►
Description
A show that features the people and technology that make Red Hat® Enterprise Linux® into the world’s leading enterprise Linux platform.
A
A
Yeah,
I
am
all
for
it
right
like
consider
this
an
office
hour
about
linux,
file,
system
directory
structures,
and
we
will
happily
take
you
down
any
rabbit
hole
you
want
to
go
down.
We
are
ready.
B
So
we've
talked
about
this
a
while
ago,
and
it's
like,
like
I
said,
a
foundational
topic
that
a
lot
of
folks
don't
always
have
their
arms
around
and
I
figured
we'd
start,
which
is
talking
about
like
what
that
top
level
directory
structure
looks
like.
So
let
me
share
my
screen.
B
Did
yeah
so,
let's
just
take
a
look
at
that
top
level
directory
and
what
I
did
was.
I
just
took
this
list
of
directories
and
I
put
them
over
here.
B
Like
it
and
and
putting
chris
short
on,
the
spot
is
always
fun
to
do
so.
I
figured
we
would
take
turns
talking
about
what
we
expect
to
find
in
each
of
these
directories.
A
A
Level
binaries
there's
other
places.
To
put
you
know
your
personal
binaries
or
you
know,
artifacts
from
your
your
your
software
that
might
be
running
but
like
this
is
where
cat
and
ls
and
rm
and
all
those
you
know
like
core
linux
commands
those
binaries
live
here.
B
If
you
take
a
look
at
the
graph
environment,
variable
every
user,
if
they
log
in
with
a
bash
shell
and
use
the
standard
shell
scripts
that
come
with
rel,
will
actually
have
their
own
personal
bin
directory
added
to
their
path
by
default.
A
B
A
B
These
are
a
logical
representation
of
devices
attached
to
the
system,
so
this
is
where
we
find
these
special
kind
of
files
that
represent
hardware
and
generally
they're
indicated
on
whether
they're
a
character
or
black
device,
and
then
there's
like
the
major
and
minor
numbers
instead
of
number
of
links
and
some
other
pieces
of
data
about
them
in
their
inode
but
more
more
generically.
A
All
right,
so,
let's
see
I
mean
I'll
just
do
home
home-
is
where
all
your
users
live.
Essentially
right
like
they
will
work
inside
home,
their
home
directory
is
theirs
like
unless
you're
root
on
a
system,
one
user
can't
look
at
another
user's
directory
by
default,
at
least
so
it's
kind
of
like
your
private
playground
to
do
whatever
you
need
to
do
in
you'll,
see,
interestingly
enough
root
does
not
live
under
home.
It
lives
under
root.
We'll
talk
about
that
later,
but
you
know
your.
A
Will
have
your
files
in
it
right
like
if
you
use
like
the
the
gui
for
rel?
I
forget.
No,
if
you
use
gnome,
for
example,
that
home
directory
is
kind
of
like
where
everything
is
for
you,
your
downloads,
your
documents,
everything
that
kind
of
deal.
B
Yeah
and
what
goes
in
all
these
directories
is
dictated
by
a
document
called
the
linux
file
system
hierarchy
standard
and
for
for
linux.
This
is
where
user
directories
go
for
unix
at
large.
A
A
B
Yeah
so
but
home
is,
is
where
most
modern
distributions
will
put
user
stuff.
A
B
Yeah
for
rel,
we
we
include
it
right,
but
but
yes,
there
could
be
specialized
builds
of
other
linuxes.
That
maybe
would
not
include
that.
But
we
we.
B
It
in
our
in
our
file
system
level
for
real,
so
I'll
do
a
twofer
okay,
because
we
have
lib64
and
then
there's
also
liv
and.
B
Yes,
but
they're,
both
first
shared
libraries
and
the
difference
between
them
is
that
you
know
once
upon
a
time
we
were
compiled
with
32-bit,
binaries
and
libraries
and
then
at
some
point
we
started
supporting
64-bit
stuff,
so
lib64
was
created
to
hold
those
and
now
we've
kind
of
coalesced
them
back
together
again.
But
I'll
talk
about
that
coalescing
in
a
bit,
so
lib
and
lib64
are
for
shared
libraries.
B
A
But
anyways
mount
is
the
you
know,
defecto
point
for
mounting
file
systems
of
another
variety
right.
So
you
know,
if
you
have
your
iscsi
network,
this
would
get
mounted
here
as
a
subdirectory
underneath
mount.
