►
From YouTube: Kubernetes SIG Federation 20170718
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
The
only
thing
I
had
on
the
agenda
right
now
was
the
face
to
face
meeting
timing
on
the
thread
that
a
six
variation
was
cc'd
on
seemed
like
at
least
the
core
folks
that
were
saying
absolutely
needed
to
be
there.
August
8,
which
is
a
Tuesday,
seem
to
line
up
for
everybody.
So
any
objections
for
doing
that.
B
A
C
C
In
this
case
it
was
a
bug
that
was
introduced
in
the
networking
and
Signet
working
sig.
Networking
did
not
have
coverage
for
that
particular
bug,
but
we
did,
but
it
appears
this
Federation
had
an
issue,
so
we
have
had
to
backtrack
and
since
we
only
tests
on
what
I
would
call
older
clusters,
so
we
test
on
clusters
that
are
deployed
from
midnight.
So
sometimes
the
bug
will
appear
it
be
offset
by
a
few
hours,
so
it's
harder
to
track
down.
C
C
Actually,
it's
the
interaction
between
the
deploy
job
and
the
pull
job,
so
Henry
Stacey,
oh
I,
should
I
should
specify
that
we
haven't
on
calls
documents
put
together.
That
explains
all
this,
so
anybody
who's
on
caution
definitely
familiarize
themselves
with
it.
Now
we're
in
the
position
where
there's
instances
where
it
to
be
fragile
enough,
that
we
have
to
kind
of
make
sure
that
everybody
in
communities
have
familiarize
themselves
with
at
least
part
of
this,
which
is
not
you,
don't
want
it
to
be
fragile.
C
So
to
try
to
summarize
we
audit
Buster's
every
day
at
8
a.m.
because
deploying
three
clusters
takes
about
50
minutes,
so
that
would
mean,
if
you
want
to
test
Federation,
the
Federation
test
suite
on
every
PR
would
run
for
50
minutes
plus,
however
much
time
it
takes
to
run
Federation
tests.
So
the
way
to
amortize
that
is,
we
deploy
convenient
requesters
every
day
at
8
a.m.
C
and
the
PR
testing
is
exclusively
bring
up,
Federation
control
plan
and
run
the
Federation
tests,
and
that
only
that
part
takes
25
minutes
and
that's
acceptable
for
testing
PRS
methods,
since
we're
only
using
recycling.
Essentially
those
Buster's
once
a
day.
We're
only
testing
against
the
core
kubernetes,
as
is
from
the
8
km,
deploy
job.
So
if
a
bug
is
introduced
in
core
kubernetes
and
it's
not
caught
by
any
existing
tests,
like
was
the
case
for
a
networking
bug,
then
it
will
only
appear
once
Federation
sees
it
potentially
is
next.
C
So
in
this
case
we
were,
we
were
on
call
to
try
to
fix
the
Federation
issue
which
turned
out
to
be
a
networking
issue,
so
it
just
puts
more
pressure
on
us
needlessly
and
right
now
the
code
base
isn't
changing
a
lot.
So
every
time
we
kind
of
look
at
ourselves
in
suspicion,
we're
kind
of
pretty
sure
that
it's
not
a
Federation
issue,
because
not
much
not
much
movement
in
okay,
but
it's
still
on
us
to
kind
of
speak
out
and
help
identify
others
fix
their
bugs.
E
C
Yeah
I
could
hear
just
enough
to
know
that
you're,
nothing,
a
good
decisions,
talk
but
okay,
I
mean
I.
Think
at
this
point
she
was
fairly
well
understood.
We
have
or
Regan
wears
on
fall.
We
have
a
rule
on
call
and
I
think
we
have
shishi
on
your
phone.
Maybe
it
would
be
good
if
if
somebody
else
could
work
on
the
infrastructure
portion,
which
is
the
interaction
between
the
deploy
and
the
test
jobs,
make
that
process
a
little
bit
better.
So
it's
not
such
a
burden
for
four
people
on
call.
C
I
think
what
I'm
saying
is,
if
you
can
do
something
like
find
a
way
to
maybe
in
recycle
clusters,
more
often
and
have
maybe
parallel
jobs,
recycle
plus
triggers
on
the
PR
tests,
basically
just
a
bit
more
parallelism.
The
issue
comes
that
the
deploy
portion
and
if,
while
you're
deploying
a
cluster,
somebody
tries
to
test
their
PR,
it's
going
to
fail,
so
there's
some
form
of
locking
that
has
to
be
done
there.
So
it's
essentially
infrastructure
code
that
has
to
be
written.
F
So,
for
now
what
I
can
at
least
committed
I
was
not
part
of
on
call
so
far,
so
maybe
I
could
be
added
to
that.
I,
don't
have
much
exposure
to
this
infrastructure
and
very
little
tests.
However,
she
has
so
I'll
try
to
get
some
knowledge
from
him,
and
maybe
a
week
later
or
so,
I
could
also
be
part
of
that
on
call
bTW,
stuff
and
I
think
I
see,
and
if
there
is
one
there
are
one
or
two
other
guys
who
are
partly
part
of
this
Federation
sig.
G
I
K
K
So
the
one
this
is
the
email
I
sent
out
to
Mikael.
Actually
Nicholas
is
regarding
the
Federation
art,
the
design
document.
I.
Remember
you
sent
out
about
six
months
ago
and
I
know
there
were
a
bunch
of
iterations
going
on
and
you
had
some
interaction
with
the
aft
sig
as
well.
So
we
wanted
to
find
out,
because
this
is
important
for
us
actually
be
planning
to
roll
out
a
production
release
by
October,
and
we
would
like
to
see
if
you
can.
I
I
I
K
F
K
F
J
K
F
A
Cool
is
there
anything
else
that
anyone
wanted
to
bring
up
to
discuss.
A
Of
course,
if
anything
comes
up,
you
can
always
feel
free
to
send
out
an
email.
We
can
always
chat
over.
You
know,
especially
now
that
the
meetings
are
bi-weekly
again.
More
of
that
would
probably
be
good,
rather
than
waiting
for
this
to
necessarily
talk
about,
especially
for
blocking
things.
I
think
emails
are
be
good
for
that,
if
you're
just
blocked
on
something
small
or
the
slack
channel,
rather
than
waiting
for
this
meeting
cool.
Alright,
everyone
have
a
good
week.