind
a
feature
flag.
So
we
don't
anticipate
that
to
affect
normal
operation.
If
you're
not
using
it,
but
we
still
invite
folks
to
play
around
with
the
beta
and
if
you're
interested
in
csi
support
on
your
platform
definitely
give
this
a
shot
on
your
non-production
environments
and
that's,
I
think,
that's
all.
I've
got.
E
I
was
intending
to
put
together
a
wiki
page.
We
had
recently
activated
the
github
wiki.
I
was
intending
to
put
together
a
little
wiki
page
with
a
maybe
a
table
of
the
versions
I
was
intending
to
test
and
the
platforms.
So
not
not.
Every
major
provider
has
gotten
their
csi
driver
out
there.
Yet
I
know
I'm
following
an
issue
with
google
that
they
haven't
brought
their
csi
driver
with
the
beta
snapshotting
apis
up
to
their
stable
release.
A
D
A
H
Hey
everyone,
so
we
received
a
few
feature:
requests
in
github
that
I'll
be
kind
of
sharing
on
the
screen
here
around
restoring
rustic
volumes
by
default
so
effectively,
including
all
volumes
to
be
restored
by
rustic
and
then
providing
mechanisms
for
exclusion,
but
still
retaining
the
same
feature.
Functionality
and
behavior.
That's
expected
for
existing
users
of
the
rustic
plug-in
where
in
by
you,
would
exclude
the
volumes
to
be
restored
by
default
and
then
annotate
those
volumes
you'd
like
to
include
so
jonas.
Is
it
all
right?
If
I
share
my
screen
here.
H
It's
related
to
issue
605
and
I
believe
2375,
but
at
the
bottom
of
the
issue,
you'll
see
we've
provided
links
to
the
topic
in
the
agenda,
as
well
as
a
detailed
product
requirement,
I'm
going
to
click
through
onto
the
detailed
product
requirement.
This
is
just
a
10
000
foot
view
of
what
the
requirement
is.
H
I
should
note
that
in
some
of
these
cases,
though,
some
of
the
flags
are
going
to
be
example,
flags
that
may
not
be
what
we
actually
use
in
the
definition
when
it's
complete,
but
this
is
just
to
kind
of
illustrate
what
the
experience
will
look
like.
So
after
a
user
has
back
opted
into
the
backup
case,
2
would
be
essentially
the
new
behavior,
which
is
to
back
up
the
volumes
with
rustic
by
default.
H
H
So
if
there
are
any
questions
on
the
10,
000
foot
view
of
the
requirements
spec
or
if
you
have
any
feedback
to
share
or
any
questions
about
what
the
experience
could
look
like,
please
feel
free
to.
Let
us
know
we're
very
interested
in
hearing
back
from
everyone,
and
we
are
planning
to
work
on
this
and,
as
part
of
the
1.5
release,.
I
I
had
a
kind
of
a
high
level
question
about
this,
and-
and
this
is
kind
of
related,
because
right
now
we're
using
it
where
we
we
have
it,
give
it
giving
users
a
choice
to
either
use
rustic
or
snapshot
volumes
so
and
with
the
current
behavior.
That
means
you
know
for
rustic
we
put
the
annotation
on
there
and
then
and
if
we're
not
putting
annotation,
then
the
behavior
is
snapshotting,
and
one
thing
I
was
realizing
was.
I
There
may
be
a
case
for
a
third
third
situation
of
explicitly.
I
don't
want
to
back
this
up
either
be
a
snapshotting
or
rustic,
I'm
not
sure
if
that
fits
into
this
kind
of
redesigns
game,
where
you
know,
if
you're
opting
in
you
know
saying
rustic
is
the
default
now
and
but
there's
really
two
choices.
If
you
don't
want
to
use
rustic
either
you
want
to
use
the
snapshotter
or
you
want
to
leave
the
volume
alone
and
not
back
it
up,
and
is
that
going
to
be
explicit
in
this
as
one
possibility.
E
J
I
So
we
want
the
kubernetes
resource
to
be
backed
up
and
restored,
but
not
actually
copying
the
image
bits
or
snapshotting
and
for
nfs
it's
fine,
because
there's
no
volume
snapchatter,
so
that
that's.
But
we
realized
that
in
our
for
ebs
volumes,
for
example,
when
we
were
trying
to
just
move
the
definition
to
a
new
cluster,
we're
actually
snapshotting
and
because
there's
not
really
a
way
to
tell
valero
back
this
resource
up
but
don't
snapshot.
It.
C
Yeah,
so
we
don't,
we
don't
really
have
a
way
to
control
that.
I
think
it's
probably
a
separate
feature.
I
mean
because
that's
you
know
you,
you
can
control
with
the
annotations
whether
or
not
to
get
a
rustic
backup.