You
know
your
nfs
shares
would
get
mounted
for
system
use
in
the
mount
directory
and
notice.
A
It
says
m
t
and
I'm
saying
mount
that's
because
like
they're
one
in
the
same
kind
of
deal
where
media
is
a
mount
point
for
removable
media,
so,
like
you
insert
a
thumb,
drive
or
a
cd
dvd,
whatever
even
certain
kinds
of
card
readers,
that
kind
of
thing
right,
like
that'll,
show
up
in
media,
but
it
does
use
mount
to
put
those
things
into
media
and
media
is
kind
of
newer
than
mount
in
the
directory
structure
by.
B
B
Proc
is
a
in-memory
file
system;
it's
actually
not
stored
on
disk.
So
anything
you
look
at
in
that
directory
is
actually
you're,
looking
at
components
in
the
kernel's
memory
in
the
runtime
and
so
there's
all
kinds
of
stuff.
In
there
we've
actually
done
two
shows
on
it.
One
was
yeah.
A
B
Overview
and
another
one
was
a
deeper
dive
and
tunables
that
are
there
right,
but
it
got
its
name
because
originally
it
was
where
we
could
find
process
information
so
like
when
you
look
in
there
you'll
see
all
the
pids
of
the
processes
running
on
the
system.
A
That's
a
funny
funny
comment.
I
like
that
one
there's
the
proc
episode,
I'm
dropping
in
the
chat
now
and
then
the
tunables
episode
I'll
grab
here.
Real
quick,
come
on
youtube
search.
B
So
I
I
mentioned
that
there
was
a
newer
place
that
noam
was
using
for
file
system,
removable,
media
stuff
and
that's
under
run.
These
are
four
mount
points
that
people
need
during
the
run
time
of
an
application.
So
when
you
plug
a
usb
thumbstick
into
gnome,
that's
running
it's
running
application.
It's
like!
Oh!
I
need
to
make
a
mount
point
to
mount
this
usb.
Stick
that
just
got
plugged
in
for
my
user,
so
I
can
run
the
files
application,
so
it
creates
this
subdirectory
to
run
and
then
attaches
that
device
at
that
location.
A
A
As
far
as
you
know,
your
petition
goes
sometimes,
and
it's
supposed
to
sit
in
theory
at
the
front
of
the
disk,
but
it's
kind
of
hard
like
in
the
virtual
world,
to
imagine
the
front
of
the
disk
right
like
especially
with
ssds
right.
A
That
was
because,
on
spinning
media
spinning
rust,
as
I
like
to
call
it
nowadays,
it
would
be
faster
access
times.
So
now,
with
ssds
nvme
drives
like
the
access
time
is
consistent
across
the
entire
storage
device.
So
it
doesn't
really
matter
as
much,
but
boot
is
where
the
inet
system
lives
well,
not
all
of
it,
but
some
of
it
and
like
how
linux
starts
on
your
system
that
lives
here.
The
kernel
like
lives
here
and
everything
gets
loaded
at
startup
from
boot.
B
Yeah
and
it
used
to
be
like
way
way
back,
you
put
the
bootloader
in
the
master
boot
record
under
disk,
which
was
the
first
sector
of
the
disk
right.
B
We
can't
put
a
whole
lot
of
smarts
in
one
sector
of
space,
so
we
had
to
create
boot
and
make
sure
that
it
was
at
the
front
of
the
disk,
because
you
could
put
a
little
binary
in
there
to
be
your
boot
loader,
but
it
could
only
have
so
much
address
space,
so
you
had
to
store
those
files
that
it
was
accessing
within
a
certain
amount
of
your
disk,
because
it
just
didn't
have
the
address
space
to
find
them
if
they
were
stored
later
on
that
this
one
is
one
that's
kind
of
still
a
little
funky,
because
in
some
places
like
cloud
images,
a
lot
of
the
marketplace,
images
and
stuff
have
moved
away
from
this.
B
You
just
have
the
slash
file
system
and
that's
it
yeah,
however,
for
like
disa
stick
compliance.
B
B
B
A
A
I
had
been
bitten
by
this
before.
Yes,
because
the
boot
partition
was
so
small
and
then
the
kernel
that
was
upgraded,
didn't
work
so
yeah,
yeah,
good
stuff.
B
That's
how
it
was
initially
done
back
in
like
the
system,
five,
and
actually
it
was
berkeley
that
started
that
that
directory.
