►
From YouTube: TGI Kubernetes 080: Velero 1.0
Description
Show notes: https://github.com/heptio/tgik/tree/master/episodes/080
00:00:00 - Welcome to TGIK!
00:04:00 - Week in Review
00:19:00 - Velero and why it was renamed from Ark
00:21:00 - Let's get started, starting with the docs
00:25:00 - Creating an S3 bucket & IAM Account
00:34:00 - Run velero install
00:38:00 - Install nginx example workload
00:48:00 - Take a backup using velero backup create
00:52:00 - Look at contents of backup
01:03:00 - Create scheduled backup
01:07:00 - Install velero into cluster 2
01:09:00 - Migrate nginx app from cluster 1 into cluster 2
01:13:00 - Roadmap
Come hang out with Joe Beda as he does a bit of hands on hacking of Kubernetes and related topics. Some of this will be Joe talking about the things he knows. Some of this will be Joe exploring something new with the audience. Come join the fun, ask questions, comment, and participate in the live chat!
This week we will look at Velero (https://velero.io). Velero (formally called Heptio Ark) is a backup and disaster recovery tool for Kubernetes. We've been working on this for quite a while and it recently hit 1.0.
A
All
right
welcome,
welcome
welcome
to
yet
another
episode,
80
of
t,
GI,
kubernetes,
I'm,
Joe,
Beda
and
I'll-
be
your
host
for
today.
For
those
who
don't
know,
t
GI,
kubernetes
RT
gik
is
a
weekly
ish,
YouTube
livestream
that
we
do
where
I
or
one
of
the
other
folks
at
the
VMware
team
here.
Take
a
new
technology.
Take
something
that's
interesting
or
happening
in
the
in
the
kubernetes
world
and
play
with
it.
A
Live
oftentimes,
I,
don't
actually
know
what
I'm
talking
about,
or
at
least
I,
don't
know
the
thing
that
I'm
presenting
super
well
and
so
it's
a
good
chance
for
us
all
to
learn
together
as
we
go
through
this
stuff.
So
let
me
just
start
out
by
saying
hello
to
everybody.
That's
joining
me.
You
know
a
lot
of
times,
folks
will
say
hi
on
the
chat
here,
so
we
have
Lou
Maddy,
saying
hi
who's,
always
a
regular
murugesh
nice
to
see
you
from
Bangalore
tim
from
frankfurt.
A
Shahar,
says
hello,
fellow
llamo,
architects,
so
torn
about
the
mo
thing,
Jo's
from
London
good
to
see
ya
so
yeah.
So
the
way
that
we
usually
set
this
up
is
that
I'll
spend
the
first
10
15
minutes
going
into
what's
new
around
the
kubernetes
ecosystem.
What's
sort
of
the
news
of
the
week
and
then
after
that,
I'll
do
a
deep
dive
into
a
specific
topic.
The
project
that
we're
looking
at
this
week
is
called
Valero
and
I'll,
give
a
little
bit
about
the
history
and
the
purpose
of
that,
but
real,
quick.
A
A
So
it's
easier
to
find
stuff
in
the
episode
so
yeah
that
link
is
in
the
chat,
so
feel
free
to
give
that
a
try.
What
have
I
shared
it
I
shared
it
right.
Yes,
okay,
so
folks
that
are
signed
in
can
go
ahead
and
start
playing
around
there,
so
yeah.
So
let
me
say:
hi
to
some
more
folks.
We
have
Kenneth
from
Delaware
M&A
from
Strasbourg,
it's
its
yak
from
Bangladesh
I,
just
love
that
everybody
is
all
over
the
place.
A
It's
awesome,
Lorraine
Hart
from
Germany
Nolan
who's
on
the
Valero
team
and
Steve
is
also
on
the
viler.
The
Valero
team
Steve's
in
Boise,
Nolan
I,
don't
even
know
where
you
live
man,
oh
Indiana
I
knew
that
I
knew
that,
but
I
didn't
know
that
Syed
Erik
Saeed
from
London
Erik
from
Belgium
Mahmood
from
New
York
new
video
quality
is
great.
Really
ok.
A
Was
it
bad
before
I
hope
not,
but
always
trying
to
improve
this
stuff
Martin
from
the
Netherlands
Matt
from
Madison
awesome,
awesome,
yeah,
I,
know
like
and
also
you
know,
a
huge
shout-out
to
those
who
are
calling
in
from
all
over
the
place
it.
You
know.
1
p.m.
Pacific
time
seems
like
a
great
time
to
do
something
like
this,
but
it's
like
either
mid.
You
know
middle
of
the
night
or
weekend
in
the
rest
of
the
world.
So
it's
it's,
maybe
not
the
best
time
live
and
learn
on
this
stuff.
A
Alright,
let
me
go
ahead
and
get
started.
I'm
going
to
jump
right
in
so
the
big
news
this
week
is
kubernetes.
1.15
was
released.
There's
a
blog
post
on
the
kubernetes,
a
blog,
our
very
own,
George,
helped
to
pull
together
all
the
stuff
for
the
blog
post
and
and
I
just
I
want
to
hit
some
of
the
highlights
real
quick
here,
I
mean.
Obviously
you
can
go
through
this.
You
can
go
through
the
release
notes,
which
is
also
a
huge
another.
A
Big
task
of
you
know
that
we
do
when
we
pull
together
these
releases,
big
stuff
that
we're
actually
seeing
here.
So
there's
a
lot
of
work
going
on
in
the
part
of
the
project
that
we
called
API
machinery,
which
is
essentially
the
core
sort
of
API
work
and
storage
work
in
kubernetes.
You
know
kubernetes
at
its
core
and
I'm.
You
know
happy
to
give
a
little
bit
more
into
this
at
at
some
point.
A
But
kubernetes
at
its
core
is
a
database
with
an
API
sort
of
layer
wrapped
around
it
and
and
so
we're
continuing-
and
you
know,
I
should
say
we
as
in
the
project,
is
continuing
to
to
enhance
what
that
API
layer
does.
A
lot
of
these
things
are
around
custom,
resource
definitions,
and
so
that's
ways
that
you
can
extend
the
built-in
objects
in
kubernetes,
add
new
objects,
and
so
we're
continuing
to
to
do
interesting.
Stuff.
A
There
I'll
go
down
like
some
like
one
of
the
things
biggest
problems
that
we
have
with
CR
DS
customer
resource
definitions
is
being
able
to
version
these
things
and
upgrade
so
now.
There's
a
wet
web
hook.
Conversion
so
that
you
can
actually
start
doing
version
upgrades.
How
do
you
get
schema
information
react
to
that
dynamically.
A
lot
of
work
going
on
around
that
you
know.
Do
we
store
fields
that
we're
originally
part
of
the
schema?
A
So
a
lot
of
defaulting
like
if
you
write
something
and
you
don't
include
a
field-
is
there
a
default
value
for
that
field?
Does
that
get
recorded
all
sorts
of
really
really
interesting
stuff
going
on
here?
