►
Description
Join Javier Pena and Yatin Karel for an overview of RDO's Packstack project. Learn more about this project and how to install it, tricks for configuration, and how you can use it for Proof of Concepts and learning more about how to use OpenStack through the dashboard and the CLI.
A
B
Amy
hello,
hi,
everyone,
I'm
javier
pena,
I'm
a
principal
software
engineer
in
red
hat
and
we
are
going
to
discuss
backstack
today,
so
we're
going
to
see
what
backstack
is
we're
going
to
see
also
what
it
is
good
for
and
what
it's
not
a
good
fit
for
packstack
some
history,
so
we
can
be
back
stacking
context
and
finally,
we'll
have
a
demo,
and
since
the
demo
takes
a
few,
some
time
to
finish
will
have
an
ask
me
anything
in
the
meantime,
no
backstack
is
a
simple
rdo,
installer
utility,
some
people
confuse
rdo
with
backstack
or
rdo,
with
triple
o.
B
So
we'll
just
discuss
it
here.
So
rdo
is
a
distribution
of
openstack
based
on
what
the
openstack
community
provides.
We
just
package
some
of
the
openstack
projects
into
rpm
packages
and
then
distribute
them.
Backstack
takes
care
of
it.
They
are
setting
up
an
openstack
cloud
cloud
based
on
on
those
packages.
Well
triple
o
is
another
utility
we'll
discuss
it
later,
so
pack
stack
is
based
on
the
puppet:
open
stack
module,
so
we
are
building
on
top
of
the
openstack
community
efforts
in
this
ca
in
this
site.
B
We
we
also
participate
in
the
puppet
openstack
community
by
maintaining
some
of
the
public
modules
and
also
helping
the
the
community
with
tests
on
top
of
rdo
and
it's
mostly
a
good
fit
for
ci
and
test
deployment.
So
the
pack
stack
features
make
it
a
very
good
utility
to
do
test
deployments
rather
than
production
deployments
and
we'll
see
how
and
we'll
see
why
so
tax
tag
is
is
great
for
simple
and
quick
deployments.
You
will
see
that,
so
you
can
start
with
a
centos
machine
and
you
probably
get
openstack
running
in
less
than
an
hour.
B
B
B
Oh,
that
sounds
nice,
but
there
are
also
some
cases
where
tax
stack
shouldn't
be
used.
So,
for
example,
for
h8
deployments,
we
shouldn't
use
black
plugstack
to
deploy
rdo
because
there
is
no
redundancy,
the
components
deployed
by
pak
stack.
So
in
the
openstack
world
we
have
the
concept
of
controller
node,
where
you
can
set
up,
for
example,
the
machine
that
controls
all
the
services
in
openstack.
B
So
packstack
will
only
deploy
one
machine
for
that.
We
cannot
have
multiple
machines
and
then
have
failover
in
case
one
of
them
fails.
It's
also
not
meant
for
for
production
deployments,
because
backstack
lacks
some
production
by
the
features.
It's
meant
to
be
a
quick
and
easy
tool
to
spin
up
an
openstack
cluster,
but
we
don't
have
all
the
features
that
you
will
want
to
have
on
a
production
deployment.
We
don't
have
ssl
everywhere,
for
example,
also
for
day
two
activities:
it's
not
a
great
fit,
because
there
is
no
support
for
upgrades.
B
For
example,
you
know
you
cannot
upgrade
from
a
train
openstack
deployment
to
a
usuri.
You
can
stack
deployment
using
backstack.
B
It's
another
community
project
that
is
specifically
made
to
to
to
work
on
all
those
environments.
So
it's
meant
to
have
a
cha.
It's
meant
to
be
production,
ready,
it's
meant
to
have
the
two
activities.
So
this
is
you
see
that
we
have
different
targets
for
each
of
the
tools.
