►
From YouTube: GitOps Happy Hour (Ep 6): Storing (g)it
Description
Join Christian Hernandez, GitOps Extraordinaire, for a journey through how to achieve GitOps in any number of ways. The occasional Red Hatters and special guest will join us too.
Slides: https://speakerdeck.com/redhatopenshift/gitops-happy-hour-storing-g-it
A
Good
morning
good
afternoon
good
evening
and
welcome
to
another
episode
of
the
get
ops
happy
hour
where
we
don't
drink
unless
it's
diet,
coke,
at
least
for
me,
I
am
chris
short
executive
producer
of
the
open
shift
tv.
I
am
joined
by
the
one
and
only
get
ops,
extraordinaire
christian,
hernandez,
how's
it
going
christian.
B
B
Yeah,
whatever
it
is
yeah,
but
I
I
still
yeah
it's
good.
I
can't
believe
the
year
is
almost
over
getting
getting
ready
for
2021
right.
We
announced
a
bunch
of
summit
stuff,
that's
going
to
be
in-person
summit
stuff.
So
it's
like
wow
what
we're
going
to
do
stuff
in
person
right.
They
didn't
even
dream
of
doing
that
in
2020,
but
towards
the
end
of
2021,
so
check
your
calendars,
we'll
be
doing
actually
live
stuff.
So
I'm
excited
about
that.
I
don't
know
about
you,
chris,
but
I'm
a
little.
B
B
Yeah
yeah
so
definitely
check
that
out,
you'll
be
hearing
more
as
the
year
comes
more
and
more
info
about
the
summit
and
what's
coming
up
there,
so
yeah
and
also
something
else,
cool
that
came
out
you
actually
mentioned.
While
we
were,
we
were
just
chatting
before
the
before
the
stream
here
we're
talking
about
the
github's
actions.
Actually
that
came
out
for
openshift
yeah,
so
that's
actually
pretty
cool.
Maybe
that's
something
in
a
later
episode,
we'll
dive
deep
deeper
into,
but
it's
pretty
cool.
A
A
A
A
B
B
A
B
Know,
handling
storage
and
get
ops
gets
a
little
tricky
right.
So
this
is
a
question
that
comes
up
a
lot
of
the
times.
Right,
especially
when
you
know
someone
who's
you
know
is
getting
started
with
with
get
ops
gets
a
little
further
down
the
line
they're
like
well.
How
do
I
manage
storage
with
with
get
ops
like?
B
How
do
I,
you
know,
manage
things
that
need
state
right,
stateful
things,
for
example
like
a
database
right
like
if
you're
running
a
database,
you
know
your
data
is
something
you
want
to
keep
around
right,
and
so
you
know
I'm.
I
guess.
I'm
gonna
talk
a
little
bit
about
the
the
paradigm
of
persistent
volumes
and
precision
volume
claims
in
kubernetes.
C
B
Right
kind
of
just
get
that
out
of
the
way
and
then
kind
of
go
over
some
of
the
challenges
that
brings
in
in
terms
of
get
ups
in
terms
of
declarative
infrastructure
and
and
then
we'll
do
some
hacking
right.
So
this
is
this
is
one
of
the
things
is.
Is
I
don't
think
I
found
the
the
the
exact
answer
yet,
but
I
I
did
find
some
things
that
you
can
do.
B
Again,
like
I've
said
this
before
I'll
say
it
again.
This
is
some
of
the
things
that
the
canadians
have
said.
It's
like
this
is
like
an
ever-changing
thing
right.
This
is
like
a
journey
like
we
use
journey.
I
think
it's
overloaded
a
little
bit
but
like
this
is
like
truly
a
journey
like
some
of
the
things
I
think
andrew
joked
is
like
getups
happy
hour.
What's
what's
get
ups
this
week,
because
it
keeps
just
keeps
changing
right
so.
A
B
Yeah
yeah,
exactly
so
all,
and
it's
funny
I
I
do
I
I
did
a
talk
about
get
ups
like
before,
actually
literally
like
like
a
month
before,
we're
all
ordered
to
shelter
in
place.