Another
area
where
the
VMware
folks
specifically
have
been
putting
a
lot
of
time
is
a
sync
cluster
lifecycle.
The
big
project
in
cig
cluster
life
cycle
is
cube,
ADM
or
cube
admin,
and
it's
essentially
a
part
of
the
bootstrapping
process
for
getting
a
cluster
up
and
running.
A
If
you
have
a
bunch
of
machines,
cube
admin
makes
it
really
easy
to
get
kubernetes
up
and
running
once
you
have
a
bunch
of
machines
and
it's
both
a
full
experience
with
cube
admin,
I'm
enjoying,
but
also
a
toolbox
that
lets
you
do
a
bunch
of
other
stuff,
and
so
the
h.a
capabilities
of
cube
admin
are
getting
easier
and
more
stable,
and
so
that's
actually
really
cool
and
there's
also
some
new
capabilities
around
certificate
rotation,
which
is
which
is
really
cool.
The
thing
I'm
most
excited
about
is
that
cube.
A
Admin
now
actually
has
a
logo
which
is
an
atom,
cube,
atom,
cube
admin,
and
so
that's
super
cool,
so
I
want
to
like
laser-cut
this
at
some
point
soon,
Salem
Andy
says
anyone
seen
Joe,
I
refresh
the
page
Oh
No.
Are
you
not
getting
a
stream
l'm
adi,
so
other
folks
are
seeing
a
stream
okay.
So
hopefully
it's
as
something
that
you
can
that
you
can
actually
get
to
okay,
so
yeah.
A
One
of
the
things
that's
interesting
out
of
this
is
that
again,
there's
there's
a
migration
from
to
actually
make
this
stuff
be
more
and
more
pluggable
over
time,
and
so
my
entry
storage
plugin.
So
what
is
this
I
haven't
actually
seen
this
one
so
so
being
able
to
actually
have
some
stuff
in
tree
I.
Don't
I'm
gonna
have
to
look
look
into
that.
A
What
does
it
mean
by
entry
there?
But
okay,
but
then
I
know
a
lot
of
works
been
going
on
there,
and
one
of
the
big
things
coming
out
of
that
is.
Is
the
ability
to
clone
volumes
and
that's
you
know,
starting
to
treat
volumes
and
actually
surface
more
of
the
sort
of
cloud
capabilities
and
in
kubernetes
kubernetes
way,
yeah.
A
So
the
delay
in
the
chat,
the
I'm
now
using
this
like
ultra-low
latency
setting
in
in
YouTube
and
I
hadn't,
used
that
before,
because
it
used
to
be
that
when
you
enabled
that
you
weren't
able
to
actually
DVR
and
actually
rewind
and
fast-forward
and
stuff,
but
now
you
can
actually
DVR
and
do
the
ultra-low
latency.
So
that's
actually
really
cool,
so
they're
they're.
Definitely
improving
that
all
right.
So
you
know,
first
of
all,
I
just
want
to
say
like
like
getting
a
kubernetes
release
out
is
a
huge,
huge,
huge
effort.
A
There's
like
I,
think
the
number
here's
38
people
that
actually
all
get
together
to
make
this
happen.
I
want
to
say
it's
a
well-oiled
machine,
but
I
think
there's
there's
places
where
we
could
still
apply
some
oil.
But
like
it's
a
lot
of
work-
and
you
know
huge
shout
out
to
the
team
there
and
it
was
Claire
Lawrence
who's
who's.
That
pivotal
was
the
release
lead
this
time,
and
so
it's
always
it's
always
interesting
scene.
A
You
know
seeing
the
team
come
together.
You
get
a
release
out
like
that,
all
right,
and
so
there's
some
other
notes
in
here.
Around
sort
of
volume,
cloning
alpha
future
of
CR,
DS,
there's
a
blog
post
on
that
and
then
there's
a
there's,
a
set
of
things
that
are
being
deprecated,
and
so
it's
always
worthwhile
to
actually
go
through
some
of
this
stuff
and
actually
see
you
know
if
any
of
this
stuff
is
actually
going
to
hit
you
and
bite
you
in
the
butt.
A
A
lot
of
these
things
is
just
changing
some
naming
and
how
we
actually
sort
of
address
certain
types
of
resources,
that's
worth
digging
into,
so
that
those
are
in
the
notes
there.
Okay,
so
the
next
thing
that
I
think
is
really
interesting,
is
H,
a
proxy
hit
2.0,
which
is
a
big
milestone
and
I.
Think
really
interesting
thing
here
and
I
think
this
is
fascinating.
A
Is
that
it
now
there
is
a
kubernetes
ingress
controller
that
is
part
of
the
H,
a
proxy
project
hosted
at
the
H,
a
proxy
github
org
with
documentation
next
to
everything
else.
That
is
super
super
super
cool,
I,
love,
seeing
you
know,
projects
like
H,
a
proxy.
You
know
really
embrace
adapting
and
to
kubernetes
making
that
be
a
first-class
part
of
their
experience
and
I
think
you
know
this
is
one
of
those
things
that
I
think
is
interesting
from
from
a
ecosystem
point
of
view.
A
It's
like,
should
you
know
our
operator
is
going
to
be
something
like
you
know,
which
is
a
so
ingress
is
kind
of
a
little
bit
of
an
operator
type
of
thing.
Our
operator
is
going
to
be
this
thing
that
exists
outside
of
the
projects,
or
are
they
going
to
be
co-developed
with
the
projects
or
over
time
where
they
can
actually
be
built
into
the
projects
themselves?
And
so
it's
it's
fascinating.
To
see.
Projects
like
H
a
proxy,
really
bring
kubernetes
support
to
be
a
first-class
thing,
very,
very
cool.
A
A
This
is
Bjorn
here,
who's
in
I,
don't
know
where
Bjorn
is
working
for
Schenker,
so
I'm
gonna,
guess
based
on
name
and
then
maybe
maybe
he's
in
Germany
talking
about
how
he
set
up
his
sort
of
CI
CD
pipeline,
or
at
least
the
start
of
it
and
I
think
this
is
awesome,
because
it's
in
excruciating
detail,
you
know
not.
Everybody
is
gonna,
make
the
same
choices
with
something
like
this,
but
I.
A
A
Let's
see
yeah
so
and
George's
saying
that
if
folks
do
want
to
get
involved
in
the
release
team,
there's
always
looking
for
help,
and
a
lot
of
this
is
coding.
A
lot
of
it
is,
is
helping
to
sort
through
features
and
requests.
There's
a
lot
of
process
type
of
stuff
pulling
together
a
release,
notes
a
lot
of
interesting
stuff
going
on
there,
alright,
so
here's
another
blog
article.
This
is
one's
from
Danielle
libel
and
I,
want
to
say
going
into
a
lot
of
really
cool
sort
of
like
optimizing.
A
Your
cute
control
game,
everything
from
command
completion
to
using
cube
control,
the
lookups
formats,
custom,
outputs.