B
Just
some
quick
figures
about
black
stack:
it's
been
eight
years
and
one
day
since
the
first
commit
the
project
was
originally
based
from
ovid's
installer
and
then
it
grew
up
on
its
own.
But,
as
you
see
it's
been
around
for
almost
forever,
we
have
like
137
people
contribute
over
the
last
eight
years.
Right
now
we
have
a
team
of
three
people
who
are
the
core
reviewers.
B
Now,
let's
focus
on
the
demo,
which
is
the
funny
part
in
this
case,
we're
gonna,
we're
gonna,
use
a
virtual
machine,
a
simple
virtual
machine
for
the
test,
it's
a
small
one,
with
just
two
cpus
and
eight
gigs
of
ram
and
some
swap
with
centos
eight.
And,
of
course,
we
will
need
internet
connectivity
because
we
need
to
download
the
the
rdo
packages
from
from
somewhere
okay.
B
B
Yum
install
center
release,
openstack
usury,
and
this
will
set
up
the
rdo
repositories
for
the
usuri
openstack
release
for
us
for
centos
eight
we
can.
We
can
deploy
train
or
usuri,
which
are
the
two
latest
of
the
stack
releases
for
centos
seven.
We
can
start
you
can
install
train
and
all
earlier
releases
and
that's
because
centos
7
has
very
limits.
Some
very
limited
support
for
python
2
and
the
obvious
document
d
dropped
support
for
python
2
in
in
the
city.
So
we
cannot.
We
cannot
try
to
deploy
usury
on
7.
B
B
B
Finally,
in
this
case,
in
this
demo,
we
are
going
to
set
up
a
button,
an
openstack
environment
using
ovn,
with
open
virtual
networking.
It's
one
of
the
it's
currently
the
yeah,
the
default
network
component
for
neutron
in
openstack
and
we're
going
to
set
up
an
external
network
so
that
vms
will
create
inside
this
vm
and
access
the
internet.
B
B
By
default,
centos
eight,
we
only
comes
with
network
manager,
but
some
of
the
operations
we
we
do
in
pak
stack
to
to
give
us
access
to
the
external
network
require
the
network
scripts.
So
we
will
disable
network
manager
and
enable.
B
A
B
So
so
we're
going
to
pre-generate
it
because
the
default
all-in-one
installation
by
packstack
that's
some
things.
We
don't
want
to
to
do
in
this
case,
for
example,
it
will
it
will
not
create
the
external
network.
The
way
we
want
it,
so
our
vms
would
not
have
access
to
the
internet.
B
B
B
So,
as
sami
mentioned,
if
we
just
wanted
to
set
up
a
very
quick
environment
and
and
let
it
go,
we
could
just
do
backstack
minus
minus
all
in
one,
and
this
would
use
a
set
of
default
values
for
everything
and,
just
after
a
few
minutes
create
the
the
openstack
environment
for
us,
and
that
would
work,
however,
especially
for
the
network
configuration
and
the
network
configuration
it
creates,
is
pretty
simple.
It
doesn't
allow
us
to
to
exit
the
this
vm,
so
we
will
generate
an
answer
file
and
go
through
some
options
before
running
the
installer.
B
B
So
this
is
the
answer
file.
We
have
a
lot
of
options.
This
is
a
simple
ini
file,
so,
for
example,
at
the
beginning,
we
have
all
this
config,
whatever
install.
This
allows
us
to
enable
and
disable
openstack
services
in
general.
I
want
to
keep
them
all,
but
first
I
will
disable
all
the
all
the
metering
parts
so
kilometer
and
out
will
be
disabled,
but
let's
say
I
want
to
have
openstack
orchestration,
so
I
will
enable
heat
installation.
B
Another
important
parameter
we
have
at
the
very
beginning
is
the
number
of
service
workers.
This
is
useful,
especially
if
you
have
a
vm
with
a
with,
without
a
small
amount
of
ram
and
a
lot
of
processors,
because
by
default
pack,
stack
will
spawn
as
many
processes
for
each
service
as
processors
are
in
the
system.