B
But
what
has
happened
over
time
is
it's
become
the
place
where
we
put
configuration
files
because,
like
you,
put
your
binaries
in
the
bin
directory
and
you
put
other
files
in
other
directories
and
then
I
was
like
well,
I
need
this
config
file.
Where
does
that
go
yeah
I'll,
put
it
in
the
center
directory
and
over
time
that
just
became
the
place
where
all
the
configuration
files
went.
B
A
Etsy
containers
is
a
great
directory
if
you're,
using
podman
or
any
container.
You
know
orchestrator
or
driver,
or
anything
like
that.
Etsy
containerscontainers.com
will
contain
all
your
image
registries,
for
example.
A
So
that's
always
good
to
be
aware
of,
and
you
know
that
order
and
precedence
matters
as
well.
A
Is
you
get
like
like
opt-ben
and
doc
and
stuff
are
usually
in
there,
but
then
you
get
opt
package
bin
and
people
get
confused.
Sometimes
right,
like
let's
say
my
package,
is
you
know
a
go
binary
and
that's
not
where
the
go
binary
goes
right,
like
opt
is
for
specifically
stuff
you're
deploying
right,
like
it's,
not
necessarily
the
de
facto
place
where
stuff
gets
installed,
which
is
kind
of
confusing
to
people.
But
if
you
look
at
the
page
for
op,
there's
some
very
specific,
oh
there's
a
question
in
here.
A
Right,
let's
see
here
doesn't
necessarily
say
that
in
the
docs.
A
B
Time,
yes,
the
90s.
B
Only
right,
and
so
they
created,
opt
to
be
the
place
where,
if
you're,
adding
on
stuff,
that's
where
you
should
add
it
on,
because
it
was
optional
to
the
system,
and
so
because
of
that
complex
history
and
the
fact
that
things
like
bin
directories
under
opt
were
added
to
people's
paths
by
default,
that
that
line
got
blurred
really
fast
over
the
need
for
ease
of
use.
B
So
there
are
still
some
packages
that
will
do
that.
Like
a
lot
of
the
add-on
software,
you
get
from
ibm
or
oracle,
for
example,
with
their
database
suite
yeah
they'll,
put
it
down
under
opt-in,
so
there'll
be
a
directory
in
opt.
That
says,
like
you
know,
name
of
software
and
then
underneath
of
that
will
be
bin
lib.
You
know
yadda
yadda
for
the
software
that
goes
along
with
that.
A
B
But
a
lot
of
stuff
like
if
you
install
software
off
of
the
extra
packages
for
enterprise
linux
repository,
for
example,
it's
not
going
to
put
it
knobbed,
no,
it's
going
to
put
it
in
bin
or
lib
or
you
know,
wherever
else
it.
It
goes
into
the
file
system,
directory
structure
and
that's
kind
of
just
the
way
it
is
yeah.
So
and
there's
not
a
one
good
approach
like
you
know,.
B
B
A
B
I'm
taking
root,
so
this
is
the
root
user
accounts
home
directory.
We
don't
put
it
in
slash
home
because,
as
chris
pointed
out
earlier,
it's
flagged
as
optional
also
a
lot
of
times.
It's
set
up
as
a
separate
file
system,
so
it
goes
to
like
different
storage
yeah
and
if
your
system
was
having
problems
and
for
whatever
reason
that
storage
wasn't
available
when
you
log
in
as
root,
you
still
need
to
maybe
have
access
to
your
own
stuff.
B
So
slash
root
goes
as
part
of
the
slash
file
system
and
goes
on
the
storage
of
the
slash
file
system.
Other
unix's
will
actually
not
put
this
out
there.
So
when
you
log
in
as
root
on
solaris,
for
example,
you
get
dropped
into
slash
as
your
home
directory,
which
means
when
you
download
a
file
or
your
desktop,
wants
to
create
things
like
downloads
and
pictures
or
whatever
it
puts
to
the
topmost
directory,
because
that's
your
home
directory
for
root,
so
things
got
kind
of
cluttered
at
that
top
level.
A
And-
and
this
is
an
interest
like
as
we're
going
through
these
I'm
pulling
up
the
docs-
this
is
an
interesting
footnote
on
root.
We
recommend
against
the
root
account
you
we
recommend
against
using
the
root
account
for
tasks
that
can
be
performed
as
an
unprivileged
user,
what
a
great
idea
and
that
it
be
used
solely
for
system
administration.
A
What
does
this
mean
to
me?
I
read
this
as
root.
