►
From YouTube: Contour Community Meeting - Oct 15, 2019
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
I
do
I
got
it
so
we're
recording
now
today,
in
the
United
States
at
least
it's
Tuesday
October
15,
2009,
teen,
contour
meeting
and
yeah.
Here
we
go
so
we
have
quite
a
few
things
to
talk
about
today.
A
few
of
us
will
be
talking
on.
My
screen
is
the
hack
MD.
Hopefully
you
can
just
read
that
URL.
If
you
want
I
can
post
it
in
the
zoom,
so
you
don't
have
to
type
in
that
long
ID,
but
other
than
that.
Let's
get
going
so
first
on
the
agenda,
assume
that's
gonna
be
Dave.
C
Morning,
so
what
have
we
done
in
the
last
since
we
spoke
the
last,
it
was
three
weeks
ago,
so
contra
mundo
release
candidate.
The
first
release
candidate
was
released
last
week.
This
is
I've
been
talking.
Those
who
come
to
this
meeting
a
couple
of
times
been
hopping
on
pretty
strongly
about
this
notion
of.
We
are
really
in
that
the
the
polishing,
the
polishing
and
preparation
phrase
for
wonder,
though,
and
I'll
see,
one
really
is
our
first
stab
at.
C
C
B
D
C
Yep,
you
can
always
find
this
on
the
kubernetes
slack
and
FAQs,
not
sure
don't
your
drug
of
choice
you
can
always
get
hub
is
where
we
do
all
our
communication
there,
from
which
I
was
not
just
as
much
as
possible
there,
which
is
basically
everything.
So
you
can
always
reach
out
to
find
us
there
all
right,
Ruby.
C
I'm
gonna
talk
a
little
bit
about
what
has
changed
between
the
beta
1
and
the
release
candidate
1,
and,
apart
from
like
the
beautiful
bug,
fixes
that
you'll
find
in
the
release
notes.
The
big
change
is
HTTP
proxy
has
moved
from
alpha
2
V
1.
We
didn't
go
through
the
beader
change
partially,
because
every
time
you
do
this,
you
lose
people
like
we
change
the
kind
in
the
version.
That
means
your
old
objects,
don't
work.
C
We
could
have
gone
alpha
feed
v1,
but
that
means
you
would
have
just
had
to
rewrite
them
all
the
way
where
we
are
at
with
htpc
ID,
in
what
we
calling
v1
is
that
we're
now
in
position
that
we're
comfortable
with
the
state
of
the
object,
the
state
agency
ID
the
state
of
the
ml
schema
in
that
we
believe
in
as
much
as
it
that
any
engine
you
can
be
certain
of
anything
that
the
changes
we
we're
we
may
make
in
the
future.
Just
like
the
changes
to
the
v1
objects
and
communities
will
be
additive.
C
Only
we're
now
in
a
point
that
we
can
give
a
lot
of
confidence
about.
You
know
your
objects
that
you
write
today,
you
know,
should
be
proxy.
V1
I'm
gonna
keep
working
for
the
foreseeable
future
years
decades,
whatever
there
might
be
a
v2
or
they
might
be
a
new
kind
of
object.
If
we,
if
given
any
ads
there
in
risky
to
but
HP
proxy,
is
stable
and
we're
supporting
it
going
forward.
C
There
have
been
three
specific
changes:
I
need
to
call
that,
if
we're
using
beta
one,
the
first
one
is
that
the
per
service
health
check
has
moved
to
a
per
rep
route
check
and
similarly,
the
per
service
strategy.
That's
the
load
on
strategy
has
moved
up
again
to
a
program.
One
there's
a
reason
behind
this.
This
is
when
you
have
health
checking
involved,
engaged
on
a
back-end
service
envoy.
C
Has
this
behavior,
where,
if
all
of
the
backends
go
stale,
it
ignores
the
health
checking
now
this
seems
like,
like
a
kind
of
crisis
of
confidence,
the
the
software
level
that
Envoy
now
doesn't
trust
itself
that
it's
health
checking
is
working
because
all
of
them
are
failing
and
it
files
into
this
file
open
mode
where
it
ignores
the
health
check
and
since
the
information
to
the
backend,
without
getting
into
the
specifics-
and
we
could
talk
later
about
that-
we
can
look.
