►
From YouTube: 2017-12-18 Rook Community Meeting
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
B
C
C
Well,
let's
go
through
the
zero-point-seven,
let's
see
where
we
are
with
that
so
I
added
three
open
ships,
because
the
folks
over
at
the
stuff
team
are
trying
to
get
Rick
working
on
open
shifts,
looks
like
mostly
permissions
issue.
Not
you
know.
Any
policy
issues
runs
running
on
openshift,
although
I
think
to
be
friendlier
to
open
ships,
we
have
to
make
some
image
changes
in
our
containers.
E
C
B
C
It's
in
the
route
group,
but
not
UID
0.
So
there
is
there
some
guidelines
for
how
to
fix
it.
We
can
probably
fix
it.
In
general,
there
are
still
some
permission,
issues
that
were
filled
or
security
issues
that
need
to
be
resolved.
For
example,
open
ships.
Has
this
you'd
have
to
like
explicitly
a
lot
running
privileged
and
for
the
agent.
E
D
E
C
B
Quick
thing
about
the
1313
OpenShift
ever
getting
claim
I.
Think
that's
more
of
related
to
from
what
I
read
it.
There
is
no
storage
class
specified
for
this
claim
and
yeah
error
getting
claim
because
and
then,
when
you
scroll
in
a
lock,
something
like
storage
class
is
not
set
or
something
and
root
will
pick
it
up
and
try
to
do
something
about
it.
Yeah
fails
because
yeah
yeah
start
off
empty,
not
found
it.
E
B
A
B
C
D
A
C
B
C
B
B
C
B
It's
really
a
question
of
I,
think
really
kind
of
if
the
OpenShift
guys
are
so
kind
and
give
us
give
us
access
to
each
product
to
test
out
what
can
be
done
because
always
having
a
user
report.
Oh
I'm,
on
auto
ship
product
whatever,
here
it's
not
working,
every
update
a
list
isn't
that
good!
So
maybe
you
can
get
something
from
them.
A
D
A
C
A
D
B
B
C
No
I
I
understand
the
problem.
I
think
the
the
default
is
either
make
it
empty
derp,
which
means
that
by
default
were
not
persistent
or
guests
at
a
directory,
which
will
likely
be
wrong
in
some
environments
on
the
default.
I
honestly
think
this
is
another
one
of
those
where
local
storage
is
the
solution,
but
I
we're.
You
know
we
should
talk
about
that.
Ten
point.
D
Settings
in
the
mo
that
we
have
a
default
where
it
works,
a
mini
cube
like
we
don't
use
a
racial
coding,
Lee
replication,
one
in
this
ya
Milles,
but
then
in
the
documentation.
We
talked
about
well
three
node
cluster
and
user
researcher
coding.
So
this
feels
like
maybe
a
place
where
you
know
and
it's
an
empty
Durand,
the
checked
enamels
but
the
documentation.
All
talks
about
you
know
setting
that
path.
I.
E
Definitely
don't
want
to
go
back
regress
back
to
having
a
default
of
you
know
empty
der,
because
then
we
lose
persistence
right
and
that
just
causes
all
sorts
of
other
problems.
I
wonder
if
this
could
just
be
solved
with
a
really.
You
know
naive,
dumb
way
of
like
having
a
rook
cluster
mini
cube
cluster
openshift
diameter
ones
that
you
know
need
special
settings
or
settings
that
will
work
by
default
in
those
environments
and
just
have
a
couple
of
different.
You
know
clustered
IMO
files
for
the
different
environments,
so
it's.
A
So
I'm
seeing
a
three
solution
so
far
and
let's
see
which
one
works
best
I'm
seeing
have
a
dedication
file
just
for
the
mini
cube
environment,
where
we
said
the
right
conflict
and
right
configurations,
I
work
with
it.
The
other
solution
is
through
use
empty
dirt
just
like
before,
and
that
will
work
anywhere
with
that.
Is
that
the
monitor
the
data
would
no
we
persist.
So
if
I
possibly
start,
you
will
lose
all
day
yeah.
So
that's
or
third,
we
documented
on
the
on
the
Korean
JAMA
file,
so
I
create
a
pull
request.
A
B
Your
comment
in
the
about
the
comment
in
the
yellow
I
think
we
should
ask
users:
what's
they
use
in
the
annals,
because
I
think
we
had
a
few
users
that
were
like
taking
the
gamble
and
being
like.
Oh
that's
commented
out,
let's
uncomment
it
and
remove
together
abdicated
part
and
I'm,
not
sure
why
they
choose,
for
example,
a
racial
code.