Is
cli
only
don't
install
a
gui
as
root
right
and
have
mail
and
all
kinds
of
other
stuff
living
under
this
directory,
because
it
is
going
to
be
a
necessary
thing
for
when
stuff
goes
bump
in
the
night?
You
want
that
to
be
there
right
and
only
really
important
things
should
just
live
in
root
for
system
administration.
I
mean
but
fair.
B
It
but
there's
a
lot
of
like
workstations
that
people
own
and
I
could
see
them
maybe
using
root
on
that,
but
again
for
system
administration
stuff.
I
think
you
typically
will
operate
on
those
boxes
as
a
regular
user
for
a
whole
host
of
reasons,
not
just
security
and
file
ownership
and
a
variety
of
other
things.
A
No,
we
have
a
comment,
slash
question
from
rapscallion
reeves,
our
friend,
and
we
now
be.
We
might
not
be
able
to
answer
this
because
I
don't
know
the
answer.
Maybe
you
just
got,
why
are
bsd
and
linux's
file
trees
so
similar,
but
so
different?
I
don't
get
why
bsd
has
user
local
bin
and
user
local
etsy,
for
example,.
B
I
think
it
ties
to
the
history
of
how
we
got
here
right.
So
once
upon
a
time
there
were
two
large
unix
distributions,
berkeley
and
system
five
and
as
they
matured
and
they
wanted
to
woo
users
from
one
to
the
other
or
back
again
they
had
to
build
up
this
kind
of
cross-populated
directory
structure
horribleness
so
like.
B
If
you
were
a
a
berkeley
person
right,
because
you
started
on
solaris
when
it
was
solaris
four
dot,
something
something,
and
then
your
company
decided
that
you
were
going
to
be
an
hp,
ux
shop
which
was
system
five.
So
you
log
into
your
first
hp
x,
machine
you're
like
where
is
everything
so
hp,
ux
or
aix?
For
that
matter
too,
from
ibm,
they
started
to
create
these
kind
of
additional
directories
to
make
it
more
comfortable
for
people
coming
from
berkeley
and
then
sun
was
like.
B
And
so
when
you
look
at
the
commercial
unix's,
I
think
a
lot
of
them
end
up
cross-populating
between
the
berkeley
and
system.
Five
standards
and
then
linux
comes
along
and
it's
the
same
deal.
They
have
to
appeal
to
people
who
maybe
were
used
to
working
in
a
variety
of
unixes,
so
they
created
this
much
more
amalgam
of
directory
structures,
but
bsd
is
still
a
legitimate
unix
distribution
with
a
couple
of
different
flavors
and
they've
stayed
pretty
true
to
their
their
file
system
hierarchy,
standard
that
they
minted
back
in
the
day.
And
so
that's.
A
Cool,
so
I
got
a
question
here:
you'll
know
this
better
than
I
do
when
is
well.
Will
rel
migrate
to
butter
fs.
B
So
we
shipped
butter,
fs,
aztec
preview
for
the
duration
of
route,
7
right
and
then
you'll
notice
that
it's
not
included
with
rel8.
B
Yeah,
like
the
biggest
thing
about
file
systems,
nobody
likes
it
when.
A
B
So
you
know
you
have
to
make
sure
that
that
stuff
is
rock
solid
like
somebody
was
just
recently
was
asking
me
about
file
system
driver
for
linux,
and
we
do
not
ship
that
as
part
of
route.
You
can
get
it
in
places
like.
I
think
people
may
have
it
yeah,
so
you
can
get
a
module
and
it'll
plug
into
your
kernel
and
stuff.
B
But
that's
because
there
is
a
extremely
unlikely
situation
where
you
can
lose
data,
and
so,
if
you're
mounting
an
ntfs
file
system
into
your
linux
directory
structure
and
then
data
starts
vanishing,
no
bueno.
A
A
We
like
to
make
sure
your
data
is
as
stable
and
intact
as
possible
as
we're
going
about
making
your
operating
systems
so
yeah
I
mean
there's,
there's
a
lot
of
work
being
done
in
butter.
Fs
fedora
is
using
butter
fs,
I'm
using
that
screen
right
there,
I'm
using
butter
fs.
I
don't
have
anything
of
significance
that
I
cannot
replace
easily
on
that
box,
for
example,
though,
as
a
result.
B
Yeah
yeah,
so
I
know
that
our
product
manager
for
file
system
stuff
is
continued
to
keep
an
eye
on
rfs.