So
if
we
have
a
system
with
eight
processors,
eight
logical
processors
or
16
with
hyper
threading,
and
all
that
it
will
try
to
create
so
many
processes
for
each
service
that
they
will
eat
up
all
ram
in
our
system.
B
So
we
we
may
want
to
reduce
this
one
and
set
one
or
two
for
the
number
of
service
workers.
In
the
case,
in
the
case
of
this
vm,
we
have
just
two
cpus,
so
that's
that's
fine
and
we
can
let
the
default
value.
B
We
can
so
that
allows
us
to
go
beyond
a
single
single
machine,
but
that's
something
we're
not
gonna
do
today
and
then
we're
gonna
go
for
the
ovn
piece,
so
the
default
options
for
network
are
like
this,
so
we
just
have
a
single
bridge
mapped
to
the
external
network,
but
nothing
else.
B
C
B
Next
thing
we
want
to
do
is
we
do
not
want
to
provision
a
demo
project
for
us.
We
will
see
later
that
once
we
connect
we
can
connect,
we
will
have
an
admin
user
and
from
there
we
can
use
multi-tenancy
to
create
different
projects
and
different
users
with
access
to
different
projects,
and
all
those
projects
will
be
isolated
from
each
other,
so
we
can
have
true
multi-tenancy.
B
C
B
And
just
wait
so
now
we
will
see
that
it
is
preparing
some
some
stuff,
it
will
install
some
packages
and
then
it
will
run
like
in
three
different,
separate
steps:
the
controller,
the
networking
and
the
compute
parts.
A
Yeah
so
phil
and
I
have
been
talking
about
the
commands
to
generate
the
answers.txt
and
how
to
then
call
it.
So
if
you
could
show
that
in
the
terminal
and
then
he
had
a
question
about
ovh
folk
ova,
obh
folks
being
in
the
pax
stack
community
and
the
community
is
open
to
everyone
who
would
like
to
participate
and
people
come
and
go
as
far
as
their
schedules
and
their
companies
allow
or
there's
people
like
me
who
continue
to
be
part
of
the
community,
no
matter
what
company
they're
working
at
or
their
job
title.
A
So
there's
no
way
of
for
us
to
say
why
someone's
no
longer
in
the
community
or
they
could
be-
and
we
just
don't
know
their
nick
on
irc
to
say
oh
yeah
so
and
so
is
no
longer
part
of
it
because
he
moved
on
to
another
company
or
position
and
so-and-so
took
their
place.
You
know
so
that's
part
of
being
an
open
source.
Community
people
do
come
and
go.
Companies
do
come
and
go.
B
B
A
Openstack
otherwise
you're
going
to
have
the
self-generated
hard
to
remember
passwords.
C
B
Yeah
right
so
so
this
is
it
so
you
have
this
config
default
password
option
that
you
can
set
to
assign
a
simple
password
everywhere,
instead
of
the
randomly
generated
ones.
Actually,
I
have
I'm
quite
fond
of
this
configuration
option,
because
this
was
the
first
one
I
my
first
real
code
stuff
I
added
into
backstage
when
I
joined
the
team
like
five
years
ago,.
A
A
Now
double
check
irc,
because
bill
was
asking
questions
over
there
earlier
just
an
earlier
question
from
irc,
and
he
was
talking
about.
You
know
the
difference
between
rdo
and
triple
o,
and
we
explained
that
earlier
like
because
even
I
used
to
as
just
go
pack
stack
equals
rdo.
So
it's
our
pack
stack
triple
o,
as
well
as
the
rpms
and
the
repos
themselves
that
people
can
use
for
the
manual
installs.
A
B
B
What
is
what
it,
what
is
100,
not
supported
and
won't
work
and
will
break
your
system
is
if
you
try
to
run
backstack
minus
minus
all
in
one
twice,
because
the
second
time
it
will
generate
a
completely
new
answer
file.