You
know
last
example,
so
I
think
we
should
ask
users
what
they
choose
when
they
look
at
the
Yama's
overall.
If
you
even
look
at
the
comments.
C
A
A
C
Will
have
I,
I
think
I
think
for
now
what
we
do
is.
We've
is
better
at
the
mo
put
some
more
comments
if
you're
running
on
MIDI
cubes
that
they
to
mount
SDA
one
if
you're
running
on
blah
blah
blah
set
this
to
otherwise,
you
know,
leave
the
default,
the
VAR
liberal,
which
is
centered
a
bear,
a
bare
metal
thing
and
then
the
true
solution
for
all
of
this
is
local
storage.
C
B
A
B
E
That
says
that
sounds
reasonable,
for
we
had
the
you
know
where
we
want
to
get
to
with
de
experience
for
this.
In
addition
to
your
idea
of
a
some
about,
you
know
comments
in
the
animal
of
saying
you
know
if
you're
on
many
cube
use
this
path,
probably
like
a
small
mention
in
the
docs,
and
you
know
the
quick
start
flow
to
you
know.
Special
considerations
would
be
wise
as
well.
I.
A
E
E
C
B
I,
don't
think
that's
about
that's
an
issue
in
what
you
mean
right
now,
I
think
it's
more
of
like
the
issue
that
when
you
tell
it,
when
you
have
I,
don't
know
five
notes,
you
will
get
this
message,
no
matter
what
you
do,
because
you,
for
example,
have
fun
cool
with
100
P
cheese
as
default
right
now,
and
then
you
yeah,
you
split
it
by
five
and
you
are
below
the
30.
So
right.
C
E
C
Yeah
and
there's
supposedly
a
whole
slew
of
work
that
slated
for
mimic
that
could
help
with
this.
So
I
mean
it's.
It's
a
it's
one
of
those
things
where
I
think
we
have
to
hold
for
Brooke
scenarios
may
I
ready.
This
is
one
of
the
one
of
the
issues
that
everybody
dislikes
about
session.
There's
managing
placement
groups.
E
E
I
didn't
I
was
gonna,
be
my
comments
as
well
as
that
near
they
had.
They
have
expressed
intensive.
You
know
kind
of
abstracting
away
placement
groups
that
users
won't
even
have
to
worry
about
them
and
that's
kind
of
you.
It
sounds
like
our
goal
as
well.
Right,
though,
making
sure
that
the
work
that
is
going
into
mimic
is,
you
know
something
that
is
going
in
the
right
path
and
it's
something
that
will
help
us
is
gonna,
be
smart.
B
B
B
I'm,
for
example,
I've
only
one
pool
in
my
cluster
I
think
free
but
well
I
have
one
pool,
it
doesn't
change,
so
I
would
go
for
many
peaches.
Yes,
I
have
I,
don't
know
it's
nodes,
so
I
use
the
PG
calculator
and
I
get
us
the
value.
I,
don't
know
one
Falls
know
something,
but
for
people
that
are
just
yeah
I
want
one
pool
for
storage.
They
just
hit
this
message
with
two
kind
of
too
many
notes
with.
B
E
They
have,
obviously
that
would
be
just
kind
of
more
of
an
experience
thing
and
a
very
short-term
thing
that
it
doesn't
address
the
long-term.
You
know
placement
group
management's
desire
rate,
but
if
you
wanted
to
just
address
that,
that
would
be
very
simple
to
just
kind
of
in
the
Brooks
detail:
status
output.
So
could
you
know
if
either
filter
that
not
just
you
know
choose
not
to
display
that
message?
Essentially
yeah.
B
A
E
B
C
B
Somewhere,
oh
I,
think
I
posted
this
earlier.
I
posted
it
in
the
pull
request.
Well,
the
thing
is
that
route
API
set
up
with
running
a
lot
of
self
comments,
is
kind
of
yeah
to
make
it
sound
bad,
it's
kind
of
broken
as
I
added
resource
limits
to
my
cluster,
especially
to
the
route
API,
and
if
you
set
them
too
low
the
comments
kind
of
need
to
peak.
B
So
if
you
set
the
limits
too
low
for
a
CPU,
especially
thick,
you
can't
get
any
metrics
anymore,
because
the
time
so
at
least
when
you're
on
a
trip
to
five
seconds,
scrape
interval
so
to
come
back
to
this
issue.
The
problem
right
now
is
that
there
is
no
the
standby
manager
doesn't
take
any
connections
so
to
look
at
it's
like
from
a
Prometheus
but
perspective.
You
always
have
one
house
with
that.
