►
From YouTube: OpenCrowbar Planning 2014 06 18
Description
Minutes: http://pad.opencrowbar.org/p/community-planning-meeting-2014-06-18
OpenCrowbar community website, will take pulls since this is github!
List Documentation Changes Needed
OpenStack cookbooks
IPMI integration
100 Node Challenge
Squid Proxy Feature (latest pull)
Process & Backlog Review
A
Good
morning,
everybody
welcome
to
the
june
18th
planning
call
for
the
for
the
open,
crowbar
community.
A
Sorry,
the
thing
we
should
do
is
actually
start
with
things
that
ads
that
were
things
that
were
accomplished
in
the
last
sprint,
so
we
should
start
with
review
victor.
Do
you
want
to
talk
about
the
the
pull
changes
that
went
through
and
some
of
the
capabilities
that
were
brought
back
or
added
from
that.
B
Yeah,
so
the
main
thing
that
I
did
over
the
last
over
the
last
week
was
to
pull
back.
It
was
to
fix
up
the
the
support
that
we
had
for
deploying
and
using
a
chef
server
that
it
fit
rotted
a
little
bit
and
got
that
back
up
in
shape
and
arranged
for
everything
to
use
it,
so
that
my
life
would
be
much
easier
trying
to
make
berkshire
from
multiple
bar
plants.
For.
B
Because
previously,
the
first,
the
we
kind
of
on
using
burp
shelf
to
help
us
manage
and
pull
in
upstream
cookbooks
that
we
don't.
We
didn't,
have
to
do
things
like
making
our
own
fork
of
the
apache
cookbook,
because
life
is
too
short.
However,
it
turned
out
that
using
that
in
combination
with
chef
solo
everywhere,
burke's
files
aren't
composable,
so
I
could
just
take
about
all
the
work
files
that
we
needed
and
glom
them
together
and
have
a
reasonable
expectation
of
it
working.
C
A
D
B
You
know
mess
around
with
run
lists
or
custom
attributes
we're
going
to
blow
them
away
the
next
time.
We
do
something
with
this.
B
A
B
There
functions
well,
there
are
functions
that
we'll
have
to
implement
in
crowbar
a
little
bit
just
for
use
by
the
roles.
For
instance,
we
don't
have
a
provision
to
create
encrypted
data
bags
and
I'm
thinking
that
the
easiest
way
to
do
that
for
things
that
will
require
them
is
to
let
the
role
take
the
attributes
that
it
wants
to
put
in
an
encrypted
data
bag
and
put
them
in
there.
B
So
we'll
need
to
expose
some
of
that
functionality,
but
the
chef
gym
provides
apis
to
do
that
and
we
can
just
package
them
up
in
ones
that
are
slightly
simpler
to
use
or
let
people
grapple
through
the
chef
gym
on
their
own
and
figure
it
out.
A
It
just
depends
on
what
people
see
is
invaluable,
yes,
cool.
You
want
to
talk
about
the
squid
proxy
stuff
because
to
everything
I've
been
hearing
is
that's
it's
just
it's
a
really
important
function
and
that
we
sort
of
put
it
in,
and
I
need
to
do
some
better
work
promoting
why
it's
important,
but
yeah.
B
This
is
what
isn't
quite
merged,
yet
it's
in
my
latest
pull
request,
but
one
of
the
things
that
I
did
there
was
to
lift
the
we've
in
open
chrome,
where
we've
always
set
up
a
good
proxy
as
part
of
deploying
the
provisioner
server,
which
is
you
know,
the
apache
server
that
we
use
to
do
all
of
our
internal
cluster
related
hdb
file
serving
stuff
and
we've.
C
B
Had
a
squid
proxy
in
place
so
that
you
could
deploy
a
working
cluster
without
needing
to
let
the
compute
nodes
have
an
ip
address
that
was
routable
to
the
public
internet,
so
the
mode
of
operation
that
I've
always
done.
My
development
on
was
the
compute
nodes
that
I
test
on
when
I'm
deploying
them
as
kvm
knows,
they
don't
get
bound
to
a
bridge
that
has
access
to
the
outside
world.