No
more
travel,
you
know,
and
you
know,
a
lot
of
questions
came
up
and
I
always
say
it's
like
this
is
like
as
soon
as
more
and
more
eyes
get
on
it.
More
and
more
people
have
opinions
and
other
people
go.
Oh,
I
never
thought
about
that.
That's
actually.
B
C
B
A
B
A
C
B
Has
already
like,
if
you're
using
abs
volume,
that
already
has
a
csi
in
front
of
it,
so
so
a
lot
of
these
things
are
moving
over
to
csi,
but
but
if
you've
been
around
kubernetes
a
while
right,
there's
this
concept
of
like
the
pv
persistent
volume
and
pvc
percent
volume
claim
right,
so
it
is
essentially
persistent.
Volume
is
nothing
more
than
a
definition
on
how
to
connect
to
storage
right
like
how
and
what
kind
of
storage
that
storage
provides
right.
B
B
Options
what
modes
right
like
can
more
than
one
instance
mount
this
right,
because
if
you're
using
block
storage
you
can't
or
shouldn't
and
if
you're,
if
you're,
using
like
nfs
or
file
based
storage-
like
you,
can
it's
it's
a
file,
it's
a
network
file,
storage.
So
right
and
there's
a
one-to-one
mapping,
usually
right.
So
there's
a
persistent
volume.
There's
a
claim
that
says
hey!
I
need.
I
need
storage
of
x
capacity
of
x
has
these
parameters
right.
B
B
You
know
at
the
base
level
there's
no
way,
you
know
you
don't
define
this
pv
matches
this
pvc
right,
I
mean
just
like
just
conceptually
right.
Obviously
there
is
ways
to
do
it,
but-
and
essentially
what
happens
is
that
once
that
that
you
know
once
you
define
the
volume
and
once
the
claim
matches
the
volume,
the
node
actually
mounts
that
right?
It's
the
node
that
actually
mounts
that
storage
and
then
presents
it
to
the
to
the
container
right
inside.
So
this
is
kind
of
kind
of
a
diagram
kind
of
showing
like.
B
A
B
B
It
does
feel
like
forever,
though,
like
like
this
stuff
was
like
you
would
have
to
do
a
lot
of
the
stuff,
and
now
I
mean
you
can
do
it
this
way
now.
Obviously,
but
this
had
to
be
done
manually
right,
so
you
would
have
to
have
you
have
an
admin
that
defines
persistent
volumes
right?
You
have
to
create.
B
B
B
End
user
task
right,
like
the
admin
has
to
define
the
pv
and
the
and
the
user,
defines
the
claim
once
once
a
match
is
found
the
claim,
then
the
user
has
the
right
to
claim
that
storage
right.
Obviously,
so
that's
why
they
call
it
a
claim,
and
then
the
mount
happens
transparently
right
to
the
admin
into
the
user.
B
The
the
admin
just
defines
how
it's
mounted
the
actual
mounting
happens
via
via
the
api
right,
so
the
the
kubernetes
api
so
and
then
came
dynamic
storage
right,
so
this
is
kind
of
this
was
like
actually
really
really
cool
like
when
it
first
came
out.
I
thought
this
was
like.
Oh
my
god.
This
is
like
really
cool
that
this
came
out
is
essentially
you
create
something
called
a
storage
class
right.
B
Originally,
a
storage
class
was
a
definition
of
how
you
tiered
your
storage
right
because,
as
a
storage
admin,
you
have
different
tiers
and
of
storage.
Like
I
know,
I
know
I
have
different
tiers
of
storage
right,
so
I
had
you
know
you
had
fast
ssds
right,
like
fiber
channel,
you
have
like
slow,
spinning
discs,
you
have
stuff,
that's
like
near
line,
storage
right
and
you
have
tiers
right
depending
on
what
the
application
needed
right.
You
didn't
want
to
like
give
everyone
ssds
right.