You
know
switching
between
clusters
and
namespaces.
What
is
he
use?
Does
he
is
a
Cuban
s?
He
talks
about
how
these
things
work
and
then
my
guess
is
I
wonder
if
he
yeah
cube
contacts
is,
is
a
tool
that
actually
makes
switching
this
stuff
a
lot
easier.
We
should
have
built
this
into
cube
control.
A
We
probably
should
build
it
into
keep
control,
so
that's
kind
of
cool,
and
so
a
lot
of
really
cool
tips
and
tricks
going
on
here,
which
is
pretty
cool.
They're
like
talking
about
plugins
and
we're
starting
to
see
the
crew
project
for
being
able
to
search
and
install
plugins
is
pretty
cool
too,
should
probably
do
it.
That
would
be
fun
to
actually
do
an
episode
on
that
maybe
write
a
plug-in.
Alright
next
thing
is
a
bonsai
cloud.
I
think
this
is
like
is
that
the
right
link
for
that
yeah?
Okay,
there's
just
the
github
thing.
A
This
is
essentially
a
toolkit
for
managing
vault
with
with
kubernetes.
So
there
looks
like
there's
a
lot
of
really
cool
stuff
here.
I
haven't
actually
played
with
this,
but
you
know
vault
sort
of
has
its
own
way
of
doing
things,
and
you
know
having
a
toolkit
to
sort
of
bridge
the
gap
between
kubernetes
and
Baltus
is
pretty
darn
cool,
and
so
this
is
the
bonsai
cloud
folks
putting
this
together
and
then
and
then
the
last
thing
I
wanted
to
mention
our
very
own
project.
Contour.
A
A
What
I
love
is
that
there's?
You
know
a
lot
of
folks
from
from
the
community
of
sort
of
putting
features
into
into
contour,
which
I
think
is
really
cool.
So
it's
good
to
see
it's
good
to
see.
People
go
through
and
you
know
scratch
their
itch
and
and
make
this
stuff
work.
So
we
had
and
then
Nick
Young,
just
joined
VMware
and
he's
he's
helping
out
working
on
their
project
they're.
Also
so
big
news
here
is
session.
A
Affinity
is
supported
being
able
to
have
services
with
external
name
so
that
you
know
contour
essentially
becomes
a
reflector.
At
that
point,
let's
see
a
couple
of
deprecations
and
removals
that
some
clarity
on
some
stuff,
h-2b
tcp
idle
timeouts,
continue
to,
like
you
know,
see
this
in
production
to
make
sure
that
it
work
and
well
and
that
were
we're
taking
problems
like
memory
leaks
and
such
and
really
making
sure
we
fix
those
ipv6
improvements.
So
a
lot
of
good
stuff
there
to
take
a
look
at
so
yellow
Mattie's,
saying
like
I
bought
with
kubernetes.
A
A
We
did
do
this
one
there
we
go
yes,
November
16th,
but
yeah
I.
You
know
I'd
love
to
go
back
and
find
like
the
episodes
that
we
should
update
and
redo,
because
there's
always
there's
always
new
stuff
going
on,
and
it's
like
these
things
are.
You
know
we
live
in
a
super
fast
moving
world.
Alright,
so
let's
see
so
our
15
minutes
in
alright.
A
So
I
want
to
start
talking
about
Valero
and
give
it
a
little
bit
of
history
behind
Valero
and
then
we're
gonna
sit
and
actually
play
with
it
and
see
if
we
can
make
it
work
for
us.
So
Valero
was
the
project
that
we
started
at
hefty,
oh
and
at
the
time
we
called
it
hefty
Oh
Ark,
as
part
of
you,
know,
moving
and
rationalizing
a
bunch
of
stuff
into
VMware.
It
got
renamed
from
hefty
Oh
Ark
to
Valero,
so
no
VMware
in
front
of
it.
A
Just
plain
old,
Valero
and
and
the
goal
of
Valero
is
to
be
able
to
back
up
my
grades
and
essentially
clone
applications
and
cluster
data
right.
How
do
you
actually
sort
of
in
a
principled
way,
look
at?
What's
going
on
inside
of
a
cluster,
be
able
to
capture
that
and
then
do
interesting
stuff
with
it?
The
the
interesting
thing
about
Valero
is
that
doesn't
just
capture
application
state.
It
also
actually
can
go
ahead
and
coordinate
with
the
underlying
cloud.
The
underlying
infrastructure,
if
necessary,
to
be
able
to
capture
capture
persistent
volumes.
A
Also,
it's
going
to
be
interesting
as
we
get
first-class
support
of
things
like
cloning
and
snapshotting
in
the
CSI
exposed
up
through
kubernetes.
You
know,
I'm
sure
we're
going
to
be
looking
to
make
sure
that
Valero
can
take
the
fullest
advantage
of
those
two.
So
Lou
man
he's
asking:
why
did
we
rename
an
arc
was
a
killer
name?
Well,
so
here's
the
deal.
So
there
was
this
other
company
out
there
that
had
a
backup
product
aimed
at
consumers
called
arc.
A
So
that's
why
it
got
renamed
to
Valero,
so
Valero
is
a
sailboat
in
Spanish,
so
you
know
we're
keeping
the
the
nautical
theme
here,
at
least
alright.
So
the
original
sort
of
idea
behind
Valero
is
early
early
early
on,
like
probably
the
first
month
that
we
formed
hefty.
Oh
one
of
the
things
that
we
did
is
we're
like.
Let's
just
start,
you
know
getting
in
the
trenches
with
some
potential
customers
help
them
be
successful
with
kubernetes,
and
as
we
do
that,
we'll
really
understand.
A
What
are
the
problems
that
they're
facing,
and
so
one
of
the
first
engagements
that
we
did,
they
asked
us
hey.
How
can
I
go
through
and
they
were
using
AWS
and
make
sure
that
when
I
create
a
volume
via
kubernetes,
it
gets
tagged
in
a
certain
way
with
using
AWS
tags,
and-
and
so
you
know,
the
fascinating
thing
is.
Is
that
anytime?
Somebody
asks
you
for
a
feature
if
it's
not
immediately
obvious,
you
want
to
ask
why
and
so
I
asked
them
I'm
like
hey.
So
why
do
you
want
to
actually
tag?
A
What's
the
scenario
where
you're
tagging
the
underlying
EBS
volumes,
let
me
go,
we
have
this
script
that
goes
through.
It
takes
automatically
take
snapshots
of
EBS
volumes
if
they
tagged
a
certain
way
and
we
want
to
make
sure
that
our
kubernetes
volumes
are
included
with
that,
and
that
was
like
an
aha
moment
because
we're
like
nobody's
looking
at
backup
in
disaster
recovery
for
kubernetes.
This
is
probably
two
and
a
half
years
ago,
something
like
that
and
so
arc
slash.
Bolero
started
off
as
a
EBS
volume
snapshot.