B
It's
a
completely
isolated
thing
and
whenever
they
need
to
pull
packages
from
the
internet,
they
always
do
through
the
proxy
that
lives
on
the
admin
that
gives
us
a
kind
of
a
layer
of
installation
from
the
internet.
It
also
gives
us
a
nice
point
for
auditing
and
caching,
because
if
you're
deploying
a
thousand
notes
question
you
don't
want
to
pull
down
a
thousand
copies
of
you
know
all
the
latest
updates
to
centos
every
time
you
want
to
let
squid
pull
down
one
and
cache
it
for
everyone
else
to
use.
B
So
the
latest
changes
that
I've
made
in
this
proxy
support
was
to
actually
list
that
out
from
being
deployed
by
the
provisioner
and
deployed
as
part
of
the
initial
bootstrap
process.
When
we're
setting
up
the
other
servers,
the
other
services,
that
crowbar
relies,
so
it
goes
that
we
download
and
install
the
packages
that
we
need
as
part
of
bootstrap.sh,
then
when
we
get
to
actually
standing
up
a
production
or
a
development
server.
C
A
B
Finding
that
every
crowbar
does
winds
up
all
the
js,
all
of
its
hd
http
traffic
to
the
outside
world
all
goes
through
squid,
and
so
it
all
gets
cached
properly.
A
Right
and
then
we're
also
saying
that
the
admin
node
can
be
the
out
the
gateway
for
the
host
operations
right.
B
E
What
so
the
when
you
operate
when
you
pull
the
new
bits?
Does
it
clean
removes
all
on
all
bits,
old
versions,.
E
B
We
tell
squid,
hey,
you
can
use
extra,
we
can
use,
you
can
use
x,
amount
of
disk
space
for
your
caching
purposes
and
once
you
start
filling
it
up
start
throwing
away
things
that
haven't
been
used
in
a
while.
A
All
right,
I
won't
one
one
thing:
one
thing
to
note
is:
if
you're
using
a
docker
admin,
then
your
squid
and
the
squid
cache
that's
getting
set
up
on
that
docker
admin,
I
believe,
is
getting
blown
away
every
time,
but
so
you
want
an
external
squid
cache
on
your
in
your
in
your
network.
In
addition
to
the
one
we're
going
to
set
up
on
the
admin
node.
B
Yeah
it'd
be
easy
enough
to
arrange
that
not
to
happen
and
if
we
wind
up
wanting
to
support
docker
a
docker
admin
as
a
way
to
actually
run
in
production.
We
need
to
do
that
anyways,
but
yeah
for.
C
A
I
I
suspect,
a
true
dockerite
duckerson.
I
don't
know
what
to
call.
It
would
tell
would
tell
you
that
we
should
really
deploy
crowbar
with
like
six
different
containers
and
then
just
swap
the
crowbar.
A
They're
welcome
to
contribute
that
work.
Oh
tmux,
that
that's
I
like
that,
I'm
still
getting
used
to
it.
B
Yes,
in
the
docker
admin
node,
everything
runs
under
everything
runs
in
a
tmux
session,
so
you
can
detach
from
it
reattach
to
it.
Spin
up
new
shells
running
inside
the.
A
A
The
other
thing
is:
there's
a
pull
in
there
with
docks
working
again
that
was
held
off
until
I
got
bdd
passing
to
100,
okay
again,
so
that
that's
in
the
other
thing
that
we
accomplished
that
I'm
really
excited
about
is
we've
been
using
the
issues
tracking,
so
we
put
all
the
stories
in
and
then
filing
filing
and
fixing
bugs
in
here,
which
is
pretty
cool.
A
Yes,
and
I
I
have
to
say,
it's
doesn't
have
all
the
promise
that
a
lovely
jira
atlassian
integrated
system
would
have,
but
it
really
is
nice
if
you
put
pound
and
then
the
id
numbers
it
cross,
references
for
you,
which
is
pretty
darn
handy.
A
And,
let's
see
oh
and
then
we
started
down
the
process
of
the
travis
ci
that
scott
had
suggested
somewhere
in
here
hold
on.
Oh
I'm,
in
the
dock
section.
