►
From YouTube: Ceph Testing Weekly 2018-09-05
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).
B
B
D
C
B
A
B
Remember
that
we
discussed
that
I,
don't
remember
URI
URI,
mind
I,
think
it's
just
sitting
on
the
tooth
ology
server
machine
in
the
land
somewhere
before
I
went.
E
B
A
B
A
Example
for
both
downstream
Susy
tautology,
the
way
it's
deployed
right
now,
it's
using
the
tulgey
opens
with
old
OpenStack
who
that
is
deploy
Nate.
Why
shell
script?
Everything
then
is
deploying
on
the
Punto,
which
has
pretty
different
set
of
packages
that
recently
unix's
recent
Linux
is
cares
like
it
doesn't
use
systemd
and
does
it
use?
You
know
it
just
using
some
custom
in
it.
A
A
C
A
C
A
E
A
C
A
But
yeah
they
do
run
but
on
which
machines
they
do
run.
This
is
the
question
and
they're
not
the
another
thing
that
guys
who
said
we
were
saying
that
it's
not
always
passing,
and
even
when
I
was
the
meeting
some
PR,
it
was
failing
first
time
and
the
second
time
it
passed,
so
they
they
have
usually
not
expected
results,
not
always
passed,
not
because
of
the
you
know
the
part
itself.
C
Getting
back
to
adding
some
openSUSE
coverage
do
upstream,
there
are
several
things
that
have
to
happen.
Obviously,
before
before,
openSUSE
could
be
a
fully
supported
distro,
but
we
have
to
I'm
I'm
thinking.
We
have
to
start
somewhere
and
move
intimately
and
the
the
lowest
of
the
low
hanging
fruits
seemed
to
me
to
to
be
you
get
a
an
openSUSE
image
is:
what
is
it
called?
Fod
fog
is
what's
used
to
yeah.
B
C
A
E
D
B
Keep
in
mind,
I
mean
part
of
part
of
the
deal
was
there
was
like
maintenance
help
with
it
too.
So
not
not.
Just
like
adding
a
new
thing
to
the
people
to
the
to
the
constraint
people
we
already
have
so
you'll
probably
need
to
like
probably
able
to
guide
you,
but
you
guys
will
need
to
like
get
in
and
like
figure
out
how
to
build
images
or
at
least
tell
him
how
to
get
it.
Working
like
a
script.
X
you
get
to
get
of
working
one
or
something
yeah.
C
C
D
C
C
Then
then,
once
I
have
that
then
I
can
push
all
of
the
scripting
that
we
use
for
testing
deep
sea
deployments.
I
can
push
that
upstream
and
and
demonstrate
that
it
actually
works
in
the
upstream
environment
on
openSUSE
and
then
and
I'm
I'm
planning
to
work
very
incremental
II.
You
know
and
and
not
you
know,
create
any
huge
waves.
I
don't
want
to.
You
know,
start
by.
You
know
breaking
breaking
tooth
ology
or
something
like
that.
B
C
C
C
These
are
just
deployment
tests
that
deployed
use
deep
sea
to
deploy
so
right,
and
then
they
check
that
that
the
resulting
cluster
is
saying
and
working
so
it
the
tests
are
go
from
the
very
simplest,
the
thing
that
just
creates
a
cluster
with
OS,
DS
and
moms,
and
a
manager
and
checks
that
the
it's
in
health,
okay
and
that's
the
end
of
the
test,
all
the
way
to
more
complicated
things
like
NFS,
Ganesha,
I,
scuzzy
gateway
and
open,
attic
and
and
each
tests
after
deploying
the
cluster
does
some
some
validation.
You
know
to
make.
C
You
know
to
assert
that
the
resulting
cluster
is
is
somehow
saying
or
functional.
That's
as
far
as
I've
gotten,
but
but
I'm
very
keen,
like
I,
said,
do
they
have
you
know
if
if
we
enable
SEP
ansible
for
the
QA,
Suites
and
and
there's
a
some
kind
of
an
architecture
for
that,
then
I
would
like
to
plug
in
deep-sea
as
well
yeah.