A
So
essentially,
how
could
we
go
through
start
doing
EBS
volumes,
but
do
it
in
a
way
that
was
sort
of
kubernetes
centric
and
was
something
that
could
could
stretch
across
just
eat
AWS
and
over
time
we
looked
at
sort
of
the
larger
problem
around
disaster
recovery
and
at
how
we
wanted
to
manage
that
and
that
ended
up
turning
into
Valero
over
time.
But
Steve
Cris
is
the
who's.
A
The
lead
on
the
project
now
was
a
was
was
the
the
first
author
on
this
also,
so
he
was
there
all
the
way
from
sort
of
the
AWS
snapshot
or
days
when
we
were
getting
started
off
with
this
stuff
all
right.
So
let's
go
ahead
and
start
getting
started
and
if
you
hit
documentation
here,
there's
a
there's
age
where,
where
is
it
a
quick
start
within
cluster?
Many,
oh,
is
that
what
we
want?
A
This
is
not
what
we
want.
I
actually
have.
It
here,
set
up
full
arrow
on
your
platform,
okay,
so
so
the
way
that
Valero
works
is
that
it
all
works
through
CR,
DS
and
so
there's
a
server
component.
That's
a
controller
that
operates
with
dis,
ER,
DS
and
then
there's
a
client
component
to
make
it
easier
to
be
able
to
manipulate
those
CR
DS.
But
as
you'll
see
as
we
start
playing
with
this,
those
CRTs
are
perfectly
accessible.
The
the
API
and
cube
control.
A
Also,
it's
just
that
the
the
Gewehr,
oh
you
I,
just
makes
it
a
little
bit
of
a
smoother
experience
to
be
able
to
manipulate
these
things,
and
then
you
set
up
Valero
so
that
it's
actually
hooked
up
to
to
your
cloud
provider
so
that
it
can
actually
then
call
through
to
actually
make
the
snapshots
happen.
So
you
have
to
have
some
sort
of
access
to
the
underlined.
A
Let's
see
so
Alemanni
says
that
chris
Nova's
talk
on
backing
up
and
recovering
across
providers
with
ARC
was
amazing.
Should
we
add
that
yeah
add
that
to
the
notes,
if
you
have
a
quick
pointer
to
that
so
yeah,
so
the
so
you
actually
so
that
argA
actually
has
to
talk
to
the
underlying
snapshot
provider
and
then
arc
also
has
to
talk
to
a
blob
store
and
that's
where
the
actual
configuration
essentially
extracting
all
the
gamma
of
kubernetes
gets
written
into
that
blob
store.
A
You
know
some
sort
of
object,
store,
someplace
else,
and
so
some
of
this
sort
of,
if
you
want
to
play
with
this
with
mini
cube,
you
can
use
this
system
called
Mineo,
which
provides
an
s3
work
alike,
type
API
locally,
but
it's
also
supported
on
top
of
AWS,
Azure
and
GCP,
and
then,
if
you
don't
have
access
to
the
underlying
volume,
snapshotting
services,
because
not
all
environments
have
that,
then
it
also
supports
this
setup.
Using
this.
A
A
A
A
There
we
go
this
actually
links
through
all
the
different
things,
so
these
are
the
ones
that
are
built
in,
but
also
there's,
there's
a
way
that
you
can
actually
go
through
and
extend
this
with
plugins,
not
something
I
mentioned
yet,
and
so
there's
a
whole
bunch
of
plugins
that
you
can
go
ahead
and
make
work
here.
So
it's
great
to
see
these
things
continually
to
be
updated
and
have
this
be
something
that
other
folk
can
plug
into?
A
A
So,
there's
a
lot
of
options
there
in
terms
of
what
might
make
sense
if
you
actually
want
to
take
this
on
premises.
Okay,
so
this
is
where
we're
at
okay.
So
the
first
thing
we're
gonna
do
is
we're
gonna
run
Valera
on
AWS,
because
that's
what
I
have
and
you
can
get
the
the
official
release
of
velaro
here
and
put
it
in
your
path.
You
can
also
use
homebrew
I.
Think
that's
what
I'm
using
here.
Oh.
A
A
Okay
and
then
I
have
a
simple
cube:
control
I
have
a
simple
two
nodes
in
a
mass
per
cluster,
so
a
small
mini
cluster
running
in
AWS,
and
so
the
first
thing
we're
gonna
do
is
we're
gonna
create
a
bucket?
What
did
the
creep
but
location
constraint
region?
Does
that
mean
it
doesn't
actually
leave
that
region?
What
is
it
I'm,
gonna,
I,
don't
know
what
that
is.
Does
it
actually
talk
about.
A
A
The
region
where
the
bucket
resides,
but
that's
actually
people-
maybe
you
got
to
do
that
because
there's
also
region
here
I'm
a
little
bit
confused,
but
let's
just
go
ahead
and
create
it
we're
good
all
right.
So
we
now
have
a
bucket
called
TGI
Cavallaro
1
s,
3,
amazon,
AWS,
calm,
alright,
we're
up
and
running
there
and
then
and
then
ok.
So
this
is
interesting,
so
we
want
Valera
to
be
able
to
act
as
a
user
when
it
when
you
actually
use
it
right.
So
this
is
creating
the
service
counter.
A
A
That
may
be
something
that
automatically
works,
but
I
think
you
know
in
this
case,
creating
a
specific
users
for
Valero
probably
makes
sense,
and
so
that's
what
the
instructions
here
have
you
do
so
we're
gonna
go
through
and
actually
run
this,
so
we're
gonna
create
a
new
new
user.
Oh
I
already
had
one.
Oh,
let
me
clean
this
up.
A
A
A
The
next
thing
we're
going
to
do
is
we're
going
to
give
it
a
policy,
and
this
essentially
is
saying
what
this
user
is
allowed
to
do,
and
it's
allowed
to
muck
with
volumes,
create
and
delete
snapshots
and
all
that
and
then
go
ahead
and
do
a
bunch
of
s3
stuff,
but
only
in
this
specific
bucket.
That's
where
this
resource
here
is,
and
then
we're
also
allowed
to
list
what's
going
on
in
that
bucket.
So
that's
all
good
stuff,
it's
a
relatively
sane
scoped
policy.
You
might
be
able
to
scope
it
down.
A
Even
more
I
think
you
might
be
able
to
give
a
tag
constraint
here,
but
I'm
not
sure.
Okay,
so
we
created
that
and
then
we're
gonna
actually
put
that
policy
on
here
that
work.
Look
at
that
alright!
So
so
now
the
velaro
user
only
has
very
scoped
permissions,
and
now
the
next
thing
is
that
when
you
create
a
user
by
default,
it
doesn't
get
any
sort
of
password.
So
I'm
gonna
go
through
and
do
this.
Actually,
you
know
what
I'm
gonna
do
is
I'm
going
to
switch
just
to
my
face.