A
So
I
actually
have
the
open
crowbar
and
I
flipped
the
on
switch
and
added
a
travis
ci
file,
hey
and
it
passed
hot
dog.
Look
at
that
just
echo,
but
that
was
the
goal
for
the
sprint.
So
richard
accomplished,
I
actually
did.
I
don't
think
it's
voting
yet
into
github,
but
that
was
this
is
the
start
I
like
that
it
actually
tested.
The
first
thing
it
tested
was
adding
itself
so
yay
process.
B
So,
what's
the
ultimate
goal
for
testing
with
travis
ci?
Is
it
going
to
be
capable
of
doing
our
whole
docker
admin
thing,
or
is
it
just
going
to
be
doing
the
bdd
stuff.
A
A
D
You
might
want
to
keep
it
external
to
the
to
the
to
the
installation.
Scripts
travis
can
run
those
things
in
sequence
right,
so
you
know
whatever
needs
to
be
done
for
the
initial
setup
of
the
system,
and
then
you
know,
let's
just
start,
adding
a
library
of
test
sequences
that
we
want
to
have
done.
Bdd
being
the
first.
A
D
It'll
it'll
kill
it
when
it's
done
right,
so
I
mean
it'll
it'll
detract
the
virtual
machine,
but
in
any
case
we
can
talk
about
it.
A
And
not
hard
to
play
with,
so
this
is
something
for
next
sprint.
A
So
we
got
that
talked
about
that.
I
did
go
through
and
do
some
of
the
docs
listing
and
try
and
list
out
missing
documentation
areas.
A
One
of
the
things
that
I
would
love
to
have
somebody
do
is
hey.
I
did.
This
too,
is
to
actually
go
through
the
deploy
guide
and
then,
while
you're
going
through
it,
put
in
some
screen
pictures.
A
I
I've
been
playing
through
some
of
the
stuff
and
building
I've
been
trying
to
work
in
the
deployment
guide,
and
so
the
deployment
guide
actually
walks
through
the
whole
create
process,
and
but
it
would
it
could.
It
would
really
be
helpful
to
have
screenshots
in
this
to
help
people
and
potentially
break
this
into
sub
documents,
and
then
I've
been
adding
things
as
I
find
them
like
troubleshooting
tips
for
deployment
and
then
victor,
I
could
use
some
help
just
building
out
the
the
port
mapping
list
of
all
the
services
that
we've
exposed.
A
A
Oh
and
for
the
mid
for
the
design
meeting
I
want
to
talk
about,
I
actually
had
a
screen
a
concept
to
simplify
the
first
step.
I
think
I
finally
figured
it
out.
D
So,
to
follow
up
on
a
couple
of
those
items
there
right
work
on
the
deployment
guide
consists
of
run
through
and
and
screenshots
travis
ci.
I
think
there's
there's
effectively
two
pieces
to
that.
The
first
one
is
is
a
full,
a
full
build
run
through
of
bdd
tests
and
then
a
design
for
the
further
voting
infrastructure
and
how
we're
going
to
how
the
how
the
process
flows
are
going
to
work.
A
D
A
B
B
B
Home
I'll
be
heading
down
to
the
fedex
facility
in
north
austin.
To
pick
it
up.
C
Hey
rob:
where
was
that
documentation
that
had
the
port,
ids
and
so
forth,.
A
All
right
other
other
deliverables.
I
know
we
made
some
progress
on
the
100
node
100
word.
I
keep
calling
100
words.
I
don't
know
how
many
words
are
in.
It
definition
see
if
I
can
find
it
and
I
think
we've
got.
We've
got
a
lot
of
text
for
the
100
node
challenge,
so
I'm
going
to
formulate
that
into
a
broader
community
thing.
So
actually
this
next
sprint.
A
B
So
we'd
be
looking
for
someone
who
just
needs
to
get
the
os
on
a
bunch
of
servers.
A
Yeah,
I
you
know,
and
I
think
ideally
we
should
talk
about
ipmi
and
it
depends
on
when,
when
the
timing
is
and
what
what
they
want
to
accomplish.