So
I
I
don't
remember
if
we
put
it
into
centos
stream,
nine
centos
stream.
Nine
was
had
their
first
build
just
like
a
week
ago,
and
that
should
be
a
pretty
good
representation
of
what
we
expect
to
land
in
in
rel
next
year.
B
When
we
release
rail
line,
I
have
not
looked
for
butt
rfs,
but
that
would
be
if
I
was
going
to
look
to
see
if
it
was
maybe
going
to
land
in
there.
That's
right.
First
mm-hmm.
B
Yeah
and
centos
stream
keeps
a
pretty
close
relationship
with
rel.
So
when
you're,
looking
at
centos
stream
you're
looking
at
a
pretty
good
representation
of
what
you
can
expect
in
the
next
version
of.
A
Oh,
so
espin
is
another
one
of
those
throwbacks
I
feel
like.
Maybe
maybe
not.
I
don't
remember,
which
came
first.
A
It
definitely
was,
and
it
came
from
the
need
to
have
like
back
in
the
day
you
partition
the
hell
out
of
disks.
Nowadays,
you
might
not
so
much,
it
might
be
one
or
two
partitions
right,
so
you
would
want
s-bend
to
be
in
a
similarly
fast.
You
know:
file
system,
slash
space
on
disk
as
boot
was
right,
because
you're
gonna
need
fsck
at
boot.
Time.
Sometimes
right
so
like
there
was
that
for
a
little
bit
and
people
kind
of
went
away
from
that,
as
this
got
faster.
B
So
like
partitioning
disks,
yeah
exactly
and
like
way
way
back,
we
could
do
things
like
change
the
permissions
on
it,
so
that
people
couldn't
access
them,
but
we
also
for
regular
users.
Don't
include
the
s-pen
directories
on
their
path,
so
when
they're
looking
for
commands,
they
probably
won't
find
these
with
things
like
tab
completion
because
they're,
not
system
administrators,
they
don't
get
the
aspen
directory
and
then
over
time
we
we
figured
out
that
that
wasn't
a
great
plan,
because
people
would
still
add
it
to
their
path
and
find
it
anyway.
A
Yeah
question
in
chat
here
are
some
links
between
ben
and
espin.
Do
they
exist
like
they
are
between
ben
and
user
ben?
That's
a
good
question.
B
So
I
was
gonna
handle
the
sim
link
question
at
the
end,
because
that's
a
change
that
we
made
in
rel.
Seven,
yes-
and
I
was
gonna
flip
back
over
to
my
actual
box,
speaking
of
which
I
should
probably.
B
Right,
so
if
you
look
in
our
our
directory
here,
right
you'll
see
that
bin
lib,
lip
64
and
s
been
are
all
symbolic
likes.
B
And
they
point
to
their
user
counterparts,
so
this
is
a
functionality
of
combining
the
linux
file
system
hierarchy,
standard
out
of
berkeley
and
system
five.
So
in
berkeley
they
put
all
of
the
bin
lib
s
bin
at
the
top
level
root
directory,
whereas
in
system
5's
directory
file
system
hierarchy
standard,
they
put
it
under
user
for
unix
system
resources
and
then
it
became
like
well
as
time
went
on
and
we
started
to
cross
populate
between
those
two
directory
structures.
B
B
We
just
went
ahead
and
merged
them
together
and
created
these
sim
links.
So
in
reality
everything
in
rel
goes
into
user
and
then,
if
your
package,
when
you're
you
know
installing
an
rpm
or
whatever
says,
slash
bin,
that's
fine.
It
just
follows
the
sim
link
and
puts
it
in
user
bin.
So
yeah
we
created
those
sim
links
to
still
maintain
that
compatibility.
A
And
I
missed
this
question
from
rapscallion
reeves.
One
thing
I
still
don't
understand
about
fedora,
34
and
butterfs,
since
compression
is
enabled
by
default.
If
you
check
the
encryption
box
during
f34
install
will
that
mess
you
up?
How
can
butter
fs
encrypt
compress
encrypted
data?
Let's
and
let's
not
conflate
compression
and
encryption
right,
they
are
vastly
different
things.
An
encryption
algorithm
masks.
A
The
data
using
the
you
know,
prescribed
algorithm
compression
uses
a
different
algorithm
to
make
more
space
on
the
disk
right,
like
that's,
essentially
the
way
to
look
at
it,
but
rfs
uses
compression
in
the
file
system,
which
should
be,
in
theory
more
efficient
than
like
using
external
processes
to
make
that
compression
happen,
and
then
decompress
and
recompress
in
and
out
kind
of
deal.