It
says
on
the
issue.
C
We
may
have
to
change
the
configurator-
the
configuration
in
Envoy
such
that
we
can.
We
take
all
the
back-end
services
the
list,
then
we
would
merge
them
on
the
Envoy
side
into
a
single
cluster,
and
that
would
mean
the
health
check
and
load
balance
properties
apply
it
to
that
cluster.
This
is
not
I'm,
not
saying
we'll
do
that,
but
we
may
have
to
do
that.
If
then,
it's
the
only
way
we
can
resolve
this
issue
and
because
of
that
we've
and
because
we're
moving
the
HTPC
idea
to
be
one.
C
We've
moved
those
two
properties
to
the
place
that
would
reflect
that
change
now.
What
I'm,
hoping
is
that
we
don't
have.
That
is
that
when
we
resolve
this
issue
in
figuring,
how
to
configure
envoy
to
not
have
this
bizarre
fail
open
behavior,
we
will,
we
will
be
able
to
retain
one
back
in
cluster,
given
any
service
and
my
intent.
C
There
is
this
the
we
will
then
add
back
health
check
policy
and
low
balance
policy
at
the
service
level,
so
you'll
be
able
to
specify
them
at
the
rat
level,
which
is
basically
for
all
of
them,
or
if
you
want,
you
can
you'll
be
able
to
specify
at
the
individual
level.
We
just
don't
know,
but
this
is
an
insurance
policy
to
make
sure
that
the
first
the
first
week
or
two
out
of
the
gate
when
we
resolve
this
underlying
problem
keeping
great
issue,
we
do
a
deprecation,
PC
ID.
C
C
C
C
But
we
can
come
back
to
it
and
come
back
to
it
in
the
Q&A
section.
Similarly,
prefix-free
right,
I'm,
just
going
straight
up
I
did
a
bad
job
when
there
that,
a
year
ago
it
has
some
unfixable
problems
so
in
in
beetle,
in
so
in
HP
proxy
v1,
it
has
been
removed.
There
is
currently
no
replacement.
The
workaround
is
to
continue
to
use
ingress
route,
which
has
the
broken
prefix,
rewrites
or
whatever
values.
C
If
you
made
manage
to
make
it
work,
my
suspicion
is
that
we
will
have
when
we
add
it
back,
it
won't
come
back,
is
just
prefix
trying
to
come
back,
there's
more
complicated
object,
because
there
is
this
problem
of
the
trailing
slash.
Do
when
you
rewrite
slash
food
to
slash.
So
you
you
want
to
present
just
a
web
application
that
honor
on
a
sub
URL
there's
a
problem:
does
it
end
in
a
slash
it
ends
in
a
slash.
C
What
can
happen
is
that
the
euro,
presented
at
the
back
end
may
have
two
slashes
in
a
row
now
economically
in
the
URL
terms.
That's
that
doesn't
matter,
but
certain
web
applications
not
naming
any
names.
Nodejs
don't
deal
with
this
properly.
They
treat
two
slashes
in
row
different
from
business
in
book,
slash
to
do
canonicalize
the
URL.
So
to
deal
with
this,
we
cannot.
C
C
Moving
onto
HTTP
proxy
I
recognize
that
too,
that's
it's
a
blocker
now
for
us
moving
deprecating
and
removing
ingress
route,
and
it's
a
priority
for
us
to
fix,
but
we're
going
to
do
it
properly
so
that
it
works
rather
than
in
this
this
case,
we're
had
very
limited,
limited
application,
so
I'll
take
a
pause
again
and
I
should
I
should
link
to
fishes
I'll.
Do
that
in
the
minutes
afterwards,
yeah
I'll
take
a
look.
Are
there
any
questions
about
this
change?.