E
A
A
E
B
A
But
the
the
point
is
the
first
part
of
the
servers
arriving
was
that
we're
starting
we're
bringing
back
the
ipmi
workloads?
So
we
had
a
really
good
design
session
last
week
about
this.
A
Here
it
is,
this:
is
the
list
it's
over
here
on
this
pad.
We
talked
about
some
role,
naming
and
bringing
back
the
conventions,
and
then
I'm
actually
trying
to
use
this
to
help
document,
creating
a
new
workload.
A
A
Because
there
are
some
potential
changes
to
make
where
I
was,
I
was
starting
to
see
where
we've
injected
the
the
term
bar
clamp
into
paths
and
stuff
that,
if
we
can
change
without
much
lift,
I
would
change,
but
I'm
tempted
to
just
leave
it
in
place.
C
A
I
I
think
that
what
we
could
do
is
we
could
leave
the
old
ones
in
place
and
like
let
it
be
dual
path,
but
I'm
I'm
going
to
look
into
that.
That's
that's,
I
think
a
good
topic
for
the
design
meeting,
but
the
goal
so,
but
the
goal
for
the
ipmi
workload
is
to
basically
bring
in
the
old,
bring
bring
over
cb1
code.
B
A
That's
what
I
thought
too,
so,
yes,
you
did.
A
I
didn't
did
he
do
it
in
open
crowbar
or
did
he
do
it
in
the
v2
repos.
A
Easy
enough
easy
enough
to
find
out.
I
think
it
was
mostly
working.
It
was
the
proxy
bridge
that
was
was
causing
problems.
I
don't
think
we
need
that
anymore,
so,
okay,
so
that
that's!
This
is
something
we're
working
on.
If
we
actually
have
some
hardware
to
test
it,
then
we
should
be
able
to
work
against
physical
hardware.
E
Victor
you're
old,
you
know,
nick's
stamp
is
still
available
and
it
had
the
you
know
all
the
different
servers
in
it.
E
A
Yeah
and
that
so
there's
other
there's
topics
that
I'm
postponing
talking
about
the
idrac
and
the
idrac
proxy,
which
I
have
now
that
I
have
a
server
I
can
play
with
a
little
bit
more
easily.
A
E
E
A
Because
I
like
to
do
that
challenge
so
there's
there's
actually
a
fair
bit
of
text
in
this
dot
in
this
document.
Let
me
pull
it
up.
We.
B
A
A
B
D
C
Up
that
merge.
A
I
did
make
those
changes
that
you
asked
for
so
these
things
should
be
good,
so
this
these
are
actually
both
going
to
be
fixed
together,
and
I
think
this
is
right,
but
I
was
going
to
add.
I
was
going
to
add
a
test
for
it
before
we
close
it,
and
so
that
gets
us
back.
This
is
actually
a
design
discussion.
A
We
need
to
review
that
we'll
review
that
on
wednesday
hadoop
ready
day
prep
mike.
I
think
that
there
was.
D
A
A
All
right,
so
that's
different,
so
I
think
we're
going
to
need
a
story
for
the
this
is
fixed
I'll
close
that
afterwards
I
added
a
new.
A
small
judd
found
a.
A
A
A
Community
website
did
we
talk.
I
think
we
did
this
in.
A
So
one
of
the
so
this
this
is
all
this
wonderful
style
is
default,
github
stuff,
so
I
I
pulled.
We
cleaned
up
this
trying
to
remember
if
we
covered
this
two
weeks
is
a
long
time
for
me
to
remember
what
happened.
This
was
a
result
of
a
design
meeting
a
little
while
ago,
where
we
split
these
two
into
into
different
pieces.
A
A
A
A
You
know
one
one
of
the
things
I
like
about
github.
Is
it
kind
of
renders
the
markdowns
as
if
they
were
readmes,
but
the
thing
I
don't
like
is
it
doesn't
give
us
any
navigation,
any
reasonable
navigation,
so
I
I
did
some
work
about
six
months
ago
to
take
some
of
that
markdown
stuff
and
actually
build
it
into
a
site.
You
know
single
source
for
the
docs
a
little
bit
of
a
wrapper
around
it.