A
Let's
see
and
then
we're
going
to
create
a
credentials
file
in
the
local
directory,
and
this
is
something
that
valero
install
is
going
to
use
and
it's
gonna
actually
create
a
secret
and
upload
that
so
I'm
now
going
to
switch
back
I'm
going
to
create
the
credentials
file
around
that.
So
give
me
a
second
here:
I
apologize.
A
A
Here,
okay,
so
now,
I
have
one
of
these
things.
One
of
these
files
called
credentials
Valero
and
now
we're
ready
to
actually
install
Valero,
and
so
we're
gonna
go
through
and
paste
this
in,
and
so
what
this
is
actually
doing
is
it's
going
through
and
creating
the
right
gamal
putting
the
right
sort
of
configs
in
there
and
just
installing
Valero
straight
up.
A
So
like
one
of
the
things
that
that
I
would
do
it's
like
like
I,
would
like
do
this
to
the
lair
adagiamo
and
then
and
then
make
sure
that
you
know
I'm
totally
cool
with
everything.
That's
actually
happening
here
as
it
does.
It
goes
that
creates
its
no.
You
can
see
it's
creating
the
secret
and
stuff.
A
So
these
are
looking
at
like
where
do
I
want
to
put
my
config
and
like
where
do
I
want
to
you
in
a
cloud
specific
way?
How
do
I
want
to
manage
snapshots
and
then
the
actual
velaro
controller
gets
installed?
So,
let's
look
at
the
one
thing
that
I
think
is
is
interesting.
Here
would
be
the
we're
all
binding
yeah.
A
Okay,
so
we
are
good,
oh
and
here's
how
you
can
actually
use
things
like
cubed
a
so
you
don't
have
to
set
up
so
that
you're
using
rules
instead
of
instead
of
actually
uploading
a
secret
key
like
that.
So
that's
actually
pretty
cool
all
right,
and
so
then,
if
we
go
through
and
we
look
at
getting
started
here.
A
We
can
go
through
and
there
is
an
example.
Now
this
is
the
mini.
Oh,
let's
see
am
I
looking
at
the
right
one
there's
an
example,
workload
that
we
can
actually
get
started
with,
and
it's
in
the
Valero
source
code.
So
this
is
one
of
those
things
feedback
for
the
Valero
team
like.
Where
is
this
examples
right?
Do
I
have
to
get
cloned?
Something?
Is
this
something
that's
available
elsewhere?
I
think
it's
I
do
have
to
get
cloned
Valero,
just
I
think
this
was
and
here's.
A
A
A
Yeah
the
released
her
bowls
are
in
github,
but
like
okay,
so
here
we
go
so
yeah,
so
here's
the
here's,
the
actual
code
for
the
thing.
Okay,
so
you're
saying!
Oh
so
examples
are
in
the
release:
tarball,
okay,
which
I
didn't
do
because
I
used
I
used
brew.
So
I
don't
actually
have
the
release
tarball,
but
I
can
do
a
git
clone
of
this
and
I
assume
that
this
actually
has
examples.
Also,
the
engineer
nginx,
app
and
stuff
like
that:
okay.
A
On
Amazon,
this
one's
called
gp2,
so
what
we
got
to
do
is
we
have
to
come
in
here
and
we
have
to
modify
the
width
with
P
V
so
that
we
can
actually
say
where
to
go
where
to
go
your
storage
class
name
here
all
right.
So
let's
go
through
sort
of
like
what
a
minimally
stateful
app
looks
like
that
Valero's
gonna,
actually
back
up,
because
I
think
there's
some
interesting
stuff
that
you
see
come
out
of
this
okay.
So
we
have
a
name
space
which
is
nginx
example.
A
We
have
a
persistent
volume
claim,
which
is
essentially
saying.
Please
attach
me,
you
know
give
me
a
persistent
volume
that
I
can
work
with.
This
is
set
up
with
Auto
provisioning,
so
it'll
actually
create
the
persistent
volumes
just
in
time
for
us
to
be
able
to
use
I
had
to
look
up
what
the
storage
class
name
is.
We're
gonna
put
that
in
there
and
I'm
saying
I
need
this
thing
to
be
15
megabytes,
so
nothing
super
big
and
then,
and
then
here's
something
that's
really
interesting
is
that
anybody
who's
done.
A
Backup
knows
that
what
you
want
to
do
is
actually
have
your
application
as
much
as
possible
right
all
the
data
that
it
might
have
to
disk,
make
sure
that
everything
is
in
a
consistent
state
and
then
actually
do
the
snapshot
and,
if
you're
doing
stuff
with
ec2
it's
up
to
you
to
core
sort
of
like
hey
database,
make
sure
you
actually
flush
all
your
buffers,
hey,
filesystem
flush,
all
your
buffers.
Okay,
now
do
the
snapshot.
A
Okay
now
go
through
and
let
things
continue
again,
and
so
that's
a
dance
that
you
have
to
do
right
to
really
do
block
level
snapshots
in
something
like
ec2
I.
Think
a
lot
of
folks
just
go
through
and
sort
of
do
the
snapshot
across
their
fingers
and
hope
for
the
best
velaro
actually
helps
to
automate
this,
and
so
one
of
the
things
that
you
can
do
here
is
that
you
can
tell
it
to
actually
run
a
command
in
one
of
your
containers,
both
before
and
after
it
actually
takes
a
backup.
A
You
need
to
do
to
be
able
to
get
yourself
ready
to
snapshot.
So
in
this
case,
it'll
call
FS
freeze
with
the
freeze,
parameter,
take
the
snapshot
and
then,
after
the
snapshot
is
has
has
returned,
it'll
actually
unfreeze
it,
and
so
this
is
a
really
really
cool
feature
that
it
actually
takes
care
of
making
sure
that
everything
has
quest
and
you've
written
everything
down.
And
so
that's
done
with
annotations
on
the
deployment.
A
And
then
let's
see
these
actually
know,
these
are
annotations
on
the
pod
that
actually
get
passed
through
from
the
deployment
and
then
and
then
there's
an
extra
image
being
being
run.
Just
so
that
it's
around
to
be
able
to
do
that
and
you'll
see
that
this
is
running
as
a
privileged
thing
to
be
able
to
do
the
snapshot,
hopefully
over
time,
we'll
be
able
to
make
these
things
B
be
less
privileged
and
then
there's
a
service
with
type
equals
load
balancer
to
be
able
to
get
at
this
thing.
A
A
A
A
A
A
A
A
A
Get
our
log
and
gen-x
so
you
can
see
our
access
log
here,
it's
not
getting
lost
so
so
far,
it's
like.
Why
do
you
usually
port
forward
instead
of
going
to
the
load,
balancer
eggie,
the
load,
balancer
head
and
actually
initialized?
Yet
it
takes
Amazon
quite
a
while
to
be
able
to
actually
get
it
to
act
together
and
get
the
load
balancer
configured
and
so
I'm
moving
too
fast
for
it,
it's
probably
actually
up
and
working
now.
A
If
we
do
the
service,
let
me
copy
this:
let's
see
if
this
is
actually
up
and
running
now:
okay,
yeah!