So
it
sounds
like
the
missing
piece
is
just
an
annotation
to
control,
whether
to
take
a
snapshot,
if
you're
not
taking
a
rustic
backup,
and
there
is
a
volume
snapshotter
that
could
right.
I
Right
and
again
that
doesn't
really
belong
to
this
change.
I'm
just
thinking
in
the
context
of
this
change
because
we're
because
right
now
you
have
an
annotation
for
rustic
or
if
there's
no
annotation,
you
do
a
snapshot,
and
this
change
might
reverse
that.
So
you,
you
know,
no
annotation
means
rustic.
Annotation
says
I
skipped
this,
which
means
volume
snapshotter,
but
I
I'm
just
wondering
in
the
in
this
do
not
back
up
annotation.
C
I
D
D
A
K
C
So
the
backup
progress
feature
that
we
that
we're,
including
in
1.4,
doesn't
have
any
way
for
plug-ins
to
indicate
progress.
I
mean
we're.
You
know
we're
basically
taking
an
approach
of
trying
to
count
the
number
of
kubernetes
resources
that
we're
going
to
back
up
and
then
providing
updates
on
how
many
of
them
have
been
backed
up
so
far,
yeah.
K
C
E
K
C
Would
just
just
to
close
the
loop
on
that?
I
would
say
if
you're
interested
in
that
you
know,
definitely
file
an
issue
with
the
kind
of
the
enhancement
request
and
we
can.
We
can
go
from
there
and,
if
you're
interested
in
working
on
a
design
dock
for
it
we'd
be
happy
to
kind
of
help.
You
work
through
that.
K
K
We
have
the
plugin
for
our
c
store
engine,
so
we
have
backup.
We
have
two
crs
custom
resources,
basically
cisco,
backup
and
c
store,
restore
the
two
crs.
Are
there
where
we
update
the
progress?
I
mean
the
plugin
updates
the
progress.
So
if
these
details
can
be
synced
to
the
okay,
the
the
main
need
of
this
progress
is
like
for
others
to
know
that
what
is
happening
so
and
at
the
same
time
like
at
the
restore
side,
we
won't
be
knowing
this
crs,
so
we
don't
know
how
much
the
backup
size.
K
The
way
currently
bsl
controller
syncs
few
of
the
pvb
kind
of
things
to
is
the
object,
storage
and
then
they
again
sync
back
to
the
destination
side
regularly
regular
intervals
right?
So
is
there
a
way
to
like?
Okay?
These
are
the
other
few
crs.
I
want
to
keep
turning
into
object
and
keep
syncing
other
destination
side.
C
C
K
K
So
the
third
one
I
have
is
on
the
finalizers
for
backup
plus.
Basically,
this
is
about
again
coming
to
the
opening
bsc
store
the
valet
clicking,
so
it
provides
incremental
nature
due
to
the
zfs.
In
the
background
it
is
using
so
the
the
backups
which
happens
or
incremental,
and
so
the
one
is
dependent
on
other
ones.
So
it's
like,
even
if
you
have
something
broken
in
between
correlated
in
between
by
user,
creating
the
delete
volume
request
on
that
kind
of
cr.
D
D
D
G
K
K
C
K
Yeah,
so
actually
we
have
an
operator
also
over
this
year's.
Basically,
let's
say:
there's
a
finalizer
on
the
oliver
backups
here.
The
point
is:
when
we
have
this
great
volume
request
comes
in
the
data
can
get
deleted
on
the
object,
storage
right.
So
I
want
to
I
mean,
I'm
thinking
of
as
a
production
tool.
K
J
C
And
if
so,
not
try
to
start
deleting
things
from
object,
storage,
perfect!
Exactly
that's
a
that's
a
requirement!
I
I
got
you
because
we
don't.
We
don't
actually
delete
this,
try
to
delete
the
cr
until
the
last
step
of
the
deletion
process,
but
by
that
point
we
may
have
actually
deleted
all
the
underlying
data.
K
L
This
is
dylan.
I
just
briefly
wanted
to
mention.
I
want
to
introduce
another
person,
I
call
alei
patel
from
red
hat.
He
he
has
been
working
with
me,
scott
and
sam
in
the
past,
but
sam
has
moved
on
to
a
different
project,
so
elai's
been
working
with
us
in
the
next
coming
months,
and
I
think
we
mentioned,
unlike
a
couple
of
community
calls
back
one
of
the
big
features
that
we're
looking
to
start
working
on.
C
L
C
A
All
right
with
that,
we,
I
think,
we're
at
the
end
of
the
the
discussion
topics.
That
means
that
the
we
are
at
the
end
of
the
meeting.
We're
gonna
do
a
bunch
of
contributor
shout
outs
next
week,
we're
gonna
collect
all
of
them
for
the
next
release
and
stuff
like
that.
So
thank
you.
All
for
joining
today
have
a
wonderful
rest
of
the
week
and
see
you
all
next
week
have
a
good
one.