►
From YouTube: Kyma Core SIG community meeting 20190410
Description
Meeting notes: https://docs.google.com/document/d/1vWleTon7sJIk0teee4SoVezS4mR3K8TlkvXkgLJwRD8/edit#heading=h.ds9remfq6dch
A
D
E
A
Yeah,
please
guys
feel
free
to
add
your
name
to
the
attendees
and
eating
notes.
You'll
find
the
link
to
our
meeting
notes,
also
on
slack
or
on
a
github
issue,
so
I
think
we
are
ready
to
start
so
thanks
again
for
joining
people
today,
we
will
be
mainly
focusing
on
release
announcements
because,
since
its
release
date,
we
usually
talk
about
the
previous
release
and
the
newly
print
release.
Also,
we
have
an
interesting
demo
that
we
will
do
on
a
feature.
A
The
tools
I
think
since
to
release
is
already,
but
it
is
interesting
to
to
give
a
damn
about
it,
which
is
using
exactly
a
tile
in
keema
for
dynamic,
DNS
configuration
and
during
the
installation.
Also,
we
have
a
quite
an
interesting
discussion.
Thank
you
guys
for
adding
your
comments
and
suggestions
and
feedback.
I
think
we
can
also
discuss
this
if
you
want
and
if
the
time
will
allow
it.
That
would
be
great,
so
I
think
we
are
ready
to
start.
I
will
be
taking
notes
today
as
well,
since
I
will
do
very
less
talking.
E
I'm,
finding
char-char
screen
he's
unsure,
yeah,
okay
and
then
I
confirm
my
screen.
So
first
of
all,
kima
0.9
florence
is
released
and
I
just
want
to
give
you
a
couple
of
highlights.
You
can
check
it.
Of
course,
yourself
in
details,
a
big
point
right
now.
We
finally
managed
to
embed
a
native
as
default
eventing
engine
in
Chema's
or
when
you
install
kima,
you
have
you
get
clean
it
if
out-of-the-box
for
free
and
it
works,
it's
integrated
with
the
newest
stack
of
other
components.
So
we
have
also
East
your
upgraded
to
one
one
version.
E
We
have
upgraded
all
the
monitoring
stock
from
it
use
graph
ana
and
everything
is
integrated
with
the
newest
k
native,
so
yeah
you
have
really
brand
new
stuff
with
one-click
installation.
Second,
improvements
will
not
talk
about
that.
Much,
because
Jakob
will
present
that
xip
IO
wildcard
DNS
is
integrated,
so
no
more
handling
with
your
custom
DNS
settings
with
opening
certificates.
E
Couple
of
other
improvements
are
mentioned
below,
but
yeah
you
can
see
full
changelog
in
the
team.
Our
release.
One
point
here:
we
have
0.9
released,
but
today
probably
we
will
have
open
91,
which
fixes
some
strange
park
in
the
UI.
If
you
provision
some
service
from
from
service
catalog-
and
you
do
not
click
on
a
name
of
this
of
this
service
instance,
it
won't
be
created.
So
this
is
some
glitch
we
found
and
but
can
be
really
annoying.
That's
why
we
will
release
the
patch
right
now.
E
E
So
the
next
topic
is
to
really
introduce
new
version
release
one
zero.
It
will
be
soon
so
actually
it's
will.
It
should
be
available
in
about
two
weeks
about
24th
of
April,
because
in
open,
if
we
focus
also
a
lot
on
of
on
performance
on
security,
on
fixing
all
the
bugs.
We
missed
couple
of
things
to
make
this
release
one
zero,
so
we
need
to
add
mutual
TLS
as
default
enabled
option
in
East
your
service
bash.
E
We
need
also
lock
integration
because
logging
is
not
fully
integrated
yet
because
of
some
glitches
in
deployment
and
as
soon
as
we
fix
that
and
couple
of
other
bug
fixes.
We
are
ready
to
release
one
0
and
this
will
be
released
soon.
In
parallel.
We
are
so
this
one.
Zero
release
is
more
about
the
finishing
and
polishing
0.9
release
and
make
it
really
stable
and
in
parallel
we
also
started
working
on
release,
one
one
which
will
come
in
five
weeks
and
with
that
version
we
will
have
couple
of
new
nice
features.
E
First
of
all,
we
want
to
have
better
way
to
introduce
own
extensions
in
Stalone
extension
in
keema
cluster,
so
we
plan
to
have
something
like
atom
catalog.