A
But
encryption
can
happen
before
or
after
compression
right,
like
there's
no
wrong
order
to
do
that
in,
but
you
have
to
decrypt
a
disk
before
you
can
load
it,
so
it
usually
is
compressed
and
then
encrypted
correct.
Or
am
I
wrong
in
saying
that.
B
I
think
I
think
that's
mostly
mostly
right.
Okay,
the
one
place
where
you
might
get
into
some
weirdness
is,
if
you
encrypt
first,
then
compress
because
oh
yeah,
when
you
compress
stuff
things
like
text,
compress
really
well
right,
whereas.
A
A
B
B
Kpatch,
maybe
I
think,
but
he
is
also
the
data
services
product
manager
for
rel,
so
he's
the
one
when
I
mentioned
somebody
was
keeping
an
eye
on
it,
he's
the
one
who's
keeping
an
eye
on
it,
and
that
might
be
our
next
stop
to
ask
these
kinds
of
questions.
Yeah.
A
B
See
yeah
and
then
in
terms
of
like
file
system
on
rel
by
default
we
use
xfs,
and
but
we
still
offer
ext4
as
an
alternative
and
the
big
difference
there
is
things
like.
B
I
know,
to
look
up
tables
and
numbers
of
files
that
one
can
have
in
a
directory.
Xfs
just
does
better,
with
small
files,
lots
and
lots
of
small
files
than
ext4,
but
is
about
the
same
for
normal
file
workloads.
B
So
we
have
some
folks
that,
like
run
their
web
store
using
rel
and
there
they
have
like
a
gajillion
tiny
files,
because,
like
each
product
review
is
a
file
right,
each
like
suggested
product
might
be
a
file
and
so
for
each
product
that
they
sell
in
their
store
and
there
might
be
like
thousands
of
files
underneath
of
it.
A
B
Yeah
and
if
you're
interested
in
things
like
compression
and
saving
space,
you
may
also
be
interested
in
taking
a
look
at
vdo,
which
is
the
virtual
data.
Optimizer
that
comes
with
rel,
essentially
uses
data
deduplication,
yes
to
right
to
not
store
the
same
bits
on
disk
multiple
times,
but
all
right.
So
I
think
it's
my
turn.
Srv
is
one
that
we
don't
really
use
in
rel,
but
like
debian
distros
use
a
lot.
B
A
A
B
Now
so
we
talked
about
it
a
couple
weeks
ago
we
were
talking
about
tunables
because,
like
proc,
it
is
an
in-memory
file
system
right,
and
so
you
have
things
there,
like
your
preference
for
using
huge
pages
in
the
kernel
or
power,
consumption,
settings
or
block
device
tools.
A
A
Okay,
temp
is
almost
exactly
what
you
think
it's
for.
It
is
for
temporary
files,
but
the
different
temp
has
a
relative
in
var
temp,
which
will
actually
purge
itself
on
boot.
So
if
you're
putting
files
in
temp,
they
will
last
through
a
reboot
where
var
temp
they
may
or
may
not,
depending
upon
the
os.
B
Yes,
depending
on
the
os
for
rel,
they
do
persist
across
right,
but
like
temp,
there
is
a
service,
that's
run
as
a
crown
job
right
that
will
remove
stuff
from
vartep
as
well.
A
And
I
use
temp
for,
like
literally
the
output
of
scripts,
that
I'm
running
as
like
a
diagnostic
type
thing
or
if
I'm
running
an
ansible
playbook,
and
I
need
to
actually
write
a
file
to
disk
and
then
I
know
I'm
going
to
get
rid
of
it
later.
Temp
is
where
I
do
it
right,
so
anything
that
I
know
I'm
going
to
have
to
have
some
output
that
has
to
land
on
disk,
and
then
I
want
to
erase
it
later.
That's
what
temp
is
for.
B
Yeah
and
I
think
by
default
stuff
gets
erased
from
temp
every
10
days
if
it's
not
been
used.
Oh
really,
thanks!
Yeah,
thanks
to
the
temp
watch
service,
that's
running
through
chron.
B
So
basically
it
like
goes
into
temp
and
looks
at
all
the
files
and
directories
in
there
and
looks
the
last
access
timestamp
and
the
last
access
timestamp
was
older
than
10
days
ago.
It
gets,
killed
off
and
removed.