C
E
C
Had
heard
that
week
that
there
could
be
some
heuristic
logic,
that
would
be
a
be
able
to
do
like,
like
some
some
kind
of
some
kind
of
logic
we
could
say
always
add
the
slash
case
will
always
remove
the
/k.
Something
like
that.
But
every
time
you
look
at
it,
it
doesn't
seem
that
possible.
So
we're
gonna
have
to
come
up
with
a
more
complicated
field
that
describes
the
more
nuance
is
a
prefix
rewriting
because
of
the
problems
of
collapsing
URLs
on
you
probably.
B
C
You
cool
so
in
all
of
this.
In
all
of
these
changes,
ninja
strapped
to
HP
proxy
we've
been
keeping
a
document
of
how
you
have
to
start
leaning
this
route,
how
you
change
that
into
a
to
be
proxy,
that's
linked
below,
and
the
last
thing
I
want
to
mention
in
this
section
before
I
handed
off
to
somebody
else,
is
that
we've
written
down
the
support
policy
for
contra
now
this
was
this
is
not
it.
This
is
not
a
change
in
our
support
policy
future,
just
writing
down
what
was
implicit,
which
is.
C
C
So
that's
that's
just
a
people's
preference.
So
that's
all
the
things
that
I
have
to
go.
I'm
gonna
hand
it
back
to
Tim.
Who's
got
some
other
stuff
yeah.
B
D
C
Yes,
so
in
terms
of
Robin,
looking
back
at
the
past,
looking
looking
forward
into
the
future
as
well
as
as
well
as
all
the
code
book,
we've
been
doing
forward,
I'm
getting
ready
for
one
though
I
want
project
contour
very
the
website
to
be
more
useful
as
it
is
now,
it's
a
little
about
sort
of
a
landing
page,
but
there
there
are
two
things
that
I
want
out
of
it.
One
is
to
resolve
fee
the
kind
of
impedance
mismatch
between
the
quickstart,
which
is
how
we
talk.
C
It
would
get
started
with
control,
intelligent
gonna
use
the
latest
and
the
default
presentation.
If
you
go
to
github,
which
shows
you
master,
because
if
you
change
the
default
branch
leads
getting
weirder,
which
shows
you
master,
which
basically
shows
you
documentation
for
a
thing.
That's
coming
up
in
the
future
and
we're
gonna
resolve
that
by
moving
all
of
the
user,
the
user
focused
documentation
out
of
out
of
github
into
the
website.
So
the
website,
the
version
that
it
recommends
and
the
documentation
goes
long.
C
B
C
The
old
is
this
time,
yet
this
is
marvelous
chicken-and-egg
problem
of
how
do
you
publish
a
website
for
aversion?
You
haven't
released
it
so
that
that's
that's
just
kind
of
knowing
plumbing
on
our
site
bug
the
the
goal
to
make
it
to
make
things
easier
and
more
useful
for
the
contour
users.
Is
that
the
version
on
the
version
on
the
website
that
and
the
documentation
that
it
talks
about
are
going
to
be
going
to
be
in
sync
now,
rather
than
having
to
split
split
across
this?
C
Something
else
in
doing
this
work
is
that
we've
identified
a
set
of
documentation
which
aren't
really
version
specific,
like
they
don't
refer
to
a
particular
version.
There
is
the
perennial
like
they
applied,
they're,
just
good
advice
for
for
all
country
users,
or
they
are
documentation,
like
the
support
policy,
how
we
work
in
github
how
we
did
the
research
of
great
documents
and
those
would
be
also
moved,
moved
to
the
website
just
so
that
we
we.
C
The
goal,
is
to
clean
out
that
dog's
directory
or
to
make
sure
it's
focused
only
on
developer
documentation,
not
not
so
much
in
user
documentation.
So
there's
been
moved
to
the
resources
section
of
the
website
and
similarly,
we
found
a
bunch
of
documentation,
for
example,
how
to
set
up
tell
this
between
ongoing
contour
or
how
to
set
up
certain
manager
is
I
talked
about
a
little
bit
about
this
in
the
last
community
meeting,
which
is
this.