So
this
is
more
the
way
to
to
install
own
services,
lambdas
deployments
other
things
as
as
underling
in
service
catalog,
but
we
want
to
distinguish
that
from
service
consumption,
so,
for
example,
consume.
E
Other
services
or
DCP
service
over
the
last
big
cloud
provider,
so
AWS
broker,
will
be
there,
and
this
way
you
can
also
use
services
from
AWS.
A
third
point:
assets
store
with
production,
ready
configuration
right
now.
We
have
asset
store
based
on
Mineo
component,
which
is
keeping
data
in
in
faster
in
karate
cluster,
but
we
know
has
option
to
work
in
the
Gateway
mode,
so
we
want
to
have
easy
configuration
to
use,
GCP,
storage
or
or
or
our
Amazon,
s3
or
other
storage
system
to
keep
our
assets.
E
Then
we
also
work
on
testing.
We
will
have
new
framework
for
for
our
testing
right
now
we
use
helm
tests,
which
has
a
lot
of
limitations,
and
we
started
in
my
incubator,
octopus
project,
which
will
be
our
test
runner
and
it
will
be
integrated
in
the
next
release.
We
also
focus
a
lot
of
performance
tests
we
want
to
because
we
had
already
a
couple
of
issues
from
the
community.
People
started
testing
ours.
E
Our
generic
settings
are
not
the
best
for
the
performance;
we
need
to
tweak
them
a
little
bit,
and
this
is
also
our
focus
tone
to
have
better
throughput,
latency
scalability
of
our
components.
Then
we
have
also
to
POC.
So
this
will
land
in
the
next
release,
probably
as
a
feature
toggle
integration
with
Kyary
weave,
which
is
easier
visualization,
and
we
want
to
also
out-of-the-box
support
some
of
servers
and
protect
our
services
with.
If
off,
we
want
to
use
our
Hydra
for
that
as
a
POC.
A
C
A
C
Stick
to
X
IPS
and
we're
all
clear
with
this:
okay,
yeah.
Okay,
let
me
show
my
screen.
If
I
may,
you
won't
find
anything
interesting,
really
but
yeah?
Okay,
so
yeah
does
the
documentation
we're
gonna,
discuss
it
later.
So
yeah
is
closer
to
brief
you
on
API
integration,
and
let
me
begin
with
a
remark
important
one.
C
This
issue
was
originally
a
part
of
0.8
release,
but
was
literally
discussed
during
freeze
as
G
meetings.
So
that's
why
I
bring
it
up
here
right
now,
yeah
so
as
to
X
IPE
IO
itself,
I'm
not
gonna,
labyrinth
on
this
one,
but
it
is
a
mechanism
that
provides
wildcard
DNS
for
a
given
IP
address.
So
if
you
need
to
know
something
more
for
any
further
reference,
anything
you
need
to
visit,
X,
IP,
dot,
IO
and
I'm.
Pretty
sure
you're
gonna
find
everything
you
need
there.
C
Okay,
so
let's
move
to
the
changes
we
implement
it.
So
the
default
cluster
setup
was
quite
difficult.
Previously
it
assumed
that
the
user
owns
a
domain
which
is
no
longer
the
case
and
yeah.
It
was
of
course,
it
was
possible
to
install
kima
using
the
xip
iOS
as
well,
but
it
was
not
considered
our
main
setup.
Our
default
setup
procedure
was
slightly
more
complex.
It
required
some
separate
files,
documentation,
etc,
etc.
C
The
way
it
works
now
is
basically
pretty
easy.
We
reversed
this
behavior
and
our
basic
basic
setup
is
using
ex
IP
domain.
So
if
the
user
does
not
provide
a
domain
during
the
installation,
I
mean
the
right.
Basically,
my
installer
knows
that
and
that's
where
the
xip
patch
kicks
in
it
basically
reserves
an
IP
address
in
our
DCP
project.
It
provides
a
wealth
or
dns
for
for
this
IP
and
generates
salsa
and
certificates,
yeah
and
so
the
installation
zone.
C
If
the
user,
however,
provides
their
own
domain,
nothing
will
happens
because
the
X
type
epod
just
just
keeps
the
procedure,
because
it
knows
it's
not
necessary
because
the
domain
is
is
already
there
and
so
yeah.
That's
the
main
difference.
We
we
reduced
number
of
config
files
this
way,
because
we
no
longer
needs
multiple
custom
resources.
C
So
basically,
xip
patch
is
yeah,
it's
a
separate
chart
and
it's
been
yeah.