So
all
passwords
will
be
different
from
the
first
time
and
then
everything
will
crash.
It
will
try
to
change
to
access
mariadb,
for
example,
with
the
same
password
as
it
has
generated,
because
it
would
detect
that
marietv
is
running
and
it
just
will.
It
will
fail.
A
B
Right,
that's
very
important,
because
if
something
fails
during
the
first
run,
maybe
there
was
a
an
issue
connecting
to
the
centos
mirrors
or
whatever.
Then
it's
relatively
safe
to
just
retry
pack
stack
minus
minus
sensor
file
with
the
generated
answer
file
on
the
first
run.
But
if
we
try
to
generate
a
second
answer
files,
then
everything
will
crash.
C
A
A
Now
one
thing
I
will
add
about
pacstack:
if
you
reboot,
your
machine,
packstack
will
come
back
up.
There
are
issues,
at
least
in
the
past
that
were
with
devstack,
we
used
to
say
it
was
a
one-time
shot
use.
The
pack
stack
has
a
little
bit
more
of
a
shelf
life
for
your
proof
of
concepts
than
say
devstack
does
because
devstack's
real
purpose
is
just
for
you
to
do
your
code
test
it
and
then
tear
it
down.
B
Yeah
absolutely-
and
that's
actually
that's
why
I
I
didn't
want
to
to
go
with
the
defaults
and
and
use
these
this
additional
ovn
and
and
disabling
the
demo
tenant
options,
because
when
you
set
up
the
and
you
set
up
the
networking
like
we
did
in
in
the
demo,
this
will
be
preserved
across
reboots.
B
A
A
Maybe
I
know
cloudnald
names,
all
his
machines
compute,
which
kind
of
confused
me
when
we
were
working
through
something
the
other
day
because
he
was
on
his
hypervisor
and
yet
it
was
called
compute.
So
I'm
thinking
that's
what
it
could
mean
is
that
centos
sets
it
up
as
a
hypervisor
and
call
and
is
referring
to
it
as
a
compute
versus
the
openstack
compute
controller
type.
B
B
B
So,
instead
of
just
seeing
this
the
screen,
I
mean
it's
nicer,
but
it
doesn't
tell
us
a
lot,
so
we
could
just
go
and
check
what
the
what
the
actual
puppet
code
is
being
run
at
the
moment.
A
B
A
B
A
B
Yeah
yeah
that
that's
mostly
first
there's
much
much
many
more
features
than
there
used
to
be,
and
then
we
have
usually
with
a
default
installation.
It's
usually
related
to
the
amount
of
ram
required
by
each
worker
for
each
service.
B
So
at
the
beginning,
by
default,
each
openstack
service
used
to
start
like
one
two
processes,
and
if
you
wanted
to
have
more
than
you
would
have
to
change
settings
manually
for
each
of
the
configuration
files
and
now
studio,
they
will
run
now.
If
you
have
16
cpus,
for
example,
it
will,
it
will
spawn
16
processes
per
service.
Those
16
neutrons
processes,
for
example,
make
things
much
faster,
but
on
the
other
hand,
they
eat
up
a
lot
of
ram.
D
So,
with
respect
to
setting
up
reports,
I
would
like
to
add
one
thing
that
we
should
avoid
mixing
the
rapports
additional
reports
like
e-pail
or
rpm
factory
with
the
audio
records,
because
they
do
not
play
well.
You
can
use
those
wrappers,
but
before
enabling
audio
repo,
you
should
disable
those
as
there
might
be
some
issues.
Look
at
the
package.
B
A
A
A
B
Well,
we
try
to
keep
resource
consumption
to
a
minimum
from
the
backstack
point
of
view,
but
of
course
I
mean
it
will
only
require
as
many
resources
as
openstack
services
do
so.
