►
From YouTube: Contour Community Meeting - Sept 1, 2020
Description
September 1, 2020
News
Status output performance issue found by @mattmoor
[youngnick] I propose we fix it and cut a 1.8.1 point release.
What have we been working on?
Refactoring work
[stevek] DAG builder refactoring part 1
Issues
[stevesloka] External Name Regression
[zianke] Add clickable anchor links to the website
[jonasrosland] Will look into CII Badge Silver status
A
Hello,
everyone
and
welcome
to
yet
another
episode
of
the
contour
community
meeting
today
is
september
1st
2020
first
of
september,
getting
a
little
chillier,
which
is
good
for
for
those
that,
like
it,
I'm
gonna
start
off
with
some
news
here.
I
think
nick
you
got
the
first
portion
here.
B
I
do
I'm
just
adding
a
link
to
the
finish
doing
that
there
you
go
that's
now
link.
Could
you
click
that
open
for
me,
so
matt,
moore
from
k
native,
has
found
a
performance
slow
down
under
heavy
churn.
That
looks
like
it's
always
there,
but
it's
unless
it's
only
really
noticeable.
It's
a
massively
noticeable
on
the
heavy
churn.
So
yeah
he's
found
that
he's
found
what
appears
to
be
most
of
a
fix.
B
There's
a
second
fix
that
we'll
bring
in
as
well
there's
pr
there
that
looks
good
to
me
and
james.
So
once
we've
got
those,
I
think
we've
merged
that
one
now,
once
there's
once
we've
got
the
two
fixes
merged
into
master.
I
am
proposing
that
we
cherry
pick
those
back
to
the
release,
1.8
branch
and
cut
a
release
1.8.1.
B
I
suspect
that
we've
had
some
other
reports
of
weird
problems.
We,
where
sometimes
the
status
the
status
doesn't
update
and
things
take
a
long
time
to
converge.
This.
This
performance
problem
is
because
the
way
that
we
do
status
at
the
moment
blocks
on
the
status
update
actually
being
applied
to
the
api
server,
and
so,
if
there's
a
lot
of
status
updates,
you
can
hit
the
api
server
rate
limit,
which
means
that
then,
the
dag
updates
cannot
complete
until
the
api,
server
status,
updates,
train
and
so
there's
two.
B
So
there's
a
fix
here
where
we
matt's
fixed
that
he's
his
pr
in
is
that
we
check
the
lister
cache
to
make
sure
we've
got
the
most
recent
version
of
the
object
before
we
try
and
update
it,
which
massively
cuts
the
number
of
updates
that
are
attempted,
and
then
the
other
thing
to
do
is
to
make
a
channel
unbuffered
so
that
we're
not
blocking
in
the
dag
on
that
sort
of
thing,
so
they're
pretty
small
patches.
B
I
think
it
should
be
not
a
big
deal,
so
I
just
wanted
to
socialize
that
and
give
everyone
a
chance
to
say.
No,
please
don't
do
a
performance,
an
extra
performance
release.
I
doubt
that
will
help
but
over
to
everyone,
for
your
thoughts.
C
It
seems
reasonable
to
me
I'm
not
exactly
sure
what
the
channel
change
looks
like,
but
I
did
review
the
the
pr
that
matt
put
up
around
using
the
listers
and
that
that
looks
good
to
me.
D
Sorry,
sorry,
nick
it's
going
to
say
whoever
merges
this.
Can
you
touch
up
the
commit
message
subject
as
you
squash
it
please.
B
Yeah,
so
it's
actually
not
this,
it's
not
this
pr,
there's
a
second
one
that
is
much
smaller.
It's
right
down
the
bottom.
B
That's
the
one
yeah
yeah,
that's
the
new
one
yeah
I
I
will
merge
that
and
I'll
lgtm
and
emerge
it
and
fix
up
the
commit
message.
B
Yeah
and
then
I
will.
I
will
pr
in
opr
in
the
change
to
to
the
channel.
It's
a
I
mean
it's
literally
there's
one
spot
where
we
actually
make
that
channel
steve
and
it's
just
changing
the
mate
call
to
you
know,
add
a
channel
capacity.
D
B
Yes,
I
think
so
too
yeah,
so
that
that's
one
of
the
reasons
I
I
don't
know
if
you
heard
james,
but
I
propose
that
we,
we
actually
cut
our
1.8.1
with
these
two
fixes
included.
B
B
So
the
second
piece
will
be
making
the
channel
on
buffer-
oh
okay,
sure
yeah
yeah
yeah.
So
I
will
pr
that
in
this
morning
and
then
once
those
are
in
and
we
go
through
the
cherry,
picking
motions,
then
then
I'll
I'll
push
the
steps
to
cutter
1.8.1
and
send
some
details
to
you.
Jonas,
once
it's
done.