It's
been
integrated
into
our
default
custom,
resource
and
yeah.
Then
again,
if
it
always,
it
always
is
installed.
But
if,
if
the
domain
is,
if
a
domain
is
provided,
then
the
punch
does
nothing
ready
and,
as
a
result,
we
are
able
to
install
action
on
a
cluster
in
a
simple
way.
The
installation
itself
is
simplified,
it's
easier
to
do
to
try
our
product
out
of
the
box.
C
One
more
thing
except
integration
is
a
nice
stuff
really,
but
it
is
not
suitable
for
production
environment
so
yeah,
it's
mostly
for
for
us
developers,
as
a
mean
of
you,
know,
developing
our
stuff
and
testing
it,
but
but
it's
not
suitable
for
your
production.
Thank
you
guys,
Thanks,
that's
pretty
much.
It.
A
A
B
Right
I
think
I
think
probably
Neil
Neil
is
more
up-to-date
than
I
am
regarding
Kyra.
Really.
The
question
for
my
understanding
that
Jason
was
asking
is
well
on
a
contractual
support
kind
of
level,
because
there
we've
got
lots
of
different
components
that
aren't
very
interesting.
However,
we
might
not
need
all
of
them,
so
what
does
it
mean
if
we
deactivate
or
disable
one
of
them
contractually?
What
does
it
mean?
Does
it
mean
that
the
whole
thing
gets
invalidated
or
you
can?
Can
you
maybe
shed
some
light
on
that
for
our
own
understanding.
E
E
Answer
the
second
in
custom
component,
installation
I,
think,
is
the
one
you
can
go
so
our
custom
resource
declares
which
components
you
want
to
install
you've.
If
you
do
not
need
this
component,
you
just
remove
the
line.
So
this
is
the
example.
If
you,
if
you
do
not
want
to
have
application
connector,
we
just
remove
that
and
below.
There
is
a
link
to
the
full
template.
Installer
yeah.
E
E
So
if
you
remove
Easter
or
cluster
essential,
anything
everything
will
stop
work
and
because
we
depend
on
on
that
a
lot
if
you
removed
X,
you
will
have
headless
cluster,
so
no
UI
will
work
there
because
you
cannot
sign
in
to
the
console
if
you
remove
application,
connector,
of
course,
optical
or
this
event,
application
connectivity
won't
work,
but
some
components
are
not
not
mandatory.
So
if
you
removed
from
its
use,
you
can
but
have
keeping
gras
fauna
in
place
also
doesn't
make
sense.
Because
then
you
so
monitoring
will
not
work
or
but.
D
E
B
All
these
dependencies,
some
documented
somewhere,
so
that
we
we
know
that
you
know
we're
done.
We
want
to
do
things
by
the
book
and
if
we,
if
we
had
a
clear
documentation,
that's
lessons,
for
example,
as
you
as
we
indicated,
if
we
deactivate
I,
don't
know
Prometheus,
it
doesn't
make
sense
for
us
to
have
a
fauna
or
things
like
that.
It's
that
indicated
clearly
with
a
clear
like
dependency
tree
or
something
I.
E
So
you
have
what
is
the
key
in
keema
light,
so
you
what
you
can
remove
here,
so
this
these
things
are
our
two
predefined
configurations,
so
we
can
safely,
for
example,
remove
all
this
logging
tracing
Turing.
If
you
don't
need
that,
we
do
not
install
that
in
the
in
the
mini
cube,
because
there
you
have
access
to
all
these
components.
It
doesn't
make
sense
to
take
resources
for
that,
of
course,
in
cluster.
It
does
make
sense
to
have
these
components
because
it
gives
you
observability.
E
A
E
B
D
D
So
it's
easy
actually,
once
you
know
it
is
so
straightforward,
it
always
is
in
the
beginning
and
your
way
you're
just
depending
on
some
documentation
like
okay
help
me
yeah.
E
D
I
would
like
to
install
some
help,
charts
well
I
kind
of
made
the
decision
by
now
after
looking
and
working
with
this,
that
your
stack
should
remain
like
totally
separate
from
my
own
repository
I
will
not
clone
in
your
stack
and
add
on
to
it
unless
I
was
a
developer
of
keema,
unless
I
wanted
to
extend
keema
course,
my
plugins
that
take
that
come
within
the
custom
installer
for
instance.
So
personally,
I
would
like
an
out-of-the-box
experience
that
consolidates
my
Chema's
stack.
I,
don't
have
to
touch
it.
I
don't
have
to
modify
it.
