►
From YouTube: 2019-04-17 :: Ceph Testing 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
A
I
was
hoping
I
could
figure
out
some
more
abstraction
for
the
remote
process
that
we
use
internally
in
the
QA
Sweden
to
ecology,
as
I
think
that's
gonna
play
badly
with
rook,
but
looking
at
it,
some
more
I'm
playing
I'm,
just
gonna
shoot
and
shoot
an
emo
today
talking
about
sort
of
the
interfaces
we
need
and
what
exists
and
I
may
I
may
be
able
to
narrow
it
down
a
little
bit,
but
just
to
see
what
the
options
are
before
I
try
and
make
it
a
smaller
interface
like
too
much
but
I
think
I'll
get
some
of
that
stuff
out
today.
A
B
A
Well,
so
they
we're
just
trying
to
set
up
interfaces
for
all
that
because
the
tactics
are
like
poking
down
and
it's
or
the
tests
are
like
poking
into
some
of
the
stuff
directly
and
I'm,
not
entirely
sure,
and
so
one
of
the
things
a
lot
of
some
of
them
do
is
like
look
at
the
actual
standard
out
from
a
process
on
a
remote
node
and
wait
for
it
to
close
or
something
and
so
I
need
to
add
it
a
little
more
carefully
exactly
like
what
they're
doing
there.
A
A
But,
okay,
we
need
to
know
bit
like
yeah.
We
need
to
know
what
we
can
do
to
replace
that
and
rook
and
I'm,
and
so
that'll
be
a
discussion
and
then
simultaneously
that
I
also
want
to
like
once
we
know
have
a
good
I
like
once.
I
got
an
arrow
interface
I
want
to
just
on
top
of
the
stuff.
We
have
actually
build
those
narrow
interface
narrower
interfaces
and
make
sure
nothing
breaks.
A
B
A
Only
like
a
couple
days
ago
that
several
of
the
smoke
tests
are
just
failing
and
apparently
have
been
for
a
little
while.
That's
he's
asked
just
on
IRC
I,
don't
know
I
think
maybe
he
was
gonna,
try
and
fix
them.
Cuz
he's
working
on
a
big
refactoring
needs
to
know
that
it
worked
but
I'm
not
certain.
He
could
pay
him
and
asked.
A
B
C
A
So
there's
a
couple
of
things
there
so
yeah.
It
would
be
good
to
have
a
smoke
sweeper
just
tooth
ology,
but
I
mean
a
big
part
of
tooth
ology.
Is
that
it
supports
what
the
ACEF
QA
suite
is
doing
and
there's
not
and
and
since
our
interfaces
are
so
loose
and
even
if
they
weren't,
like
writing
tests
to
asked
all
those
interfaces
that
didn't
actually
involve
stuff
would
be
a
big,
a
big
thing.
A
But
yeah
we'd,
like
it
it'd,
be
nice.
If
we
had
some
that
would
like
smoke
tests
that
were
that
covered
most
of
the
stuff,
or
at
least
you
know
a
few
critical
bits
that
we
could
just
run
separate
from
set
itself.
Looking
the
smokes
weed,
there's
the
there's
like
a
couple
directories
in
it:
there's
the
OneNote
directory
and
that
uses
set,
deploy,
IC
or
basically
just
the
one
better
spatially.
Just
one
test
get
the
basic
folder.
That's
got
a
whole
bunch
of
stuff
that
deployed
the
normal
tooth
ology
way.
C
D
B
C
C
C
C
So
it's
not
interfere
because
the
totality
of
the
development
of
the
I'm
you
know
they
did
ask
actually
will
have
own
instances
of
pedals,
and
only
if
we
run
this
in
parallel,
then
we
need
to
somehow
select
this
a
dream
machines
in
order
to
test
all
these.
They
PMI
and
different
fog
stuff,
but
yeah.
B
B
B
A
B
B
A
B
Generally
speaking,
staying
away
from
from
smoke
because
I'm
considering
that
it's
mostly
for
like
master
tests
and
like
developers
looking
into
that
and
trying
but
I,
see
that,
like
lots
of
stuff,
constantly
failing
and
nobody's
paying
attention,
though,
and
and
you're,
actually
asked
it
and
I
think
last
meeting,
why
don't
we
run
it,
for
example,
on
named
ranges
and
I?
Don't
know
why
why
we
run
it
and.
A
B
So
the
if
we
need
it,
we
need
to
clean
it
up
to
make
sure
that
it
runs
so
they
and
actually
debate
I
run
it
on.
For
example,
if
you
look
at
that,
our
schedule
that
I
force
it
to
be
on
different
distros,
as
opposed
to
you
know,
we
can
like
add
these
trolls
and
run
it
like
how
supposed
around
and
then
another
thing
is
that
that
I
also
forcing
it
on
to
all
the
age
and,
as
I
said,
like
Dora,
wants
to
access
OVH.
A
The
most
recent
recent
master
run
looks
a
lot
better
than
the
ones
from
before.
It
failed
on
system
D
a
couple
times
and
then
deep
use
once
they're
done,
yeah
on
system
B,
but
in
in
so
deploy,
and
it
failed
once
on
see
views
which
is
not
great.
But
it's
better
than
a
lot
of
the
earlier
runs
were.