So
now
it's
up
and
running
through
the
load.
Balancer
I
love
k-9s
for
debugging
accessing
pods
and
kubernetes
I
haven't
played
with
k-9s
I'll.
Take
a
note
of
it.
I
don't
think
I've
used
that
one.
Yet
we
have
some
other
fun
stuff
coming
sometime
soon,
which
is
which
is
going
to
be
super
cool
for
those
two.
Ok,
so
now
I
can
go
ahead
and
I
can
hit
this
if
I
go
through
and
my
exec
back.
A
A
And
yeah,
so
that's
super
cool.
Ok!
So
now
we
have
that
I
like
it,
but
you
know
it's
like
hey.
If
my
pod
dies
I
want
to
make
sure
you
know
that
that
access
log
doesn't
get
lost
and
that
access
log
could
be
a
database,
it
could
be
other
files.
You
know
there
could
be
all
sorts
of
different
things,
so
this
is
just
a
super
simple
thing
to
actually
look
at
here.
So
now,
what
we're
going
to
do
is
we're
going
to
actually
take
a
back
of
this,
and
so
when
we
run.
A
So
Valero
back
up
create
so
we're
creating
a
backup,
called
nginx
backup
and
we're
making
sure
that
we
include
the
namespace
nginx
example,
and
so
this
thing
says:
okay,
you
know
the
backup
request
has
been
submitted
successfully
and
so
now
what
we
can
do
is
we
can
go
through
and
we
can
do
we
can
like
do
it
describe
of
that
backup
and
here's
a
bunch
of
information
that
we
have
about
it.
So
so
it's
a
and
this
is
using
the
velaro
command.
A
It's
called
nginx
backup
it
has
some
labels
in
it
in
terms
of
where
the
storage,
location
and
stuff
like
that
is,
it
says,
completed
so
there's
a
phase
that's
going
on
here,
which
namespaces
are
included
excluded,
so
the
parameters
coming
in
and
then
there's
a
TTL
on
here
of
720
hours.
So
these
things
will
automatically
get
deleted
after
a
certain
amount
of
time
and
then
there's
information
about
how
long
it
took.
So
it
took
us
two
seconds
to
actually
take
the
backup
relatively
quick
super,
quick
all
right.
A
So
what
happened
in
those
two
seconds
while
we
were
taking
this
backup?
Okay,
so
here's
what
happened
Valera
on
the
server
so
I
talked
to
the
API
server,
did
a
walk
of
all
the
resources
in
the
cluster
captured
those
things
I
think.
Actually
it
did
a
walk
of
the
resources
for
the
application,
plus
any
cluster
level
resources
that
it
needed,
captured.
Those
things
put
him
into
a
tarball
uploaded
that
tarball
to
s3.
A
While
it
was
doing
that,
it
also
found
that
we
were
using
this
PVC.
This
persistent
volume,
this
EBS
volume
behind
the
scenes
it
did
an
exec
into
the
workload
into
the
nginx.
Did
the
FS
freeze
told
EBS
to
take
a
snapshot
and
then,
after
that
snapshot
was
done.
It
then
actually
called
FS
freeze
to
unfreeze
the
the
filesystem.
So
that's
the
stuff
that
went
on
here
and
if
we
start
looking
in
the
AWS
console
we'll
start
seeing
some
evidence
of
what
happened.
A
So
the
first
thing
we
can
do
is
we
can
look
at
snapshots
and,
oh
look.
We
have
a
snapshot
here,
called
kubernetes
dynamic,
see
you
know,
blah
blah
blah
some
random
number,
but
you
know
you
can
see
that
there's
information
here
in
tags
in
in
EBA
in
AWS,
talking
about
sort
of
like
you
know
which
cluster,
which
PVC
name.
A
A
Okay,
then
Steve
says
you
can
also
use
Valero
backup
download
to
actually
get
the
tar
ball
as
well,
and
so
here's
what
we
can
see
is
that
it
actually
exploded
and
there's
a
directory
structure
here
that
mirrors
a
bunch
of
the
stuff
that
you'll
see
inside
the
inside
of
the
kubernetes
cluster
that
you're
backing
up.
So
if
we
look
at
explode,
there's
metadata,
which
is
like
version
of
the
file
of
the
schema
and
then
resources
here
for
each
resource,
let's
see
we
do
it
grabs
the
resources
that
we
actually
grabbed.
A
So
here's
like
in
this
namespace
there's
cluster
information,
persistent
volume,
claims
and
stuff
like
this,
so
I
can
actually
then
go
through.
Look
at
my
deployments
and
here's
the
nginx
deployment
saved
as
a
JSON
file,
and
so
one
of
the
really
nice
things
here
is
that
Valero
is
great
to
do
stuff
automatically.
But
if
you
do
need
to
actually
sort
of
dig
out
sort
of
like
the
raw
data
of
what
you
were
running,
it
has
a
useful,
like
you
can
actually
explode
inspect
to
see
all
the
information
that
it's
backing
up.
Also,
alright.
A
So
so
we
took
a
backup.
We
dug
into
sort
of
what
that
backup
looks
like
under
the
covers,
and
so
now
let's
go
through
and
like
like
say:
okay,
something
bad
happened
and
we're
gonna
go
through
and
delete
that
namespace
and
as
we
delete
that
namespace
it'll
take
down
our
load
balance
or
it'll.
Take
down
the
it'll,
take
down
all
of
the
the
PVS
that
we
created
and
stuff
like
that.
A
So
if
we
look
at
volumes
here,
I
think
we
have
so
we
have
a
dynamic
when
it
takes
a
little
while
to
actually
go
ahead
and
delete
this.
But
let's
give
it
a
second
here,
and
this
could
be
either
deleting
the
namespace
or
it
could
be
deleting
the
cluster
as
a
whole.
So
we
can
actually
we'll
see
how
we
can.
We
can
move
this
stuff
across
cluster
to
so
this
thing
is
now
deleted
and
I
believe.
If
we
do
like
a
refresh
here,
okay,
it
may
take
a
while
I
think
for
it's
no
longer
attached.
A
A
It
gone
yeah,
okay,
so
it
got
its
act
together.
Now
that
that
PV
is
totally
gone,
and
so
now
we
can
actually
go
through
and
do
a
restore
from
backup,
and
so
you
create
a
restore
request,
saying
which
backup
you
want
to
restore,
and
so
it
gives
it
an
automatic
name
based
on
the
date
and
time,
and
you
can
see
that
at
this
point
up,
Oh.
A
I
see
what
happened
here.
It
said
it
was.
It
was
trying
to
create
the
name
Spade,
the
the
deployment,
and
there
was
a
race
between
the
deployment
object,
creating
the
replica
set
and
Valero
recreating
the
replica
set.
So
there's
some,
you
know
it
captures
a
lot
of
raw
information,
but
the
kubernetes
is
resilient
enough
that
this
does
actually
go
ahead
and
settle
down,
and
so
now
we
actually
have
this
thing
up
and
running
again
if
I
go
through
and
you
get
pods
I
have
this
pod
here.