That
is
down
the
step
by
manager.
B
B
B
C
B
Well
done,
there
was
a
block.
There
was
a
dashboard
that
you
could
just
import
and
you're
ready
to
go.
That's
the
thing
I'm
saying
and
with
the
new
metrics
from
a
manager,
they
are
completely
different
to
what
we
expose
and
I
think.
Also
the
digitalocean
suffix
pata
expose
us,
and
that
would
require
new
dashboards
from
the
users.
So
the
users
can
just
simply
import
dashboard
and
are
good
to
go
and
see
their
metrics
right.
E
Basically,
just
you
know
it:
the
community
there's
one.
Basically
one
graph,
you
know
SEF
graph
graph
on
a
community
dashboard
that
everyone
uses
and
in
with
their
that
needs
to
format
from
you
know
the
way
that
we
were
collecting
metrics
before
the
using
this
man
module
so
they're,
just
in
the
community
that
exists,
and
apparently
there's
not
one
that
works
yet
for
that
at.
C
B
Not
a
problem
of
the
exponent
because
x1
is
running
in
the
manager.
The
problem
is
that
most
users
have
to
rule
something
like
up.
If
up
is
zero,
when
up
is
zero,
then
every
service
is
down,
and
the
problem
with
that
is
that
by
default
uses,
your
Ward's
gets
alerts
from
the
Prometheus,
because
a
whole
system.
B
A
B
D
B
Not
one
solution,
I
could
think
of
is
spawning
and
nginx
as
a
proxy
in
front
of
the
managers
or
overall
having
a
changing
the
changing
the
manager
part
to
start
to
containers
and
a
second
one
is
and
nginx
the
proxies
to
the
first
and
to
the
second
one,
but
I
think
for
that.
We
would
need
to
make
more
modifications.
B
D
D
A
C
C
B
C
B
E
Could
be
okay?
Is
you
I
just
want
to
end
up
in
an
experience
where
people
you
know
set
up
their
metrics
according
to
the
it's
simple
instructions
and
our
Docs,
and
they
end
up
with
something
to
look
at
I.
Don't
care
if
it's
Griffin
or
if
it's
you
know,
yeah
ash
board,
I,
don't
care
I
just
want
a
good
experience
where
people
can
see.
You
know
visualization
of
the
health
of
their
cluster.
C
We
upstream
is
going
to
have
support
for
Prometheus
through
a
manager
module.
We
want
to
use
that
manager
module
for
Prometheus
and
we
want
it
so
that
when
changes
are
made
in
upstream
that
we
just
they
work
with,
you
know
Brooke
and
others
and
Agnes.
So
it's
that
extent
I'd
say
we
should
figure
out
how
to
make
the
manage
module
upstream
work.
Well,.
D
B
Wondering
about
the
change
is
that
it's
kind
of
breaking
existing
clusters
if
the
months
aren't
updated.
First,
is
the
problem.
I
noticed
it
with
a
PDS
dad's
yeah,
the
closer
break,
the
moms
break.
If
you
update
operator
first
and
but
you
have
to
update
in
months
and
then
the
months
and
then
a
man,
interest
and
end
operator,
because
if
before,
if
it
enables
the
module,
everything
is
kind
of
coming
crashing
down.
B
B
D
B
B
They
don't
have
to
module,
they
fail,
at
least
at
what's
happened
to
my
trust
amongst
them.
I
had
the
cluster
running
was
like
yeah,
I,
just
updated
and
I
should
restart
operate,
they
operate
and
it's
enabled
to
module
and
a
month
began
crashing
because
of
the
module
not
being
in
it
or
more
of
the
commands,
not
in
a
manager's
being
available.
They.
E
B
D
A
C
E
D
D
B
D
D
D
D
One
for
him
yeah,
we
need
three,
but
now
we
have
five
in
there.
So
there's
room
for
consolidation.
It's
not
gonna
cut
a
lot
off
just
right
and
for
the
performance
test,
I
would
looked
a
little
bit.
I
was
gonna
use
Prometheus
as
well
and
I
have
didn't
know.
There
were
issues
with
Prometheus,
but
the
idea.
B
D
C
D
E
But
some,
let
me
let
me
give
you
an
update
on
this
one.
So
you
know
the
there's
multiple
ways
to
attack
this.
The
one
that
we
first
started
with
was
a
trying
to
do
a
fix
and
upstream
kubernetes,
implement
the
feature
ourselves
and
drive
that
with
this,
through
the
storage
sig
for
having
multiple
default
locations
and
then
did
not
go
you're
not
gonna,
prove
to
not
get
much
traction
so
that
did
not
make
it
into
the
1.9
time
frame.