C
This
is
a
set
of
documentation
which
I
think
represents
kind
of
guides
like
how
to's
like
you've
any
information
on
how
like
I,
have
a
particular
problem
or
I'm
an
operator
and
I
need
to
make
sure
that
this
traffic
is
secure.
On
my
network,
like
you
come
you
come
looking
for
a
solution,
so
we're
in
the
process
of
moving
and
rewriting
that
documentation
to
directly
to
direct
directly
answer
that
and
I'm
not
sure
whether
it's
it's
merged
I
push
the
PR
last
night,
it's
a
section
called
guides
shamelessly
stealing
the
name.
C
The
project
will
put
an
end
up
in
light.
You
can
have
an
off
the
credit.
The
little
loves
contributions
for
people
who
are
switch
case
studies,
but
if
there's
a
kind
of
analog
of
a
case
study
of
like
here's,
how
I
did
of
thing
with
contour,
you
would
love
to
love
to
help.
You
write
that
up
as
a
guide,
so
that
other
others
can
learn
from
learn
from
your
experience.
I'm
pretty
sure
I
know
how
to
do
this
in
Jekyll,
but
will
give
you
an
author
credit.
You
put
your
name
up
in
lights.
C
B
Very
cool
Dave
I'm
really
excited
about
the
how
I
call
it,
how
do
I
documentation
I
think
that
that's
that's
really
going
to
give
you
super
valuable.
The
project
I
can
take
the
next
one.
It's
really
pleased
to
see
one
that
please
try
RC
one.
You
know
there
are
changes,
we're
locking
down
the
API,
so
we
would
love
your
feedback
the
earlier
the
better,
because
our
c2
is
coming
out
in
ten
days.
C
Yeah
around
the
25th
always
depends
on
Friday
put
the
finish
riding
through
all
these
notes.
On
the
spirit
on
in
the
spirit
of
dates
and
releases,
wonder
oh,
is
scheduled
for
early
November.
The
actual
milestone
says
the
1st
of
November,
and
the
reason
for
that
is.
If
you
go
back
in
time
and
look
they'll
hit,
a
blog
post
you'll
see
the
contour
first
version.
Contour
was
released
on
the
1st
of
November
2017.
C
So
there's
a
little
bit
there's
a
little
bit
of
nostalgia
there,
but
more
practically
there's
a
bunch
of
other
into
other
interactions
we
have
to
have.
There
might
be
some
some
some
emotional
work
that
we
do
over
those
couple
weeks.
The
date
will
be,
you
know
at
the
start
of
November,
but
that's
we're
still
on
track
for
that.
C
C
It's
kind
of
this
big
party,
it's
called
coupon,
it's
gonna,
be
at
the
end
of
the
month
when
there's
kind
of
another
big
party
called
Thanksgiving
for
chaplains
quite
soon
after
that,
the
the
it
the
end
result
of
this
is
that
looking
being
realistic,
we
are
not
expecting
to
do
a
wonder
one
until
early
next
year.
The
reason
I
mentioned
this
is
I've
said
this
a
few
times
that
you
know
what
we
want
to
get
English
out
is
deprecated.
C
It's
not
gonna,
get
any
any
fixes
we're
just
going
to
maintain
it
and
I
call
is
we
want
to
you
want
to
remove
it
like
we
want
to
move
forward.
There
are
now
it's
now,
just
not
just
a
matter
of
saying,
please,
which
you
move
off
it,
but
they're
also
features
which
users
need,
which
we
need
to
have
an
analogue
in.
In
HTTP
proxy,
but
the
goal
they've
got
what
I
wanna
mention.
This
is
I
want
to
assure
people.
C
The
English
route
is
a
not
going
away
until
we
have
replacement
the
replacement
for
a
prefix
for
you
right
and
then
any
other
features
at
the
moment
I
don't
know
of
which
are
missing
the
other,
the
other.
The
other
thing
is
that
people
are
very
worried
because
I
I've
written
in
the
past
things
like
once
wonder,
though,
is
that
English
is
going
away.