If
I
give
exact,
is
it
named
a
man?
A
It's
actually
has
the
exact
same
name
if
I
go
through
and
look
at
this
you'll
see
that
we
haven't
hit
it
again,
but
the
the
logs
are
still
there.
So
it
actually
round
tripped
this
thing
through
a
snapshot
and
back
to
be
able
to
make
this
happen,
which
is
which
is
really
really
cool,
so
yeah.
So
there
we
go.
We
went
through
that
now
interesting
things.
So,
like
that's
sort
of
the
the
main
flow,
a
couple
of
things
that
I
think
are
worth
looking
at
here
is:
you
can
set
up
a
backup
schedule.
A
Okay,
so
how
is
asking
does
Valero?
Let
you
restore
from
one
cloud
provider
to
another?
The
answer
is
kind
of.
If
you're
not
doing
persistent
volumes,
then
you
can
go
ahead
and
actually
move
things
from
one
cloud
to
another.
No
problem
moving,
something
like
an
EBS
snapshot
is
is
trickier
right,
so
I
mean
theoretically
and
I
think
this
is
something
that
you
know
we've
talked
about,
but
it
hasn't
happened.
A
I
think
it's
it's
a
workflow,
that's
doable
because
nobody's
done
it
quite
yet
to
be
able
to
automate
that
type
movement.
But
it's
definitely
a
you
know
on
our
radar,
if
you're
doing
something
like
rustic,
which
is
the
the
alternate
sort
of
non-block
base,
but
file-based
thing-
and
this
is
what
Nolan's
saying
you
can
go
through
and
actually
use
rustic
to
scrape
things
from
the
inside
of
the
VM
and
then
restore
into
the
inside
and
so
you're,
not
actually
getting
point-in-time
block-level
snapshots.
A
But
those
things
will
fundamentally
be
a
lot
more
portable
than
if
you
were
and
then,
if
you
were
trying
to
do
the
the
block
level
snapshots
all
right.
So
that's
where
we
go
so
like,
let's,
okay,
so
things
that
I
want
to
do.
Do
yet
is
let's
go
through
and
set
up
a
backup
schedule
and
then
also
make
sure
that
we
can
go
through
and
restore
onto
a
different
cluster,
because
I
think
that
ends
up
being
super
interesting.
Also,
so
the
schedule
is
actually
super.
A
Super
easy
and
the
nolan
says
we
also
don't
do
storage
class
translation,
which
is
which
is
tricky
too
one
of
the
one
of
the
interesting
things,
though,
is
that
I
think
you
know
another
thing
that
that
that
you
could
do
either
built
into
velaro
or
separately.
Is
you
know
you
can
take
that
tar
ball
that
describes
all
that
stuff
and
you
could
filter
mutate
change
that
as
you're
actually
moving
stuff
around
and
so
getting
to
the
point
where
we
think
about
like
how
do
we
want
to
actually
take
that
tar
ball
perhaps
manipulate?
A
A
Let's
see
it's
pointing
to
the
storage
location
where
this
thing
is
stored,
based
on
the
name
and
so
the
canonical
sort
of
like
the
sort
of
place
of
record,
for
what
snapshots
you
have
is
actually
a
three.
So
if
I
go
through
and
it
may
take
a
little
while
so
I'm
a
little
off
script
here,
but
if
I
go
through
and
if
I
delete
this
thing.
A
A
It
doesn't
find
any
resources,
but
what
actually
happens
and
I
think
it
takes
a
while
for
things
to
resync
is
that
Valera
will
actually
go
through
list.
The
contents
of
s
3
and
then
repopulate
the
kubernetes
resources
based
on
that,
and
perhaps
somebody
on
the
team
knows
if
there's
a
way
to
actually
go
ahead
and
kick
it
and
make
it
do
it.
I
could
probably
just
kill
the
pod
and
it'll
actually
do
it.
You
control,
ok
and
okay.
Keep
control.
A
A
Alright
you're
making
me
look
bad
Valero,
so
Steve
saying
it
takes
a
little
while
here
so
we'll
see
alright
we'll
come
back
to
that
in
a
little
bit
and
actually
make
sure
that
it's
working
all
right,
so
let's
go
through
and
actually
create
the
backup
schedule.
So
this
is
so
so
a
schedule
is
another
type
of
cid.
Another
type
of
object.
A
Ok,
no
ok,
so
thomas
is
saying
that
research
this
last
week
no
way
to
jumpstart
the
resync
it
defaults
to
every
60
seconds,
so
it'll
take
a
little
while
it's
eventually
consistent
to
some
definition
of
eventually,
and
so
here's
what
we're
saying
so
this
is
we're
creating
a
schedule
and
we're
gonna
call
it.
You
know
egi
k,
backup
and
then
the
schedule
uses
like
general
sort
of
cron
cron,
have
format
right
and
so
I
always
forget
these
things,
but
it
ends
up
being
minute
our
day
of
the
month
month,
day
of
the
week.
A
A
A
The
the
tricky
thing
here
is
that
at
some
point
you
hit
a
quote
in
terms
of
the
number
of
EBS
volume
snapshots
you
can
actually
create,
but
we'll
just
go
ahead
and
like
let
this
thing
do
its
thing.
So
we
know
uhyou
control,
get
our
Valero
get
backups,
and
here
you
can
see
it
went
through
and
created
the
backup
there.
A
A
Backups
yeah,
so
there's
nginx,
backup,
inti
gik
backup,
so
it
should
eventually
get
its
act
together
and
actually
get
this
thing
there.
It
is
okay,
so
yeah.
So
now
we
actually
see
here
we
have
both
the
nginx
backup
and
the
T
gik
back
on.
So
this
is
really
demonstrating
that
the
s3
bucket
here
is
the
source
of
truth.
A
Velaro
is
just
actually
reflecting
the
status
of
that
s3
bucket
into
into
kubernetes
as
a
C
or
D,
and
now
this
will
actually
go
through
and
actually
create
a
snapshot,
every
five
minutes,
which
is
pretty
cool
all
right.
So
the
next
thing
that
we
want
to
do
and
I
think
this
is
the
last
word
of
the
basic
stuff-
is
how
do
I
actually
go
through
and
do
cluster
migration
actually
recreate
this
application
on
a
different
cluster
altogether?
So
to
do
that?
What
we're
going
to
have
to
do
is
I.
A
A
A
We
already
have
the
bucket
created,
we
have
the
user
created.
So
all
those
things
are
the
same.
Now
I
can
just
go
through
and
install
Valero
I
have
the
credentials
there
so
now
we're
installing
Valero
in
this
new
place.
Now.
The
interesting
thing
here
is
that
we
have
two
of
these
things,
pointing
at
the
same
bucket
and
the
same
place
in
the
bucket
and
generally
that's
bad.
