►
From YouTube: cluster api working group 29 August 2018
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
Yeah
I
think
it's
come
up
a
few
times
that
I
think
the
Monday
meeting
hasn't
had
good
attendance
and
it
and
it
clashes
half
an
hour.
It
overlaps
half
an
hour
with
AWS
one
says
I
think
there
was
like
do
we
still
want
to
have
it,
and
what
do
you
want
to
maybe
make
use
of
the
freed
up
slot
from
say,
class
lifecycle
me
move
to
going
weekly,
yeah.
C
So
yeah
I
mean
I'd
like
to
attend
the
a
double
yes
one
I'm
currently
chairing
the
meeting.
I've
missed
one
on
Monday,
because
bank
holiday
in
the
UK
should
have
probably
cancelled
that
yeah
forgot
that,
but
there
was
low
attendance
anyway.
Did
anybody
turn
up
this
Monday
or
we
had
like
one
person,
two
people
muddy
or
did
you
go.
C
C
Yeah,
that's
right
so
I
mean
I,
wonder
I,
wonder
whether
we
could
turn
that
meeting
into
a
half
of
one
want
to
start
with
half
an
hour
meeting
would
start
with
right,
so
it
doesn't
overlap
technically,
so
people
can
turn
up
to
one
and
still
go
to
the
other.
C
Yeah
I
was
supposed
to
look
at
those
who
originally
proposed.
That
meeting
haven't
done
so
I'm
happy
to
do
so
now
and
figure
that
out
I
mean
there
was
a
bunch
of
people
who,
yes,
oh,
that's
right,
I've
opened
up
the
doodle
and
I
could
see
that
a
bunch
of
people
take
two
hours
that
worked
for
them,
but
it
was
kind
of
all
over
the
place
and
those
a
lot
people
and
I
wasn't
able
to
really
tell
what
is
going
on
there.
So
yeah
I,
wonder
I
wonder
what
we
can
do
here.
C
A
C
Yeah
yeah
so
right,
I,
guess
Darren
that
many
people
currently
need
it.
So,
let's
perhaps
we
cut
it
down
to
not
even
half
an
hour
but
like
20
minutes
and
see
and
see
people
start
showing
up
and
the
even
more
people
start
showing
up
and
there's
more
to
discuss.
We
could
we
could
extend
it
find
different
time
slot.
C
Anything
like
that.
Yeah
I
think
if
he
couldn't
be
a
reasonable
to
to
to
propose
that
will
shorten
it
yeah,
let's
say
20
minutes,
so
people
still
have
our
little
break
before
they
joined
the
AWS,
1
or
25
minutes.
Maybe.
C
D
If
the
meetings
proved
to
be
not
fruitful,
I
don't
think
we
should
have
them
like
because
implement
our
office
hours
exist,
anyways
for
the
different
implementations,
and
we
should
probably
if
it
doesn't
become
fruitful.
We
should
eliminate
them
and
otherwise
try
to
rally
on
the
implementations
that
folks
are
actively
engaged
in
and
use
this.
This
venue,
as
the
common
clearinghouse
for
things
that
exist
across
providers.
C
B
Yeah
I'm
happy
with
that.
I
can
update
that
PR
then
I
mean
Oh
actually
was
or
just
not
have
that
yeah
just
not
merge
that
pl.
Let
me
fine.
C
A
So
I
think
we
probably
want
to
keep
the
Tuesday
meeting
just
for
now,
because
that
does
see
decent
attendance
and
there
are
some
discussions
that
happen
there
that
cross
cut
different
providers.
There's
some
general
approach,
kind
of
discussions
that
happen
there
that
are
valuable,
but
also
aren't
overly
relevant.
You
know,
outside
of
implementations,
for
the
cluster
API
that
that,
may
you
know,
cause
some
distractions
with
this
meeting.
Okay,.
A
E
Yeah
we
it's
just
a
small
change
that
we've
been
thinking
about
for
a
little
while
we
had
a
problem
with
AWS,
where
we
kick
off
the
create
for
an
AWS
instance,
and
it's
created
it's
in
state
pending,
it
doesn't
have
you
know
public
dns
assigned
yet
and
so
our
machine
status.
At
that
point
it
would
return
and
it's
you
know,
everything's
fine.
E
We
created
the
instance,
it
would
return
and
then
your
status
would
be
stuck
on
the
machine
in
terms
of
some
of
that
information,
for
you
know
up
to
ten
minutes
until
we
resync
everything
again
so
there's
there
might
be
an
issue
with
the
error
handling.
It
seems
like
we
have
a
max
retries
of
10
and
then
we
stop
and
they
all
happen
immediately.
Sweet
I
think
the
exponential
back-off
is
oddly
configured,
possibly
where
it's
it's
very
low
at
the
start
and
we
have
10.
So
it's
just
immediately.
We
run
out
of
retries
and
then
stop.
E
So
that's
arguably
a
problem,
but
it's
in
generated
controller
code
and
I
didn't
know
much
about
that
or
or
how
long
if
that's
meant
to
stay
auto-generated
or
if
we
were
just
doing
that
one
time
anyhow,
so
the
proposed
PR
just
adds
an
error
type
that
actuators
can
return.
That
indicates
they've
liked
that
machinery
cute
for
whatever
duration
into
the
future,
so
we
could
just
come
back
and
say,
try
again
in
20
seconds
and
then
that
intercepts
the
whole
process
and
you
get
your
status
updated
correctly.
So
that's
up
for
discussion.