It
may
happen
the
IRRI
sellers.
C
C
Its
development
is
done,
finished
finito
then
finished.
It's
not
there's
not
gonna
be
back
porting
changes
to
it,
so
we
expect
everybody
to
move
on
it
under
HP,
HP,
proxy
and
perhaps
emerges
renaming.
The
type
was
a
mistake
because
it
kind
of
use
somehow
suggesting
that
they're
different
they're,
really
not
a
patriot
proxy
he's
just
be
one
of
ingress
wrap.
C
So
we're
gonna
be
talking
more
about
that.
In
the
new
year
some
some
people
have
suggested
dependent
depending
on
the
need.
We
might
write
a
translation
tool
to
help
people
translate
their
llamo
from
English
rap
to
UHP
proxy.
That
seems
this
is
a
nada
binding,
but
that
seems
like
a
fairly
straightforward
to
do
when
we
get
an
idea
of
the
corpus
of
ingress
route
out.
There
I
think
that's
probably
quite
a
bit,
but
it's
been
out
there
for
about
a
year,
but
the
sum
total
is
that
we
are.
C
We
are
motivated
to
get
that
in
your
track
code
out
of
contour
by
one
to
one
or
at
least
1.2,
because
I
hazard,
like.we
contour,
supports
I,
looked
at
it
to
support
for
different
ingress
objects,
extensions,
networking,
networking,
character,
ingress
rap
and
HTTP
proxy
engine
in
gracefully.
One
is
going
to
be
coming
at
some
point
in
the
next
couple
of
months
and
the
ingress
be
to
providing
theater
so
there's
a
maintenance
cost
in
supporting
six,
potentially
six
different
flavors
of
ingress
and
transposing
between
all
of
them.
C
As
I
said,
it's
deprecated
we're
not
going
to
be
adding
we're
not
going
to
be
back
wanting
things
to
English
that,
but
we're
not
going
to
remove
it
until
we
have
a
solid
path
forward
for
anybody
who's
using
it
because
they're
like
well,
you
just
I
can't
do
this
an
HTTP
proxy
and
that's
all
I
have
to
say
and
I
know
we're
going
to
do.
Questions
now
or
there's
some
other
stuff
to
talk
about,
but
we
can
spend
the
rest
of
the
after
the
rest
of
the
time.
B
B
B
You
know,
somewhere
between
one
and
three
months.
Is
it
no
more
than
three
months
that
is
not
yet
ratified,
but
to
help
the
development
team
and
our
community?
That's
that's
what
we're
doing
we've
run
across
folks
in
the
community
that
are
running
0.12,
which
is
you
know,
three
versions
back
really:
zero:
nine,
zero,
nine.
C
C
So
we
know
that
the
quarterly
quality
recycled,
cube
news
is
quite
aggressive
and
I
was
being
even
more
aggressive
than
that
may
not
be
actually
helping
helping
our
users,
but
the
thing
the
thing
I
do
want
to
underscore
is
the
line
after
that,
which
is
this
is
in
no
way
saying
we
are
not.
We
are
only
going
to
do
one
Bleek's
return,
that's
one
release
every
n
months,
we're
still
gonna
issue.
C
B
This
also
gives
the
team,
with
the
community
more
time,
to
sketch
up
big
ideas
and
actually
think
about
them
and
do
deeper
analysis
before
implementation.
But,
just
like
Dave
said
we
keep
fixing
bugs
along
the
way
and
patch
releases
keep
coming
part
of
this
is
we're
gonna
get
a
little
bit
more.
We're
gonna
follow
semantic
versioning
a
little
bit
better
than
we
necessarily
haven't
asked
so
you'll
have
some
pretty
strong
guarantees
about
what
a
patch
versus
a
minor
release
means.
C
Yeah
so
again,
if
you.
C
We're
actively
trying
to
get
everyone
onto
they
should
proxy,
be
one,
but
I
want
to
make
sure
we
don't
leave
anyone
behind
we're
gonna,
but
won't
be
removing
support
from
that
from
from
contour
until
there
is
a
Bible
up
use
use
case
for
the
few
people
who
are
stuck
with
the
features
that
we
have
ported
across.