But
if
openstack
services
keep
adding
features,
and
that
requires
some
more
ram,
there
is
almost
nothing
we
can
do
to
prevent
them
from
doing
so.
A
A
B
D
B
Four
gigs
of
ram-
that's
yes,
that's
gonna
be
complicated.
I
think
six
is
doable
six
gigabytes
with
enough
swap,
but
less
than
that
might
be
tricky.
C
B
A
B
B
No,
no,
no
not
at
the
moment,
and
we
would
welcome
patches.
We
had
actually
we
had
one
colleague
who
was
playing
around
with
octavia
some
time
ago,
but
never
it
never
finished.
And
now
that
we
are
supporting
ovn
as
the
default
networking
mechanism,
it
should
play
along
nicely,
but
we
need
someone
to
contribute
the
code.
I
would.
I
would
be
more
than
happy
to
review
that.
A
So
if
anyone
would
like
to
submit
patches
for
octavian
designate,
the
team
is
willing
to
review
them
and
get
them
included,
and
that's
part
of
community.
If
you
notice
something,
that's
missing
talk
to
the
maintainers,
the
cores
get
help
you
know
with
a
spec
or
a
blueprint
or
just
the
patch
itself
and
commit
it.
B
B
Let's
say
for
us
in
the
drive
specific
cinder
driver
for
third
party
hardware,
and
in
this
case
it
gets
more
complicated,
because
we
have
no
way
to
really
test
that
code
and
see
how
it's
working.
When,
in
those
cases,
we
used
to
ask
people
who
want
to
contribute
such
support
to
provide
a
third
party
ci
in
openstack,
and
that
usually
means
takes
some
time.
It
depends.
A
A
B
Yeah,
that's
good,
so
the
first
service
we
will
see
is
keystone.
That's
the
basic
identity,
provide
the
identity
service
for
openstack.
So
with
keystone
we
will
create
our
own
users.
B
Also,
each
service
will
have
its
own
user
and
it
defines
a
catalog
of
services,
a
service
catalog,
so
each
each
service
we
will
discuss,
will
have
its
own
url
inside
the
inside
dvn
and
we
can
use
it
to
access
all
these
services,
and
all
of
that
is
so
keystone
is
like,
like
the
cornerstone
of
obviously,
because
everything
relies
on
that.
B
Then
we
have
swift.
Lift
is
an
object,
an
object,
store
service
that
is
similar
to
s3
in
amazon,
and
that's
that's
been
pretty
successful.
Then
we
have
cinder,
is
a
it's
a
block,
storage
right,
block
storage
service.
B
It
allows
us
to
create
virtual
disks
that
we
can
then
attach
to
our
virtual
machines
glance
is
an
image
service.
It's
mostly
used
to
store
virtual
disks
images
that
we
will
use
to
to
run
our
vms.
So
let's
say
we
have
a
dentist
eight
base,
disc,
that
we
will
want
to
use
to
palm
multivolve
virtual
machines.
Then
we
will
store
it
in
in
glance.
B
What
else
neutron
is
the
networking
service?
We
can
define
virtual
networks
with
virtual
routers
builder
switches,
virtual
everything
we
will
see
a
little
bit
of
that
later.
B
B
B
The
one
is
oh
placement:
yeah
placement
is
it
used
to
be
a
part
of
nova,
and
now
it
was
extracted
as
a
separate
service.
It's
basically
a
service
to
let
no
one
know
where
it
should
place.
The
new
vms
or
new
object
it.
B
B
A
But
swift
cam
is
sold
as
a
stand-alone
product
there
and
I
think
they're
looking
to
do
that
with
glantz
as
well.
So
some
of
the
projects
end
up
spinning
out
of
openstack
into
their
own
thing.
Ironic,
which
is
the
bare
metal
deployment
project,
is
used
a
lot
of
places
where
it
is
isn't
used
as
part
of
openstack
anymore.
A
C
B
Can
yeah
it's
similar
right?