B
Okay,
well,
that's
that
item.
I
think
we're
on
to
the
what
have
we
been
wondering:
yep,
steve.
C
Chris
yeah,
I
just
threw
my
recent
pr
under
this
refactoring
bullet.
I
know
other
folks
have
been
working
on
refactoring
as
well,
but
I've
been
working
on
taking
some
of
the
work
that
steve
slocus
started
around
refactoring
the
code
that
does
dag
building
from
the
contour
or
ingress
apis.
It's
basically
starting
to
separate
things
out
to
be
a
little
bit
more
modular.
C
So
this
is
kind
of
the
first
pr
for
that
and
if
you're
interested
you
can
read
through
the
details
and
see
what
some
of
the
the
follow-up
areas
are,
but
yeah
that's
about
it
that
got
merged
this
morning.
A
All
right,
let's
move
to
steve
sloka.
E
Yeah
so
someone
reported
an
issue
with
external
names
and
I
did
some
work
today
on
that.
I
think
it
might
be
a
regression
where
it
doesn't
work.
So
this
is
where
you
have
an
external
name,
and
you
want
to
route.
You
know
through
contour
through
envoy
out
to
some
external
resource.
So
I
think
what
I
found
is
that
the
header
rewrite
is
not
happening
at
this
point,
so
something
poke
into
just
something:
I'm
fully
written
test
against
it
to
prove
it.
But
it's
it
looks
like
it's.
It's
not
working.
B
Okay,
well,
we
need
to
fix
that
too,
so
I'll
hold
off
on
doing
any.
I
want
to
release
stephen
to
you
until
you
get
to
the
bottom
of
that.
E
B
I
think
next
up
john.
F
B
A
Yeah,
this
is
great.
Thank
you,
so
much
I'll
almost
likely
add
this
to
a
bunch
of
other
sites
that
need
it
as
well.
So
thank
you
for
this.
Thank
you.
D
Sure
I'm
working
I'm
mainly
working
on
the
next
step
in
auth,
which
is
getting
the
extension
service,
crd
working
so
the
extension
service
cid
basically
builds
a
envoy
cluster
and
that
onboard
cluster
will
then
be
a
target
for
the
external
authorization.
D
B
Awesome
for
me,
I've
been
reviewing
prs
doing
gardening
and
I
spent
some
time
yesterday
having
a
look
into
moving
that
channel
that
we're
to
change
to
buffered
to
instead
use
one
of
the
kubernetes
client
go
work.
You
isms,
it
doesn't
look
like
it'll
work
very
easily
with
the
pattern
that
we've
got
at
the
moment,
and
we
think
the
pattern
that
we've
got
at
the
moment
is
there
for
the
reasons.
So
we
probably
won't
do
that
which
always
sucks
but
yeah.
B
So
that's
mainly
what
I'll
be
working
on,
not
a
lot
of
code
contributions
for
me
at
the
moment.
Does
anyone
else
have
anything
that
they
have
been
working
on
or
want
to
discuss?
A
So
I
got
a
small
one
so
for
for
contour,
we
just
pull
this
up
here.
A
So
for
contour
we
did
a
caia
best
practices
badge
a
little
while
ago
and
we're
at
passing
grade,
and
there
are
two
other
levels
above
it
so
silver
and
and
gold,
and
I
was
just
wondering
if
you
wanted
to
kind
of
start
looking
into
what
silver
would
look
like
the
when
you
start
looking
at
into
it.
We
already
have
a
lot
of
the
things
in
place
and
it
could
be
just
a
few
hints
essentially
on
on
how
to
make
sure
that
we
make
the
project
adhere
to
even
better
practices.
A
Many
of
them
are
already
in
there
like
have
a
code
of
conduct.
Do
you
have
a
roadmap
things
like
that
that
has
been
in
there
for
for
quite
some
time
already,
so
I'm
still
wondering
if
you
wanted
me
to
start
looking
into
this
and
see
if
we
can
raise
the
bar
a
little.
A
Yes,
all
right,
cool
I'll
I'll.
Do
that.
So
I
worked
with
steve,
chris
and
and
sloka
before.
Do
you
want
to
work
on
that
together
again.
B
Awesome,
well,
that's
all
of
our
current
business.
I
guess
the
question
that
I
would
like
to
ask
is:
does
anybody
have
any
issues
that
need
addressing
or
that
we
haven't
looked
at
for
a
while
or
are
there
any
pressing
concerns
that
people
would
like
to
look.
B
At
no
doesn't
look
like
it
lazy
consensus,
says
no
okay.
Well,
I
think
at
this
point
we
might
give
everyone
16
minutes
back
or
15
minutes
back
and
and
call
the
meeting
there.
As
always,
if
you've
got
anything,
you
need
to
ask
us.
Please
contact
us
on
slack
or
residency.