Yet.
B
And
if
there
is
community
desire
for
some
sort
of
migration
tool,
we
are.
We
already
have
some
existing
art
from
the
Bolero
projects,
which
went
through
a
one
dot,
oh
release
and
a
CRD
change
and
open
source,
the
actual
CID
migration
tool,
which
is
called
very
skillfully
seed
migration
tool
that
we
may
utilize
here
very
creatively
is
the
word.
I
was
looking
for
so
razor
shop,
so
I
can
take
the
next
one.
I'll
go
ahead.
Yeah.
E
Just
a
comment
on
the
migration
tool,
so
we
had
a
call
with
one
of
our
container
registry
people
and
it's
not
really
a
ingress
route
to
h,
TP
proxy
change.
It's
changing
default
from
nginx
to
contour,
so
they're
found
in
testing
that
unless
we
inject
in
the
default
search
string,
you
know
the
config.
You
can't
just
switch
the
in
grasses
on
our
users
and
then
network.
Our
policy
breaks
as
well,
so
we're
sort
of
thinking.
E
E
Know
so,
if
you
look
at
like
a
an
ingress,
router
or
a
HTTP
proxy
sort
of
yeah
mall,
it's
got
a
default
third
string,
and
so
we've
got
people
using
just
vanilla,
nginx
or
nginx,
vanilla,
ingress,
yeah.
No,
it
doesn't
have
that
string
and
when
we
flip
to
contour
the
console
is
configs.
Don't
work
without
that
string.
C
E
E
E
Like
in
now,
I'm
not
logged
into
my
worker
VDI
here,
but
we
have
a
string
to
the
default
search
for
contract.
This.
D
E
E
C
We
should
definitely
talk
more
about
this,
because
I
would
my
knowing
nothing
about
the
migration
plans
you
you
put
in
place.
I
would
not
migrate
both
during
race
controller
and
your
objects.
At
the
same
time,
if
they
lived
in
limitations
in
the
way
that
we've
done
ingress,
we
should
be,
we
should
be
talking
about
them
and
I'm
sure
that
there
are,
but
that
to
do
both
at
the
same
time,
that's
a
big
big
migration
burden.
The.
E
C
E
Understanding
I
understand
from
talking
to
Matt
Klein
is
the
idea
behind
it
say:
you've
got
you
know,
I,
guess
five
five
nodes
in
the
cluster
and
three
died.
It's
past
a
default
50%
threshold
at
that
point,
panics
and
says:
everything's
there.
So
you
don't
sort
of
run
out
of
capacity
talking
I,
think
but
I
believe
it's
over
aidable
in
bootstrap,
but
I
need
to
what
check
was.
A
A
So
so
what
happens
with
what
we
set
up
in
contour
is
multiple
up
streams.
So
there's
a
multiple
route
object
in
envoy
and
for
each
service
in
Kerber
Nettie's
or
could
it
could
be
anything
in
terms
of
other
situations,
but
for
today's
call
talk
about
kubernetes,
but
each
service
is
basically
a
cluster
and
envoy,
and
then
we
can
define
health
checks
against
each
cluster
and
what
happens
is
if
an
entire
cluster
it
goes
unhealthy
envoy
doesn't
take
that
out
of
the
route.
It
still
applies
weights
to
each
cluster
as
you
as
if
it
was
healthy.
A
So
so
you
have
three
different
clusters:
it'll
hit
cluster
one
which
is
healthy,
cluster
2,
which
is
healthy
and
then
cluster
3
will
just
fail,
and
then
it
goes
back
to
1.
So
I
asked
some
folks
I.
Think
I
talked
to
Matt
about
this
one
specifically
once
at
one
point,
and
their
recommendation
was
that
yes,
that's
happening,
that's
kind
of
how
its
design
or
nards
how
it's
working
today
and
the
result
to
fix
it
would
be
to
use
a
locality
based
cluster.