That's
just
not!
B
You
know
financially
feasible
right
in
some
places
where
you
can't
just
have
like
fast
storage
for
all
right.
You
need
to
like
to
your
storage.
That
was
the
original
intent
with
storage
classes.
Define
those
things
but
kind
of
the
byproduct
of
that
is
that
the
storage
class
then
actually
automatically
creates
that
storage
for
you
on
the
fly.
So,
for
example,
if
you
something
like
vmware
right,
which
is
an
example,
I'm
going
to
go
with
go
over
is
the
vdmk
is
the
storage
actually
doesn't
exist
yet
right
so
before?
B
If,
if
I
go
back
one
right
before,
not
only
did
you
define
the
persistent
volumes
but
that
storage
actually
had
to
exist?
The
storage
like
I
as
an
admin,
not
only
did
I
have
to
create
the
persistent
volumes
of
kubernetes.
I
had
to
actually
walk
over
to
the
storage
admin
and
say:
please
create
the
storage
right
or
I
actually
have
to
go
to
the
command
line
and
actually
create
that
storage
from
this.
For
it
to
exist
right
with
dynamic
storage
provisioning,
that
storage
doesn't
have
to
exist.
B
That's
the
storage
only
exists
at
creation
time,
meaning
that
when
a
user
asks
for
storage
right,
I
want
two
gigs
of
read,
write
many,
and
I
want
my
tier
to
be
a
kind
of
good
storage
right.
I
don't
want
the
best,
but
I
want
pretty
good
storage
right.
Maybe
it's
a
web
server,
and
so
that
claim
goes
to
the
aps
or
the
api
server.
That
says:
well,
the
good
storage
you
defined
it
as
the
netapp
ssd
right.
B
B
Other
storage
providers
have
their
own
api
right,
including
including
us
right
that
we
have
our
own
openshift
ocs
api
endpoint.
That
is
used
right,
so
the
api
server.
Actually,
then,
it's
kind
of
hands
off
for
the
admin
the
admin
defines
the
storage
back
ends
once
right.
I
define
it
once
and
then,
whoever
you
know,
wants
storage.
It
goes
through
the
api
and
then
the
api
takes
care
of
provisioning.
It.
B
C
A
B
From
there
on,
like
the
admins
kind
of
you
just
you
set
it
and
and
then
it
just
mounts
it
dynamically
right
and
then
then
the
whole
process
is
kind
of
normal
it'll
mount
it
in
the
container.
Then
the
container
couldn't
consume
the
storage
right.
So
so
that's
like
from
a
high
level
right.
This
high
level,
persistent
volume
percent
volume
claim
this
is
kind
of
kubernetes,
101
102,
let's
say
kind
of
stuff
right
where
it's
like.
You
know.
How
do
I
consume
storage?
Oh,
I
need
a
claim.
B
A
B
B
C
B
There
are
some
challenges
in
the
in
the
get
ops
world,
so
some
of
the
challenges
with
like
well,
we
just
thought
you
know,
I'm
gonna
go
backwards
right
talk
about
the
last
thing.
First,
is
the
dynamic
storage
provisioning?
What
we
just
talked
about
right
by
default-
and
this
is
just
like
the
default,
the
res,
the
storage
reclaim
policy
is
set
to
delete.
So
talking
a
little
bit
about
the
storage.
B
Don't
bother
with
it
because
it's
going
to
be
gone
anyway,
so
I'll
only
mention
it
because
you
might
see
it
in
the
documentation,
but
there
is
there's
delete
and
retain
right,
so
delete
is
and
that's
the
default.
So
when
me
as
a
developer,
if
that
persistent
claim
goes
away
like
you
know,
for
any
reason,
we
either
whether
it
be
a
disaster
or
whether
I
accidentally
delete
the
project.
The
back-end
storage
gets
deleted.
Yeah.