E
So
we
you
know,
may
want
to
continue
doing
something
that
Steve
had
an
idea
that
we
may
pursue
in
the
1.10
time
frame
and
then
another
place
that
we
want
to.
You
know
possibly
make
some
changes
in
is
in
the
various
upstream
environments,
like
you
know,
do
something
in
chorus
or
cube
a
diem
or
in
range
something
in
upstream
environments
to
make
the
experience
such
that
the
user
doesn't
have
to
do.
You
know
anything
themselves
manually
and
you
know,
could
remove
all
that
documentation
around
it.
E
D
E
C
A
Thing
that
wanna
add
is
that
on
the
last
meeting
last
week,
six
storage
meeting,
they
said
that
they
were
not
gonna,
do
anything
to
change
the
flex
volume
API.
So
we
not
able
to
add
anything
on
the
Jama
file
or
anything
like
that.
They
said
that
all
the
effort
should
go
to
the
PSI
and
CS,
either
pushing
it
to
be
beta
for
one
that
ten.
A
So
so
sad
said
that
if
one
that
ten
is
not
still
in
beta
and
we
still
need
some
changes
on
flex
volume
kind
of
like
robot
I
was
actually
actually
for
rob
block
devices.
They
said
that
we
can.
You
know
we
can
is
fair
for
us
to
to
change,
maybe
change
in
a
flex
money,
but
until
then
they're
they
wouldn't
have
any
changes
from
that
changes.
Api
on
the
flex
volume,
okay,.
A
So
one
this
I
wanted
to
bring
them
probably
later
in
the
meeting
is
that
at
si
si
si
si
si
right
now
is
Robin?
Is
it's
released
its
alpha?
So
you
know
we
as
a
team.
We
should
start
engaging
it
right
away
and
create
a
plug-in
and
all
that
to
ensure
that
si
si
has
everything
that
we
need
to
get
it
to
work.
So
that's.
C
What
we're
just
trying
to
like
I
think
I
think
it'll
be
a
common
CSI
implementation
between
us
and
upstream
stuff.
Okay,
so
that's
that's
the
that's.
What
I
think
that's
the
approach,
we're
advocating
and
and
I
think
we
should
try
to
figure
out
how
to
get
I
mean,
saw
the
features
that
we
wanted
to
get
in,
like
user
mode
set
of,
has
sex
views
and
all
that
that
could
go
in
this
in
this
new
CSI
plugin.
C
C
C
D
A
Slap
around
like
a
non-safe
storage
because
I'm
thinking
of
a
CSI
plugin
for
room
to
be
sort
of
non
safe
and
specific
or
not
suppose
a
lot
of
the
stuff,
specific
information
it
make
it
more
like
a
type
of
a
generic
and
in
and
in
high
dose,
implemented
safe
stuff,
safe
configuration,
I
wrapped
them.
That's
what
I
was
thinking
for
work,
yeah.
C
A
A
Yes,
so
right
now,
the
way
we
we
consume,
the
Flex
volume
or
week
is
through
the
sorry.
This
mm
is
through
a
flexible
you,
so
what
we
want
to
do
to
this
is,
to
kind
of
you
know,
used
dynamic
provision
file
system.
So
there
is
a
decimal
nation
done
using
Python
on
the
external
storage.
So
we
can
we
can.
We
can
leverage
that
so
what
it
does
is
that
it
basically
it
creates
one
file
system.
A
Once
the
house
is
a
server
and
then
from
that
server
it
it
creates
directory
for
every
file
system
the
user
needs.
So
that's
not.
Lastly,
this
dynamically
handle
of
provisioning
from
that
you
create
I,
sub
path
that
you
are
located.
Each
user,
there's
already
limitation
for
that,
like
I
said,
is-
is
to
a
Python
script.
That's
interpreter!
A
A
A
C
A
D
C
B
B
C
B
B
E
B
E
Exactly
are
you
talking
about,
then
I
know
we'll
look
at
this
pull
request
and
and
follow
up
on
it
today
already
looks
like
you
already
approved
that
okay
I
think
we
had
had
some
more
back
and
forth
on
it
or
Alex
I
made
another
commit,
but
I
will
follow
up
on
the
hub
and
I've
been
offline
a
lot
recently,
but
I'm
going
back
on.
So
sorry.
C
The
next
one
is
gonna
be
in
the
new
year
right.
This
is
the
first
week
I
think
it'll
be
the
first
week
yeah.
E
I
think
I
think
that.