A
So
what
we
would
do
is
have
one
single
cluster
in
envoy,
but
with
multiple
localities,
so
he
told
how
he
would
kind
of
match
each
different
upstream
Wow
and
then
we
can
apply
weights
to
different
to
each
different
locality
and
then
essentially,
we
have
the
same.
You
know
upstream
waiting
across
multiple
services,
but
your
health
check
now
what
would
work
so
if
an
entire
service
went
offline,
I
would
then
route
and
just
take
that
out
of
service
and
not
route
to
it.
Much.
A
E
C
I'm
sorry
I
said
we
would
be
definitely
like
merging
everything
into
a
single
cluster.
It's
not
ideal
because
we
would
have
to
share
the
health
check
across
that
show
the
load
balancing
to
clergy
and,
in
fact,
shared
the
upstream
upstream
certificate
validation,
which
in
fact
is
not
something
that
you
can
share
across
a
generic
set
of
backends.
So
we
are
going
to
work
extremely
hard
to
try
and
find
a
different
way
around
this.
This
augment
done
in
making
this
pre-emptive
change.
C
Eh
we
proxy,
is
basic
insurance,
hope
you
don't
have
to
call
owner
and
we
can
put
it
back
at
the
service
level,
but
it's
insurance,
the
envoys
behavior.
Yet
when
a
they
say
when
all
the
members
of
a
cluster
which
is
tribal,
wait
had
set,
go
unhealthy,
it
considers
them.
It
considers
itself
to
be
a
default
and
says
well
that
health
checks
valve
all
of
them.
I
must
be
at
fault
and
put
some
more
back
to
healthy
and,
of
course,
there
are
ten
five
hundreds.
E
I
I
think
that
I
think
the
way
Matt
described
it
to
me
was
slightly
different
and
it's
meant
for
larger
scaler
as
well.
So
say,
you've
got
a
five
thousand
no
upstream
kosta,
and
you
know
two
thousand
six.
Two
thousand
six
hundreds
failing
you
know
do
do
this:
it's
not
really
a
design
for
sort
of
small
Cuban.
Any
service
deployment.
A
Yeah
yeah,
that
panic
is
so
you
don't
get
a
false
positive,
so
like
if
things
start
failing,
it
doesn't
reflect.
You
know
negatively
across
the
rest
of
the
cluster,
so
yeah
you.
We
can't
change
that
I
believe
there's
a
threshold
we
can
set.
We
can
actually
turn
that
off
with
one
two
and
in
config
yeah
for
sure.
A
C
C
Hopefully,
I
mean
that
the
the
flipside
is
hopefully,
if
you,
if
you
had
a
bunch
of
back-end
I'm
back
in
services
and
you
have
the
same
health
check
and
load
balancing
strategies
across
all
of
them.
You
just
kind
of
tired
of
having
to
type
it
over
again.
Moving
moving
that
to
the
per
route
is
gonna,
be
more
convenient
for
you,
but
we
understand
that's
also
taking
away
its
and
flexibility.
The
read
the
reason
it
was
specified
in
our
service
is
for
the
gimbal
use
case.
We
actually
have
costs
made
up
of
multiple
some
companies.
C
Others
come
from
other
fees.
Also
note
then
OpenStack
clusters
and
they're
kind
of
combined
synthetic
objects.
So
you
actually
are
health
taking
independent
entities.
It's
not
just
they're,
not
just
the
same
pods,
just
in
different
groups
that
actually
I
represent
different
services
with
different
revisions.
It's
networking
characteristics,
so
that
was
the
original
design.
D
But
I
was
comment
like
who
talked
a
little
bit
about
the
migrate
migrating
from
other
ingress
controllers,
so
I
think
that'll
be
a
really
interesting
contribution
to
the
docs.
Referring
back
to
it
was
previous
doc.
D
C
Absolutely
and
we
can
meet
you
anywhere
anywhere
along
the
spectrum.
If
you
just
want
to
talk
to
us
in
Psych
or
just
one
rosebud
and
say
hey,
this
thing
didn't
work
all
the
way
through
to
you.