B
Yeah
yeah,
you
want
to
be
able
so
the
so
by
default,
so
that's
kind
of
some
of
the
things
with
get
ops
right
like
if
you
want
to
you
know
recover
from
a
disaster
using
get
ops
right
because
that's
like
the
whole
idea
of
get
ups,
I
just
reapply
my
yaml.
Well,
you
know
you
can
reply
your
yamo
all
you
want,
but
your
data's
gone.
So
it's
kind
of
okay,
so.
B
So
you
have
for
dynamic
storage,
you
have
little
control,
so
you
know
from
the
consuming
side.
You
know
if
I'm,
if
I'm
a
developer,
if
I'm
a
user
of
a
kubernetes
cluster,
I
have
actually
honestly
very
little
control
and
almost
to
a
certain
extent,
the
admin
which
is
fine
on
automation.
Right,
like
you
have
to
the
automation,
dynamic,
you
have
to
pay
the
tax
of
the
control
right.
It's
just
like
one
of
those
things
and
reclaiming
storage
after
disaster
and
expanding
it's
kind
of
like
a
manual
process
right.
B
B
B
B
Yeah,
so
you
know
but-
and
actually
I
I
can
actually
hear
the
collective
groan
of
operator
operations
guys.
I
like
they
much
rather
try
to
re,
mount
the
volume
and
see
if
it
comes
up
rather
than
like
try
to
restart
for
backup,
because
we
start
from
backup.
You
know
it.
Just
stuff
gets
lost
right.
You
can't
just
you're
gonna.
B
You
know
a
quick
story,
you
know
back
at
you
know,
dollar
sign
last
job.
You
know
we're
taking
snapshots
every
15
minutes
of
the
database
right
like
netapp
snapshots,
every
15
minutes,
so
a
snapshot
was
taken
at
the
at
the
top
of
the
hour
and
somewhere
like
seven
minutes
afterwards
or,
like
you
know
like
or
not
even
like,
even
worse
than
that,
like
13
or
14
minutes
afterwards,
a
developer
accidentally
dropped
the
table
right
and.
B
A
B
B
Just
that
could
be
a
lot
of
information
yeah
so
like
it
doesn't
you
know.
So
you
know
restoring
from
backup
is
like
a
big
thing,
so
I
I
already
hear
that
that
groan
of
of
like
oh,
what
do
you
mean
after
we
start
from
back
up
like
I
don't
want
to
do
that.
You
know
like
as
an
admin
like
because
it's
like,
I
just
know
the
pain
of
you
know
going
back
to
the
business
saying.
Well,
we
lost
you,
know
so
much
dollars
because
you
know
we
did.
We
can't
take
more
than
15.
B
B
You
know
there's
just
a
process
and
I
have
to
set
up
for
each
one
right
keeping
track
of
what
is
what
it's
a
little
difficult
right
of
like
what
is
connected
to
what
that,
even
it's
even
true
to
a
certain
extent
of
dynamic
storage
and
more
control,
you
have
like
more
control
in
a
manual,
but
it's
you
know.
More
control
just
means
more
steps
right,
there's
more
feature,
other
things
to
consider
right
so
like
file
permissions
file
permissions,
especially
in
openshift,
since
everything
runs
as
a
random
uid-
is
very
difficult
right.
B
Yeah
when
you
restore
when
you
reapply
your
your
yaml
to
a
cluster
that
you
restored
its
uid
changes,
so
trying
to
keep
track
of
permissions
is
is
really
difficult,
so
you
need
to
like
just
some
things
to
consider,
and
then
you
know
how
you
restore
from
backup
right.
So,
like
okay,
you
know,
I
reapplied
my
ammo,
I'm
ready
to
restore
for
backup
like
how
do
I
you
know.
How
do
I
do
that
in
such
a
way?
You
know.
Do
I
not
now
I
have
to
scale
down
the
application.
B
B
A
B
B
You're
trying
to
get
that
yeah
exactly
so,
even
if
you
can
mount
that
the
original
storage
you
know
you
still
may
have
to
back,
restore
from
backup
right.