A
I
could
pick
that
up
and
take
a
look
at
it
right
and
one
one
thing
that
I've
been
doing
is
I
fixed.
I
fixed
the
the
documentation
full
index
and
I've
been
tweaking
it
so
mike.
I
took
what
you
had
been
doing
and
I've
been
expanding
it
a
little
bit
so
there's
a
table
of
contents
and
then
a
full
index.
A
A
Issue
is
then
you're
maintaining
the
links
in
the
markdown
and
there's
there's
kind
of
a
meta
level
where
we
can
generate
a
lot
of
those
links
off
the
same
source
code.
I
had
built
a
prototype
with
sphinx
that
mostly
mostly
dealt
with
that
problem
that
you're
you're
you're
right.
If
you
were
to
take
these
docks
and
dump
them
into
a
site,
you
could
add
a
header
for
each
one
that
basically
had
siblings
up
siblings
children
up
down
navigation,
which
is
exactly
what
the
ui
docs
do
right.
So
this
same
concept
same
concept.
A
E
A
That
would
be
pretty
exciting,
because
if,
if
we
have
the
same
level
of
navigation,
it
makes
it
so
much
easier
and
one
of
the
things
I'd
love
to
be
able
to
see
is
stuff
like
the
principles
documentation
which
I
don't
think
is
in
here,
which
I'm
not
sure
where
it
is
here.
Would
it
you
know,
be
nice
to
see
that
just
made
easier
for
the
community
to
been
clicking
around
too
much.
I
lost
track
of
my
tabs.
A
A
Be
interested
in
us
taking
the
principles
guides
and
creating
creating
a
like
a
white
paper,
so.
A
Yeah,
so,
while
we're
on
this
topic
on
the
weekly
end
of
things,
I
just
I
I
did
a
minor
wiki
update
to
kind
of
point
people
towards
the
master
docs
and
the
in
the
source
tree,
because
I
don't
think
we
want
to
start
replicating,
have
duplicate
content
in
many
places.
A
So,
let's
see,
I
think,
wiki
page,
it's
just
a
shallow
introductory
page.
I
don't
know
if
we
want
to
link
here
to
the
community
site,
but
I
just
put
in
the
deep
link
into
the
actual
documentation.
Excellent.
Thank
you
yeah.
We
don't.
We
really
don't
want
people
editing
this.
Oh,
we
can
add
custom
sidebars
that
might
this
might
be
the
way
to
handle
some
of
that.
A
A
So
that
leaves
open
stack
community
cookbooks.
B
And
there
are
some
things
in
the
core
that
need
to
be
added
to
make
the
openstack
community
work
properly.
B
Okay,
at
the
very
least,
we
need
to
have
some
level
of
support
for
telling
crowbar
that
some
attributes
should
go
into
data
bags
and
encrypted
data
bags,
because
they
expect
that
all
password
type
information
should
be
stuck
into
encrypted
data
bags,
instead
of
being
locked
up
to
plot
them
through
roles.
C
B
That
we
that's
the
main
thing
that
jumps
out
at
me.
There's
also
some
things
where
I
managed
to
inadvertently
break
how
cluster
roles
work,
so
I
don't
need
to
be
fixed.
I've
been
planning
on
doing
some.
B
Just
to
get
the
the
theft
cookbooks
functioning
again.
A
A
B
You
know
you
only
really
have
mons
osds
and
yeah
mondo's,
these
mds's
and
the
gateway.
E
And
that's
right
so
to
be
guys,
pay
attention
that
the
self
is
going
to
release
g8
by
the
end
of
the
month,
saf
in
in
well.
The
enterprise
f
ice,
one
two
which
will
be
running
on
rail
seven
and
that
is
firefly.
One
two
code
base.
C
B
Probably
just
gonna
stick
with
emperor
until
I
get
everything
working
more
or
less
correctly,
and
then
worry
about
updating
to
updating
to
later
versions.
A
So
victor,
I
I
recognize
there's
places
where
we
we
have
technical
changes
to
making
core.
I
think
that
there
are
people
who
are
interested
in
being
able
to
drive
those
cookbooks,
but
don't
know
where
to
start
right.