You,
you
check
out
the
repo
client
going
to
the
site
director
and
add
er.
Add
a
document
in
guides
directory.
We
can
meet
you
anywhere.
You
want
I'm
happy
to
help
copy
edit.
Your
experiences
using
this
product
are
invaluable
to
others
who
are
trying
to
get
started.
B
C
Well
so
two
things
one
I
will
fix
that
and
I'll
make
sure
this.
This
is
up-to-date
with
the
new
tribulus
who've
pushed
in
over
the
last
month.
The
other
thing
is
that,
as
part
of
the
website
work,
if
you
go
to
the
community
tab,
all
of
the
previous
meetings
on
most
of
them
and
now
referenced
on
there
I
haven't
found
all
the
links
to
the
YouTube
videos,
but
are
also
posting
on
there.
So
the
community
page,
as
well
as
listing
just
the
upcoming
stuff,
also
has
references
to
the
the
previous
meetings.
C
So
if
you
kind
of
coming
in
late,
you
can,
you
can
read
up
also
on
the
references.
Tab
are
links
to
the
gears.
We
just
added
stay
çf,
webinar
and
I
believe
a
number
of
number
of
people
speaking
a
kook
on
my
think
on
North
America,
and
so
we're
again
really
wanted
to
drive
traffic
towards
project
cantata
IO,
it's
just
a
more
everyone
loves
get
everyone
not
putting
coding
in
repose,
but
there
are
different
tools
are
different
jobs.
We
really
want
to
make
the
the
website
its
information
architecture
usable
to
answer.
C
Questions
like
I
want
to
try.
This
I
have
a
specific
use
case,
I
think
a
lot
back
to
15
10
15
years
ago,
when
I
first
came
across
an
X
I,
really
like
the
way
that
they
dedicates
so
much
of
their
website,
not
to
then
this
great
big
list
of
like
here
all
the
configuration
options
for
each
setting,
but
they
had
it
much
more
accessible.
How
do
I
set
up
last
CGI?
C
How
do
I
proxy
upstream,
which
is
really
the
solutions
that
I
was
coming
for,
like
I,
was
like
I,
think
to
replace
Apache
with
a
web
server?
How
do
I
do
this
specific
thing?
How
do
I
do
proxy
load
balancing
or
how
do
I
do
sticky
sessions
or
something
like
that
and
I
want
to
make
I
want
to
create
a
space
on
the
website
to
capture
the
information,
because
think
that's
really
going
from
a
solution.
C
Ok,
I,
have
a
question:
I
need
a
solution
to
I
think
use
a
better
a
better
way
of
presenting
that
information
rather
than
just
beat
Doc's
I
love
a
HTTP
proxy
page.
It
describes
everything
about
the
feature,
can't
doesn't
say
anything
about
why
you
want
to
it's.
It's
it's
interesting
stock,
notation
and
in
the
counterpoint
to
the
reference
documentation.
You
need
to
have
documentation
and
guides
that
are
focused
more
on
solutions
and
how
to
ice.
That
and.
B
B
A
I
decided
what
you
mentioned
on
it,
but
I'm
a
bunch
of
us
will
be
a
coop
con
in
November.
So
my
dad,
Nick
and
I
are
gonna,
have
a
talk
at
envoy
on
the
Monday
before
starts
and
then
I
know.
Steve
long
and
I
have
a
talk
about
envoy
again
a
Kubek
on
itself
and
then
we'll
be
around
the
booth
and
all
around
that
week.
So
we'd
love
to
hang
out
and
chat
in
person
if
anyone's
around
you
know,
gonna
attend
well,
I'll,
definitely
be
there.
Thank.
B
C
B
B
Well,
if
there
are
no
more
questions,
we
appreciate
you
coming
in
participating
in
this
community
with
us.
Please
reach
out
on
community
slack.
All
of
the
links
are
in
the
second
D,
as
well
as
our
community
plays
page
and,
like
I,
said
before
we
love
your
contributions,
no
matter
how
big
or
small
your
oh,
it
makes
this
project
great.
So
thank
you.