You
can
you
you
can
use
hit
so
cloud
formation
files
you
would
use
in
or
the
same
api.
You
would
have
in
amazon
to
create
your
objects
inside
an
openstack
cloud.
B
Yes,
less
than
an
hour,
as
I
said
so
so
now
we
have
a
an
openstack
cloud
running
on
our
vm,
but
we
do
want
to
create
some
some
networking
inside
this
vm,
so
we
can
create
our
vms
and
have
them
access
the
internet-
oh
yeah,
tin,
is
going
to
take
over
from
here,
and
it's
going
to
show
us
how
to
do
that
from
the
command
line.
What
do
you
think
the
stage
is
yours.
D
D
So
we
also
need
to
create
allocate
allocation
pool,
since
we
are
using
an
ip,
for
example,
sorry,
so,
first
we
need
to
source
the
rc
file
and
with
where
we
have
stored
the
credentials
for
this
openstack
setup
that
we
have
just
installed
with
the
text.
Now
we
will
create
the
network
with
the
name
public.
C
B
D
D
So
the
gateway
is
same
for
what
whatever
for
this
system
is
subnet
range.
Subnet
range
is
192
160
to
122.0
same
as
this
vm,
a
location
pool
for
which
we'll
be
using
for
our
floating
ips.
For
our
vms,
we
have
a
lot
getting
from
100
to
150
and
also
we
are
not
enabling
dhcp,
because
we
are
using
this
network
just
for
a
floating
ip.
So
whenever
vm
will
use
this
network,
there
will
be
no
dhcp
running
here,
so
no
ip
will
automatically
allocate
it
to
the
vms
okay.
D
So,
since
our
network
are
created,
we
will
create
create
the
image.
D
So,
first
of
all,
we
will
download
an
image
from
zeros
vm.
We
will
download
you
can
download
any
image
like
image
here.
We
means
that
openstack
image
we
will
be
using
to
boot
the
vm,
so
you
can
download
and
ubuntu
vm
or
c
assigned
to
svm
centers
eight
cent
plus
seven
as
per
the
need.
But
since
the
resources
are
limited,
we
are
going
to
try
the
zeros
image.
C
D
C
C
D
D
D
C
C
C
A
D
We
should
avoid
to
combine
all
those
resources
in
a
production
cloud
because
download
getting
down
one
one
node
will
down
completely
down
all
the
services
and
nodes,
so
we,
for
example,
some
vms
vms,
are
running
on
compute
node.
So
if
controller
goes
down,
vm
might
have
a
possibility
to
get
till
accessible
in
case
of
controller
is
down,
but
if
all
are
on
one
all
the
vms
and
everything
will
go
down,
we
have
to
segregate
all
the
services.
A
D
D
D
D
D
Flavor
also
one
more
thing
that
there
are
many
operations
for
each
and
every
command
you
can
just
type
like.
I
will
show
you
like,
for
example,
openstack
server
create
never
help.
It
will
show
all
the
options
that
we
can
provide
to
server
create.
Also,
there
is
too
much
documentation
available
for
all
these
command
line
options.
Operations
that
we
have
just
did
also
one
more
thing.
D
D
List,
okay,
so
good,
our
vm
is
in
active
state.
So
it's
running.
So
if
you
want
to
see
on
what's
running
in
background,
we
can
just
write
debug
options
and
see
what
are
the
apis
being
running
being
hit
in
the
back
end.
D
D
D
C
D
C
C
C
A
So
thank
you,
everyone
and
we'll
be
doing
some
more
openstack
in
the
future.
We're
also
planning
a
workshop
that
will
probably
be
half
a
day
for
manually
installing
openstack
and
then
another
one,
which
will
probably
be
a
one
day
event
for
installing
triple
and
also
doing
something
shorter
like
how
to
join
the
community
and
get
yourself
set
up
so
keep
an
eye
on
rdo
project.org
for
events
and
more
information.