It's
like
you,
may
you
know,
data
still
may
be
missing
and
you
may
have
to
restore
that
data.
You
know
from
a
backup
somehow
right
so
so
there's
no
magic
bullet,
but
I
do
want
to
spend
like
this
next
half
to
just
do
a
little
exploring
right,
yeah.
Let's.
B
C
B
B
No
way
reflects
what
you
would
need
to
do
in
a
production
system
right
so
like
just
cut
that
caveat,
but
let
me
use
octotree
because
we
love
octotree
right.
So
I
have.
I
have
a
directory
called
storage,
ops
right,
so
I
have
one
I
have
my.
This
is
a
very
simple
example,
but
I
think
illustrates
the
point
right:
it's
basically
a
two-tier
application
for
an
web
database
that
database
needs
storage
right.
So
I
have
here
my
persistent
volume
claim
right
right.
B
A
B
To
my
kubernetes
cluster,
let's
see
get
pods
right,
so
I
have
oh,
I
don't
have
oc
get
nodes.
I
have
this
on,
not
iws.
I
have
this
on
on
vsphere
right.
So
if
I
do
an
oc
get
sc
stands
for
storage
class.
I
get
the
vsphere
volume
right
in
a
central
vision.
If
I
do
an
oc
get
persistent
volumes
right.
I
have
some
personal
volumes
because
I
use
mineo
for
object,
storage
here,
but
you
know
this.
It
doesn't
show
any
other
storage
right.
B
It
just
basically
says
I
have
thin
meaning
thin
provision.
I
have
this
storage
here
and
then
what
the
reclaim
policy
is
the
default
reclaimed
policy
here
is
delete.
So
remember,
like
I
said
before,
delete,
and
so
let's
go
back
to
here
right.
This
is
the
only
you
know.
Everything
else
is
pretty
much.
You
know
just
kind
of
the
status
quo,
boilerplate.
B
B
C
A
B
C
A
A
B
A
B
B
Yeah,
that
was
a
pretty
glorious
row.
Someone
said,
I
think,
that's
that's!
That's
pretty
funny.
Yeah,
indeed,
yeah
definitely
definitely
a
rope
row.
Okay,
so
I
actually
paused
when
we
went
down
so
there
wasn't.
There
was
nothing
you
missed
right.
So
if
I
do
an
oc
get
pv
in
the
pvc
right,
it'll
say
I'm
bound
here.
If
I
do
an
oc
get
pods
of
priceless
right
sort
of
the
can
I
do
oc
get
pvc.
B
B
A
B
B
C
B
B
B
B
B
Along
the
same
old
line
is
the
road
row
right
because
since
I
use
dynamic
and
because
the
the
I'm
using
dynamic
and
the
default
reclaimed
policy
was
to
delete
when
I
lose
my
my
namespace
by
accident
right,
like
maybe
I
fingered
that
maybe
I
have
a
bunch
of
priceless-
it's
gone
right.
So
if
I
click
sync,
if
I
sync
again
yeah
it'll,
you
know
it'll
create
this
quota
to
the
pvc
right,
so
yeah
it'll
create
it'll,
create
it,
but.
A
B
B
So
yeah,
so
it's
up
if
I
reload
the
page
yeah,
my
app
is
back,
but
there's
no
records
right.
That's
because
I
lost
the
storage
right.
So
this
is
a
couple
things.
You
can
still
use
this
model,
but
you
have
to
do
a
database
backup
right.
So
you
right,
you
have
to
do
my
sql
dump
or
whatever
of
the
data
you
have
to
maybe
have
a
job
or
you
know,
cron
job
going
in
kubernetes.
You
can
do
crown
jobs
of
it
just
basically
dumping
the
database.
C
B
B
If
you're
going
to
go
that
route,
that's
just
something!
You
need
to
keep
in
mind
right,
something
you
know
I'm
going
to
have
to
one
take.