A
That's
not
something
that
you
want
to
do,
and
so,
if
we
go
to
the,
if
we
go
to
the
instructions
here,
one
of
the
things
that
they're
doing
is
that
they're
saying
in
cluster
2,
which
is,
is
TGI
K
2.
You
want
to
actually
start
setting
over
a
store
only
flag
which
puts
Valero
into
a
read-only
mode.
Now
this
is
for
Valero
100,
there's
a
feature
that's
in
in
github
right
now.
A
A
A
So
you
can
do
the
dry
run:
yeah,
Mon,
yeah,
okay,
so
now
this
is
actually
fascinating.
We've
seen
that
it's
already
synced
up.
You
know
it.
This
one
happened
faster
from
the
other
one,
so
here
Valero
get
backups.
You
can
see.
We
have
three
backups
here
we
have
the
backups
that
are
sitting
over
here
also
so
now
it
actually.
We
see
both
of
these
backups
on
both
clusters
and
now
we
can
actually
choose
the
latest
one
here
and
do
a
Alero
restore
right.
A
A
A
Okay,
so
it
restored
everything
across
the
cluster.
You
can
see
that
it's
like.
Oh
some
of
these
things
already
existed
because
it's
like
did
some
calico,
so
you
can
also
scope
the
restore
just
to
a
particular
namespace,
which
is
something
that
we
could
have
done
before
and
so
yeah
so
I
think
there's
a
certain
amount
of
like
you
know.
If
you're
doing
a
restore
you,
you
know
it
captures
all
the
information,
some
of
it
may
be
stuff.
A
That's
applicable,
may
not
may
not
be
like,
let's
say
you're
using
calico
on
one
cluster
and
you're
using
you
know,
sort
of
the
native
AWS
C&I
stuff
on
another
cluster.
You
know
it
will
copy
the
calico
configuration
over,
even
if
maybe
you
don't
don't
want
it
to,
and
so
this
is
one
of
those
things
where
you
know
you
know
over
time.
A
A
You
can
see
here's
the
the
status
of
the
of
the
backup
of
the
restore,
and
if
we
do
the
lair
okay,
you
can
see
that
it
continues
to
take
snapshots
backups
every
every
five
minutes
here,
alright,
so
that
is
the
main
flow
of
velaro.
That's
the
stuff
that
I
really
wanted
to
show
off.
I.
Think
it's
worth
talking
about
some
of
the
stuff.
That's
actually
coming
down
the
pike
in
terms
of
things
that
we're
looking
at
so
the
first
one
is
that
there
is
native
snapshot,
support
being
pushed
as
part
of
kubernetes.
A
This
is
a
lot
of
work
happening
in
SIG's,
storage
and
Valero's,
working
that
I
can
actually
sort
of
operate
and
work
with
that
snapshot.
I
think
we're
still
figuring
out
sort
of.
Are
there
things
that
you
can
do
with
the
native
capability
versus
the
built-in
capability?
I
know,
there's
there's
a
lot
of
interesting
work
going
on
there
to
figure
out
the
best
use
of
that
I
know.
Nolan
who's
on
the
chat
here
has
been
working
hard
on
that
there's
the
read-only,
the
read-only
back-up
locations
for
doing
restores.
A
That's
another
thing
that
we're
working
on
Steve
Nolan.
Is
there
any
other
sort
of
up-and-coming
features
that
you
know
that
you
all
are
working
on?
I
mean
obviously
there's
a
there's,
a
plug
in
since
we're
working
with
a
lot
of
folks
to
make
sure
that
we
get
as
many
plugins
as
possible.
I
know
that
there's
a
lot
of
talk
about
sort
of
getting
something
going
with
the
with
the
vcn
folks
here
at
VMware,
so
that
they
can
actually
be
well
supported
here
too.
A
Okay,
so
working
on
moving
the
rustic
integration
to
GA,
it's
currently
beta.
So
rustic
is
it's
a
little
bit
more
complicated
because
you
actually
have
to
you
know
you
don't
have
this
sort
of
like
hey,
just
make
an
API
call
and
it
creates
a
clone
of
your
disk.
It
actually
has
to
do
some
some
relatively
interesting
stuff
from
the
inside.
A
How
is
valera
working
with
helm
so
so
helm
has
its
own
sort
of
set
of
state
and
I
believe
Valero
will
back
up
all
of
that
state.
Also,
as
he'll
move
with
helm,
three
moves
to
something.
That's
much
more
CRD
based
I
think
that'll
actually
work
a
lot
better
with
with
things
like
Valero
Owen.
That
Steve
saying
we're.
Also
thinking
about
the
idea
of
delegated
backups,
so
individual
namespace
owners
can
backup
and
restore
just
their
namespace
I
think
that'd
be
a
really
really
cool
feature.
That's
the
first
I've
heard
of
that.
A
A
So
there's
a
helm
truck
to
deploy
it
in
terms
of
intelligence
around
helm
resources
we're
not
quite
there
probably
need
some
action.
Plugins,
so
yeah
I
mean
helm
is
not
just
you
know
it's
it's
it's
more
stateful
than
I
think
it
probably
needs
to
be
I
mean
some
of
that
is
just
point
in
time
of
when
helm
was
created
and
such
so
yeah,
so
with
that
I'm
gonna
call
it
for
today's
tea
gik.
A
Thank
you,
everybody
for
taking
the
time
to
hang
out
with
me,
and
you
know
please,
please,
please,
you
know,
submit
ideas
into
the
TGI
K
repo.
You
know
we've
switched
to
the
new
channel
here.
This
is
the
new
VMware
cloud
native
apps
channel.
If
you
haven't
already,
please
subscribe,
there
I
think
most
of
the
new
content
coming
out.
There
is
really
going
to
be
kubernetes.
You
know
upstream
related
stuff,
whether
it
be
we
do
office
hours
for
contour
and
velaro
and
stuff
like
that.
A
Let's
see
also
one
other
thing
that
I
want
to
mention
is
that
next
week,
on
Friday
and
I'll
tweet
about
this.
Also,
when
we
get
more
details,
is
that
we're
going
to
be
holding
a
cloud
native
happy
hour,
which
is
something
that
that
Nick
Nicholas
Lane
one
of
the
the
kubernetes
architects
field,
engineers
that
we
have
here
does
where
get
together
and
it's
a
much
more
in-depth
sort
of
like.
Let's
explore
the
new
features
of
the
kubernetes
version.
A
You
know
he
he
kind
of
runs
it
as
a
show-and-tell
type
of
thing
versus
versus
me
doing
my
own
thing
up
here,
and
so
that's
something
to
take
a
look
at
next
week.
Stay
tuned
for
more
details
there.
So
with
that,
thank
you.
Everybody
thanks
for
joining
me
and
I.
Don't
think
I'm
going
to
be
doing
tea
gik
next
week.
I
have
a
conflict,
but
we're
working
on
finding
somebody
else
to
take
it
over
for
the
next
week
and
I
will
see
you
all
later
all
right.
Thank
you.