►
From YouTube: Kubernetes SIG Network 20170223
Description
Kubernetes SIG network meeting. Thursday, Feb 23 2017.
A
A
B
A
A
C
Not
super
useful
in
terms
of
just
trying
to
boost
the
coverage
percentages
didn't
know.
If
anyone
else
had
any
opinions
on
it,
I
don't
want,
say,
they're
worthless
test.
It
seemed
like
they
weren't,
really
adding
a
whole
lot
of
value
other
than
testing
you
know.
Did
we
get
the
right
type
of
a
thing
and
whatnot.
A
Yeah
I,
so
I
mean
I,
haven't
looked
at
the
test
and
soup
too
much
detail
recently,
but
I
assume
there's
some
places
where
the
coverage
is
missing,
because
the
coverage
is
less
important,
but
there
are
probably
areas
where
you
know.
Big
chunks
of
mainline
code
are
are
missing
coverage
and
I
would
put
more
importance
on
that.
Rather
than
a
non
mainland,
minor
or
login
code
makes
sense
right.
A
C
For
so
it
wasn't
that
the
function
was
it
really
being
tested
accurately.
It
was
just
a
couple
pieces
of
like
if
else
I,
both
logic
maybe
got
this
and
then
in
the
iptables
code.
Also,
some
of
that
there
was
booze
that
I
found
where
they're
saying
it's
not
difficult.
It's
a
lot
more
difficult
has
an
HB
I
when
I
came
to
unit
testing,
so
the
iptables
old
stuff-
and
it
was
much
better
to
test
it
in
and
then
coverage
instead.
A
Yeah
I
think
Dan
Williams
might
be
most
knowledgeable
on
what
testing
is
Eric?
Wasn't
there
he's
not
on
the
call
now,
but
I
know,
he's
been
and
working
in
that
code
base
and
I
mean
my
opinion
is
if
you,
if
you
have
done
work
here
and
you
know
no
matter
what
bits
they
are
worth,
putting
them
out
as
a
PR
and
increasing
that
test
coverage.
A
C
A
C
It's
covered
up
on
the
community
meeting
today
because
of
the
backlog
of
flakes.
They
added
functionality
to
the
bot
to
be
able
to
tag
the
specific
sig
forts.
But
then
there
was
arguments
that
this
has
been
kind
of
tried
before
and
while
it's
helpful
to
at
least
be
able
to
select
the
label
to
see
what
flakes
you
currently
have
assigned
to
your
sig,
there
was
no
real.