We
already
did
some
work
on
some
base.
Cookbooks
do
we,
you
know,
is
what
I
think
we
need
to
think
through.
If
somebody
came
in
and
said
all
right,
I
want
to
try
to
try
to
help
make
this
progress.
A
A
So,
if,
if
we
just
said
look,
this
is
how
you
use
berkshiff,
you
set,
you
know,
take
just
actually
showed
somebody
how
to
use
the
upstreams
that
we
had.
They
could
start
iterating
through
this
process.
The
things
that
you
need
to
do
to
fix
and
core
aren't
necessarily
blockers
for
somebody
just
turning
through
the
turning
starting
to
turn
the
crank
on
integrating
the
upstreams
right.
There's
fixes
to
the
upstream.
Some
patches
they'll
need
to
make
and
they're
not
blocked
on
the
other
stuff.
So.
B
Like
I
said
those
those
the
the
unbreaking
of
cluster
support
and
adding
support
for
being
declared
that
these
attributes
need
to
go
in
data
bags
is
something
that
we
need
to
have
working.
But
we
can
really
start
on
a
lot
of
upstream
stuff.
That
assumes
that
there's
a
lot
of
options
that
they
do
use
encrypted
data
bags.
A
lot
of
parts.
C
A
So
I
we
have,
we
have
a
priority
call
to
make
from
ipmi
versus
the
encrypted
database
and
which
one
which
one
would
have,
which
one
is
more
important
within
the
community.
A
But
I
I
don't,
I
don't
know
which
way
to
guide
you.
I
think
that
if
we
had
people
who
were
interested
in
spinning
up
on
the
openstack
cookbooks
and
we're
clamoring
for
it,
I
would
I
would
say
that
that
would
we
should
look
at
that,
but
for
right
now
I
think
the
ipmi
is
actually
the
bigger
the
hotter
topic.
The
people
on
the
call
feel
that
that's
the
right
call
ipmi
is
first,
yes,.
C
C
B
A
A
That'll
then
drive
whether
or
not
we
fix
the
encrypted
data
bag
thing
as
a
priority
or
not
the
the
thing
to
call
out
the
community
call
is,
we
know,
there's
a
gap
as
soon
as
as
soon
as
it
becomes
a
blocking
gap.
We
can.
We
can
prioritize
it,
but
for
right
now
I
I
want
to
make
sure
everybody's
on
the
same
page.
That
ipmi
is
actually
the
the
first
feature.
B
B
A
A
We're
we're
not
mirroring
other
open
source
communities
and
pushing
that
card
on
pull
requests.
We
do
test
them,
though,.
A
All
right
that
was
actually
the,
I
think
what
what
I
wanted
there.
I
think
this
remains
on
my
plate
to
to
build
the
dock,
build
some
of
the
documentation
or
find
it
find
where
we
had
it,
because
I
believe
that
we
actually
documented
some
of
this
stuff
I'll
produce
video
or
something
something
fun
like
okay,
other
topics,
other
items,
I
think
we've
covered
everything
that
I'd
expected
us
to
cover
a
little
bit
more.
A
B
For
real
seven
changes,
unless
I
can
manage
to
get
some
sort
of
relative
and
iphone
legitimately,
as
long
as
as
well
to
hook
up
with
subscriptions
that
will
probably
get
pushed
off
until
there
is
a
center
of
79.
So
for
me
to
play
around
with.
C
Rob
is
it
worth
putting
something
up
regarding
creation
and
distribution
of
packages
and
deployment
guidelines,
so
that
people
can
just
install
it
on
rail,
seven
and
test
it
or
on
centos,
seven
and
test
it.
A
B
C
B
A
B
That's
only
true
in
the
docker
container
and
I
about
about
a
week
before
they'll
stop
working
on
crowbar.
I
actually
wrote
a.
I
actually
wrote
a
little
kickstart
that
will
bootstrap
the
kvm
admin
mode
at
606.5
and
modifying
that
for
7
would
actually
be
a
good
thing
to
do.
B
C
B
I
thought
they
just
had
essentially
bug
testing
releases
out
right
now.