B
A
Var
is
one
of
my
favorite
directories
because
there's
so
much
going
on
right,
like
you
know,
var,
logs
var
cash
var
crash
var,
lib
var,
lock,
like
everything
kind
of
like
running
on
the
system,
is
under,
like
the
outputs
of
that
is
all
under
var
right.
So,
if
you're
looking
for
logs
somewhere
go
look
in
var
logs,
if
you're
looking
for
lock
files
that
you're
trying
to
get
rid
of
go
looking
for
lock,
but
the
the
the
purpose
of
like
var
itself,
it
contains
variable
data
files.
That's
literally
why
it's
called
var.
A
This
includes,
like
spool
directories,
so
print
jobs,
emails
things
like
that
administrative
logging,
data,
transient
files,
temporary
files.
One
thing
to
point
out
is
that,
like
some
things,
living
in
var
aren't
shareable
right
across
systems.
So
just
keep
that
in
mind.
Right,
like
you,
don't
want
to
mount
var
log
to
the
same
thing
on
every
system,
unless
you
have
a
way
to
take
care
of
that
output.
A
B
B
A
A
A
Well,
I'm
I'm
trying
to
keep
my
place
in
the
slides
here
if
that
makes
sense,
but
yeah
like
var,
shoot.
I
was
just
slipped
my
brain.
I
I
spent
a
lot
of
time
in
var,
working
with
databases,
for
example
right.
B
A
That's
where
your
data
from
your
databases
will
land,
and
you
want
to
make
sure
that
that
far
directory,
especially
for
that
database,
is
fast
very
fast
right,
like
you,
want
var,
to
kind
of
be
your
fastest
disk
at
times.
For
this
reason,.
A
So,
let's
get
all
right:
let's
just
get
all
these
questions
answered.
Would
you
consider
testing
zfs
on
fedora,
I
mean
maybe.
B
So
zfs
is
not
it's
not
open
source
well
or
there's.
A
A
We
had
an
oracle
storage
array,
that's
the
only
thing
oracle
we
bought
because
it
used
zfs
and
it's
a
really
impressive
file
system,
but
it
is
very
tightly
licensed
in
the
sense
of
like
it's
incompatible
with
open
source
software
to
some
extent
so
yeah,
that's
just
something
to
keep
in
mind.
A
Linux
has
something
for
zfs,
like
it's
making
its
way
out
into
the
world.
So
just
keep
that
in
mind.
It's.
A
A
Right
and
that's
kind
of
the
idea,
but
yeah,
oh
boy,
so
one
of
my
co-workers
just
got
another.
A
A
I'd
really
like
to
see
butterfest
be
the
new
zfs.
Yes,
I
would
too
we
covered
the
ram
thing
I
thought
exported
things
like
nfs
should
come
from
slash,
srv
sort
of
like
the
opposite
of
mount.
Well,
that's
a
good
point.
Where
do
nfs
mounts
land
typically.
B
So
it
depends
on
what
you're
trying
to
share
right.
If
you
were
trying
like
back
in
the
day,
we
would
do
things
like
nfs
share
our
user
directory,
because
we
would
install
matlab
on
a
server
and
we
want
other
client
boxes
to
be
able
to
mount
and
execute
it
locally,
and
so
we
we
share
user
as
an
nfs
share.
B
I
think
the
biggest
thing
on
rel
for
nfs
is
making
sure
that
it's
got
the
right,
sc
linux
contexts
to
be
shared.
Yes,
so
yeah,
I
think
I
still
end
up
putting
stuff
under
mount.
B
So
that
I
can
context
it
correctly
and
update
my
selinux
policy
so
that
it
knows
that
things
in
that
directory
gets
get
done
like
that.
But
you
could
do
the
same
thing
in
in
srv.
B
A
A
thing
a
standard,
but
you
can
deviate
from
that
standard
and
some
of
that
standard
is
optional
and
so
forth.
So
on
do
okay!
Here's
another
question:
do
you
use
x4
or
butter
fs
in
your
daily
life?
A
That's
just
because
I'm
using
fedora
more-
and
you
know
rel
less
here
in
the
house,
but
that
doesn't
mean
that,
like
when
I
muck
with
servers,
I'm
probably
touching
xfs
in
general.
Nowadays,
I
feel
like.
B
Looking
at
this
top
level
directory,
I'm
using
colorized
ls
output,
do
you
know
why
temp
is
green
temp.