B
B
Want
to
go
through
the
whole
thing
every
time
so
yeah,
but
you
know
once
I
have
anything
useful,
I
think
the
first
couple
days
you're
just
gonna,
be
like
there
are
some
ansible
tests
already
like
and
and
the
redhead
QV
has
a
public
rh
branch
into
college
e.
That
does
that,
although
I
haven't
looked
through
it
good
well,
there's
right.
B
E
Greg,
let
me
talk
about
defensible.
What
currently
we
have
in
upstream
and
what
we
have
in
redhead
branch.
The
upstream
one
actually
is
a
full-fledged
sensible
testing,
because
offensive
all
actually
plays
around
with
variables,
though
adding
new
test
is
basically
just
adding
a
ml
fragment
with
various
variables
to
the
sensible.
E
E
So
we
cannot
do
that,
we
can
do
so.
We
did
it
in
downstream,
but
we
at
a
very
high
cost.
The
high
cost
is
with
limited
number
of
nodes.
We
can
just
learn
one
or
two
tests
parallely,
because
you
need
eight
or
ten
nodes
to
do
the
same
amount
of
testing.
What
we
can
do
in
three
node
testing.
So
those
are
the
pieces
and
it's
I
mean
it's
a
trade-off.
So
one
has
to
decide
what
we
want
and
if
limited
coverage
can
do
it.
That's
why
yeah.
B
And
I
mean
I'm,
not
yeah,
and
more
than
that
like
wait,
maybe
you
don't
want
to
enable
that,
because
you
really
don't
want
users
doing
that,
but
you
need
your
little
developer.
Well,
we
have
a
developer.
C
We
have
a
dev
dev
or
a
yeah.
It's
called
dead-end
environment
variable
that
disables
validation,
you
know
so
normally
deep-sea
will,
if
you
specify
moms
and
OS
DS,
you
know,
or
or
or
a
very
small
cluster
under
four
nodes.
Deep
sea
will
say
no
I'm
not
going
to
deploy
that.
But
but
there's
a
way
to
you
know
get
around
that.
A
B
C
E
C
You
know
it
will
make
it
so
much
easier
to
to
run
the
tests
on
San,
Tomas
Ubuntu
right,
since
our
our
all
of
our
downstream
testing
infrastructure
is
geared
towards
us
less,
so
even
even
getting
openSUSE
coverage
in
the
downstream
infrastructure.
You
know
there's
an
uphill
battle
and
so
and
then
adding
sent
to
us-
and
we
wound
to
that.
To
that
is
it's
just
just
not.
We
don't
have
the
manpower
to
maintain
sent
to
us.
We
boom
two
images.
No.
C
C
I
just
wanted
to
mention,
of
course,
the
Ceph
task
deployment
method
also
works
on
openSUSE
yeah.
Oh.
B
C
The
David
should
know
right
and
we
can
yeah
well
I'll,
provide
an
image
and
if
it
doesn't
work
we'll
we'll
provide
a
different
one
and
I
the
the
script
that
I
use
to
create
the
images
I.
Can
you
know
I,
add
different
image
types
or
targets
you
know
and
and
define
special
sets
of
packages,
for
example
or
or
add
remove
packages
not
a
big
deal.
I.
E
E
B
Helen
next
week
there
are
some
thank
you.
Some
people
in
Red
Hat
have
been
running
around
and
using
Blake
and
AI
analysis
on
CI
and
testing
stuff.
They
want
to
come
and
talk
to
us
about
some
things,
they've
learned
when
they
ran
flake
and
there
you
know,
processing
against
the
tooth,
ology
codebase,
so
hopefully
that'll
be
interesting.
Well,
it'll
certainly
be
interesting,
hopefully,
is
useful,
but
we'll
definitely
be
here
next
week.
So
flake.