You
know
you
should
be
taking
it
back
out
your
databases
anyway
right,
but
but
two
you
know
also,
you
know
account
for
the
time
of
restoring
the
backup
before
you
go,
live,
obviously
right.
B
B
A
B
Don't
know
what
it
is:
yeah
yeah.
So,
oh
well,
that's
deleting.
Let's
look
at
the
the
other
method
right,
so
the
other
method
includes
actually
defining
a
pv.
So
let's
look
at
this
pv
and
see
what
the
definition
is
right
here
right.
So
this
is
not
using
dynamic
storage.
Obviously
this
is
using
the
manual
way,
the
old
school
way,
and
I
basically
set
the
storage
class
name
to
null
because
I'm
saying
I'm
not
going
to
use
a
storage
class.
B
This
is
a
direct
direct
connect
right,
I'm
not
going
through
the
api
that
whole
flow
of
using
storage
class
right.
So
I
set
that
to
null
right
and
then
this
is
sort
of
some
of
the
things
that
you
would
expect
right.
Storage,
one
gig,
I'm
gonna
be
using
nfs,
so
I'm
gonna,
I'm
gonna
click
file
system
right,
because
that
that
defines
what
type
of
storage
I'm
going
to
use
read
write
many
right
because
it's
nfs
all
right,
very
important,
persistent
volume!
Reclaim
policy
is
retained
right,
so
I'm
going
to
retain
the
data.
B
I
don't
want
it
to
go
away.
I
just
want
it
to
retain
right
and
then
so
there's
this
claim
reference
right,
so
meaning
that
when
someone
tries
to
claim
this,
I'm
only
going
to
satisfy
the
pvc.
That's
named
this
way
in
the
namespace
name.
This
way,
so
this
is
essentially
along
with
line
seven
and
eight
right,
which
is
labeling.
I'm
gonna
one
match
the
label
and
two
look
for
a
specific
claim
reference.
So
this
is
how
I
do
that
one-to-one
mapping.
So
I
want
to
say
this
storage
belongs
to
this
app.
C
B
A
B
B
So
here
retaining
storage
policy,
so
so
rewind.
Sorry,
if
I
were
repeating
myself
a
little
bit
here,
but
I'm
maintaining
the
storage
right,
I'm
saying
only
this
claim.
You
know
this
claim
name
coming
from
this.
Namespace
can
claim
the
storage,
and
this
is
the
nfs
storage
back-end
right.
So
if
I
go
to
my
nfs
here
right,
so
that's
the
storage
back-end
for
nfs
and
then
I
do
you
know
something
else.
I
need
to
call
out
here.
B
B
So
I
have
a
pvc
right
which
is
kind
of
the
same.
I
kind
of
want
to
say
in
my
claim
I
want
to
say
I
don't
want
to
use
the
storage
class.
I
want
to
connect
to
this
specific
volume
right
that
I've
been
defining
up
here,
right,
press
this
db,
pv
and
then
press
this
dv
right
like
I
want
to
connect
to
that
guy.
I
want
to
match
the
labels,
so
I
want
this
to
be
all
one
to
one
right.
So
so
let's
see
this
in
action
right.
B
C
B
B
B
A
B
B
B
A
B
C
B
Yeah
stuff
is
dead
right.
If
I
go
back
here,
if
I
don't
oc,
get
pv,
it'll
say
a
static,
it
was
bound,
but
now
it's
released
right
now.
I
have
it,
but
I
have
it
retained.
So
the
data
is
still
there
right
right
and
so
in
a
cluster
failure.
Actually
the
cluster
will
be
gone.
So
if
I
do
an
oc
delete,
you
know
pv
the
actual
pv
right.
The
pv
actual
will
be
gone
too,
but
the
data
will
still
be
safe
right.
The
data
will
still
be
there.
It's.
B
The
definition
is
gone,
gone,
yeah
the
definition's
gone,
but
we
had
retained
so
it
didn't
touch
it
still.
The
data
is
all
there,
so
this
is
kind
of
more
analogous
to
like
a
cluster
goal
going
down
not
like
a
whole
site
going
down,
because
you
know,
hopefully
you
know
as
me.