A
Just
ls-ld,
temp
and
you'll
see
why
it
has
the
t
at
the
end
there,
the
the
wt,
the.
A
A
B
B
Yeah
anyone
can
write
to
it,
but
the
sticky
bit
says
that
only
the
owner
of
a
file
or
directory
may
remove
it.
B
A
Doesn't
this
also
mean
that
it
will
inherit
the
these
settings
all
the
way
down?
No
matter
how
many
directories
you
create,
underneath
it.
B
That
is
the
ownership.
Well
right,
sgid
there.
A
B
A
I
guess,
instead
of
christian
hernandez
so
andrew
says
I
probably
unfairly
shy
away
from
butter
fs
as
a
result
of
the
now
quite
old
data
corruption
bugs
I
stopped
paying
attention
to
it
back
then,
but
I
understand
there's
quite
a
few
commercial
products
using
it
now,
including
synology.
He
thinks
that's
interesting
to
learn.
If
that's
the
case,
if
your
synology
box
is
running
butter
fs,
let
me
know.
B
So
things
like,
I
was
just
talking
to
somebody
about
satellite
six,
oh
yeah,
okay,.
B
Were
they
were
like
it's
like,
but
but
it's
fine
now.
A
A
B
I
don't
remember
the
version.
You
could
only
have
sixty
four
thousand
five
hundred
thirty
five
files
under
directory
before
it
would
like.
B
And
xfs,
like
it's
ridiculous,
amounts
that
I
would
challenge
someone
to
actually
achieve.
It's
like
millions
of
files
in
that
in
a
single
directory
level,
and
I
can
retrieve
it,
but
because
it's
better
at
looking
up
and
retrieving
those
tiny
large
amounts
of
files,
that's
kind
of
what
sweet
spot.
Oh.
B
Yeah
yeah
yeah
yeah,
so
I
actually
actually
have
a
lot
about
that.
B
B
And
essentially
it's
like
installing
vdo
and
actually
using
it
to
figure
out
how
much
data
is
really
being
stored
when
you
are
interacting
with
the
video
devices
so.
B
We
don't
have
time
to
go
through
all
that
at
the
top
of
the
hour,
but
yeah
vdo
is
implemented
through
a
kernel
module.
B
So
you
want
to
make
sure
that
you
are
running
the
same
kernel
module
that
you
have
on
your
running
kernel,
and
so,
whenever
this
provisions
you'll
see
that
very
first
command
here
is
the
yum
install
and
it
has
a
video
version
and
I
actually
specify
the
version
in
the
command
and
the
kvdo
kmod
also
has
a
version
that
I
specify
because
it
all
has
to
match
the
currently
running
kernel
right.
So
if
I
do
a
umay
r.
B
The
the
lab
vendor
actually
rebuilt
my
labs
without
me,
asking
them
to
a
newer
version,
a
newer
update
release
of
rel83,
so
it's
like
834
and
I
did
not
go
through
and
test
all
the
labs
after
that,
we're
about
to
update
them
to
rel
8.4,
because
that
was
released
today.
A
A
No
you're
fine,
we
mentioned
it
on
an
earlier
show.
So
that's
what
made
me
like
be
like.
Oh
I'm,
gonna
mess
with
scott,
but
yes,
rel
8.4
was
released
today
just
a
few
hours
ago,
as
a
matter
of
fact,
so
yeah
go
check.
It
out
tires
on
it
on
your
desk
systems.
Please.
B
It
is
working,
so
that
is
the
right,
the
right
kernel,
module
and
video
version.
B
Normally,
when
you
run
the
video
command
it'll
like
error,
if
you
have
the
wrong,
if
you
have
a
mismatch
for
your
module
number
nice
so
but
yeah,
the
video
stats
is
the
command
that,
like
shows
you
how
much
you're,
using
and
as
you're
like
storing
files,
that'll
report
on
how
much
is
being
used
and
that's
where
you
can
track
the
data
deduplication,
that's
being
managed.
B
Cool,
so
I
think
yeah.
A
I
thought
we
were
gonna,
call
it
a
day
yeah.
I
hope
this
was
educational
for
folks
right
like
if
you
have
any
questions,
feel
free
to
reach
out
to
me
short
at
redhat.com
or
chrisshort
on
twitter,
and
you
know
we're
very
happy
to
have
you
here
for
this
episode
and
we
look
forward
to
seeing
you
tomorrow
at
9.
00
a.m.
To
talk
about
same
show,
different
name,
the
data
services
office
hour.