►
From YouTube: Distribution Demo - Feb 11, 2020
Description
The Distribution team explores the k8s network ingresss apiVersion changes in Kubernetes 1.18
A
A
And
it
was
reported
that
I
believe
it
was
kubernetes
1.18
was
having
some
issues.
I
I
was
thinking
it's
this
api
version
causing
issues
there's
some
discussion
further
on
that
might
be
illuminating,
but
I'm
going
to
go
ahead
and
start
bringing
up
the
gitlab
chart
latest
gitlab
chart
with
that
extensions,
v1
beta1,
which
is
still
currently
what
we
have
spent
in
our
charts
and
see.
A
A
A
A
A
A
A
B
I
think
we
do
need
to
have
some
discussion
about
the
current
supported
versions
and
some
idea
of
what
versions
users
actively
make
use
of
the.
As
seen
in
the
note
just
behind
this,
the
terminal
there
in
116
it
was
still
extensions,
v1
beta
1.
in
118
it
became
networking.case.io,
v1
beta1
and
there's
an
automatic
transition
or
an
upgrade
of
the
model
from
extensions
v1
beta
1
into
networking
case
as
gerard's
about
to
show.
B
B
The
trick
that
we
have
to
be
concerned
with
is
yes,
there
are
ways
to
detect
the
version
of
kubernetes
in
use,
and
there
are
ways
to
tell
helm,
even
if
you're
not
actually
talking
to
kubernetes
assume
it's
this
version
we
just
have
to
figure
out
which
versions
we
want
to
actually
support
aka.
Should
we
still
be
supporting
14
and
15.
A
Right
and
I
mean
in
previous
discussions-
we've
talked
about
just
supporting
the
last
three
major
majors.
Give
me
three
minor
versions
of
kubernetes.
I
believe
so
at
some
point.
116
pretty
much
goes
away,
actually
almost
ready
to
go
away,
so
that
may
have
impact
on
our
customers
too.
But
that's
what
we're
going
to
officially
support.
C
Yeah,
I
think,
as
of
right
now,
we've
been
supporting.
Basically,
we've
been
basing
our
support
around
the
oldest
version
that
eks
and
gcp
are
supporting
is
what
we've
been
doing,
but
we
should
probably
take
a
look
and
see
if
we're
gathering
usage
ping
data
on
our
customers.
We
haven't
looked
at
that
in
terms
of
making
our
decisions
for
versions
prior
to
this,
so
be
interesting
to
see.
If
we
have
that
data.
A
A
B
D
A
A
B
A
E
E
B
B
A
B
A
B
A
A
B
A
B
There
is
an
issue
that
I
was
responding
to
earlier
today,
where
the
user
had
an
explicit
problem,
because
we
don't
use
the
networking.kates.io
for
the
ingress
they
are
using
a
aws
alb
and
because
the
default
behavior
is
to
flip
the
path
type.
So
in
the
definitions
of
the
actual
paths
for
the
hosts
there's
a
path
type
to
go
with
the
path
when
you
do
that,
it
automatically
provides
that
as
implementation
specific.
B
However,
there's
different
types,
there's
three
or
four
different
types
that
you
can
use
in
the
new
model
ingress,
it
can
be
a
prefix
so
that
it's
basically
treated
as
a
regex.
You
know
it,
you
literally
just
do.
Does
the
string
start
with
this
right
or
it's
implementation
specific,
which
is
what
we
get
in
the
alb?
B
B
A
A
B
And
this
comes
down
to
literally
like
the
apis.
The
comment
that
I
have
on
the
issue
that
you
were
going
to
try
to
demo.
I
don't
think
that
has
direct
links
to
the
ingress
apis,
oh
wait.
Maybe
it
does
the
other
one.
I
was
specifically
pointing
out
like
this
is
how
the
models
change
between
these
versions-
and
this
is
where
you
went.
B
B
B
B
To
do
that,
you
have
to
make
sure
that
when
you're,
using
the
newer
version
that
you
have
this
set
to
implementation
specific
and
then
you
have
to
tell
nginx
that
it
is
actually
a
regex
and
then
on
top
of
that,
you
have
to
use
explicitly
posix
compliant
regex,
because
the
implementation
methodology
and
the
way
it
has
to
pass
everything
around
the
across.
Who
knows
how
many
languages.
A
A
B
A
B
Right
now
nominally
this
isn't
like
a
huge
thing,
but
it's
a
behavior.
We
need
to
be
aware
of
throughout
all
the
charts,
because
we're
going
to
start
seeing
this
and
so
we'll
need
to
actually
see
exactly
what
happened
between
the
two
ingress
object
types
and
then,
because
119
and
120
support
v1,
we'll
have
to
see
what
the
difference
is
between
extensions
v,
1
beta
1
networking
k
to
v,
1
beta
1
and
the
final
v
1.
B
A
B
B
B
D
B
D
B
B
D
D
B
B
C
122
is
when
it
disappears,
but
that
being
said,
they've
been
making
so
related
to
what
jason
was
running
into
earlier
with
path
type.
Not
only
was
that
not
available,
but
you
know
in
119
they
also
changed
the
default,
so
there
is
no
default
as
of
119..
So
118,
I
think,
has
a
default
right
and
then
119
doesn't
have
a
default,
so
that
also
is
going
to
be
problematic
in
terms
of
a
difference
in
behavior
between
the
two.