I
know
me
as
an
admin.
Well,
hopefully,
if
I
just
restore
the
for,
you
know,
restore
the
application
and
just
mount
it.
It
should
just
work
right.
I
don't
want
to
have
to
touch
my
backups
if
I
don't
have
to.
B
C
B
B
B
So
I
got
everything
back
right,
so
the
this.
So
you
know
this
is
more
analogous
to
what
people
think
of
when
they
say.
I
just
need
to
reapply
my
well.
You
know
when
get
ops
when
we
say
I
should
need
to
reply
reapply
my
manifest
and
my
applications
back.
This
is
more
analogous
to
that
right.
This
is
more.
This
is
closer
to
that
goal
than
the
other
way,
not
to
say
the
other
way
is
not
valid.
The
elevator
is
completely
valid.
B
B
Yeah,
your
infrastructure,
your
apps,
you
know
all
those
things
to
me.
I
think
this
is
more
closer
to
get
ops
in
a
way
that
I
would
do
it
right
right.
You
know
as
and
again
depending
on
the
application
right.
If
I
have
an
application,
that's
a
stateful
application
that
that
needs
to
be.
You
know
like
I'm
running,
like
an
ecommerce
site,
or
I
don't
know
whatever.
This
is
more.
B
You
know
closely.
Actually
this
is
funny
it's
more
it's
closely
related
to
traditional
infrastructure,
where
I
have
like
a
one-to-one
relationship
between,
like
my
data,
my
database
and
an
application
mounting
a
specific
amount
for
that
database,
and
you
know,
but,
as
you
see,
it
makes
kind
of
restoring
a
little
easier
right
right.
If,
if
I
lose
my
cluster,
I
can
just
reapply
all
these
yamls
and
if
I
took
care
of
my
my
storage
right
right,
the
data's
still
there
it
should.
You
know
just
work
right
so.
B
B
B
B
B
B
A
B
You
know
like
if
you
know
you
have
a
database
that,
like
barely
touches
the
disk
because
everything's
just
trying
to
happen
in
memory,
but
that
fs
sync
man
that
takes
forever.
We
try
to
flush
those
out
so
so
yeah.
So
that's
you
know
kind
of
like
two
ways
of
doing
this.
I
I
think
the
second
way
is
more
closely,
but
I
would
love
you
know
anyone
in
the
chat.
B
And,
like
I
mean
you
can't
run
spinnaker
everywhere
like
on,
you
know
like
right.
Like
you
just
it's
just
you're
gonna
come
across,
especially
some
legacy.
Applications
you're,
gonna,
you're
gonna,
run
into
some
applications
that
you're
just
like
this
one.
It's
just
it
just
uses
the
database
and
it's
just
it's
store
space,
the
state,
it's
stateful,
and
so
so
yeah
yeah.
A
B
A
B
So
yeah
some
something
that
happens
at
red
hat.
That's
really
really
awesome
is
we
have
a
like
a
recharge
and,
and
it's
essentially
the
last
week
of
the
year
we
do
like
kind
of
like
a
recharge.
Essentially
the
company,
you
know
shuts
down.
You
know
a
lot
of
people
take
time
off
around
that
shutdown.
So,
like
I
know
myself,
I'm
taking
pto
around
that
that
that
that
recharge
day
so,
like
essentially
I'm
gone
for
you
know
18
days.
B
So
it's
actually
something
really
cool
yeah,
really
awesome
that
red
hat
does
for
its
employees.
But
that
being
said,
get
ups
happy
hour
will
return
on
the
14th
january
14th.
Yes,
so
market
calendars
will
be
back
on
the
14th
talking
get
ups,
but
you
know
we're
taking
a
little
break.
I
know
chris,
I
know
chris.
I
know
you
deserve
a
break.
I
mean
you're
doing
non-stop
streaming,
I'm
pretty
sure
you
deserve
some
time
off.