D
D
Least,
we
can
do
I,
just
have
my
own
aliases
certificates
and
all
that
and
I
never
I
never
used
the
fool
commands,
but
anyway,
what
my
am
my
issue
is
with
this
stack.
Is
that
if
you
were
to
make
it
adaptable,
you
know
adopted
by
there's
no
source
thing.
I,
don't
know
what
your
motives
are
for.
Putting
this
out
there
I
assume
it
has
many
many
motives:
I
mean
it's
like.
No,
you
wanna!
You
want
people
to
contribute.
Yes,
yes,.
E
E
D
Wanna
be
a
big
player
with
sa
PE.
You
want
to
show
something
to
the
world.
I
mean
there's
many
models
so,
and
we
are
guys
we
should
take
that.
Take
it
forward
to
the
people
contributing
to
your
open
source
repository,
so
I
would
I
would
suggest
some
opinionated
workflows
like
for
people
coming
in
from
different
angles.
Like
me,
I'm
very
helm,
opinionated
I
have
some
approaches
like
get
ups
like
or
home
file
that
I
like
to
use
in
practice,
and
it
would
be
nice
to
offer
a
way
to
show
how
we
can
deploy
extra
packages.
D
That
will
then
suddenly
be
nicely
wrapped
by
some
Android
container.
You
know
like
this
out-of-the-box
experience
that
you
offer
and
maybe
then
you
know,
expose
this
service
through
some
API
and
then
maybe
also
suggest
how
you
can
wrap
that
all
up
in
your
own
custom
home
package,
because
ultimately
this
is
what
I
would
like
to
do.
I
would,
for
instance,
just
boys
some
package,
some
some
some
open-source
home
package,
but
I
would
enrich
it
with
some,
your
CRS,
you
know
to
making
getting
picked
up
by
this
service.
Catalog
and
okay.
E
So
I
think
that
this
is
something
that
is
coming
just
in
the
next
release,
so
I
this
this
add-on
manager.
This
is
actually
this
thing,
so
we
have
a
Highland
broker
inside
of
of
kima,
so
Helen
broker
is
kind
of
open
a
broker
API
simplemente
ssin
that
is
deploying
helm,
charts.
So
the
connection
of
two
two
things
they
so
package
manager,
ham
with
open
service
broker
API,
and
we
want
to
so
it
is
supported
right
now.
E
D
This
makes
it
possible
for
me
to
install
after
I
add
a
repo
just
the
same
thing
and
I
find
my
serve.
My
services
I
can
install
it.
Yes,
okay,
then
my
next
question
would
be.
Would
you
automatically
connect?
You
know,
make
all
the
extras
available.
So
could
I
then
make
a
service
binding?
You
know
yes,.
D
His
work
for
that
you
offer
me
my
next
question
would
be.
This
is
something
important
I
want
I,
want
it
to
be
repeatable,
so
I
want
and
like
press
a
button
that
says
okay
now
these
are
the
resources
that
have
been
created
for
you,
and
you
can
add
this
into
your
own
repo
and
make
it
and
consolidate
like
like
from
a
get-ups
perspective.
So
I
want
to
test
to
run
play
around
with
this.
Okay.
E
E
E
D
E
D
Us
a
custom
extra
version
of
the
of
the
deploy
of
the
installation.
We
can,
of
course,
probably
deploy
many
installations
with
different
release.
Names,
mm-hmm
I
could
get
back
at
the
same
time.
You
know
this
is
to
release.
These
are
the
files
that
you
that
you
need
to
store,
and
then
you
deploy
this
as
this
package
on
top
of
kima.
You
will
never
get
back
to
your
current
state.
Uh-Huh.
E
D
E
I
think
that
that
is
a
good
idea,
so
I
encourage
you
to
maybe
to
try
this
hand
package
things,
and
so
we
can
start
with
that
and
actually
this
whole
package
we
can
apply
with
with
one
common.
So
this
is
maybe
because
this
is
just
a
service
class
we
can.
We
can
apply
also
from
from
from
the
CLI,
is
CI,
and
if
you
can
yeah,
maybe
just
to
some
prototype,
how
you
would
like
to
work
and
give
us
a
feedback.
D
E
A
Yeah
or
also
let
me
let
me
show
you
where
we
have
also
our
on
our
community
repo.
So
here
we
introduce
proposals,
so
our
community
repo
for
cigs
and
working
groups
for
succour
uh-huh
yeah.
You
will
see
a
directory,
that's
called
proposals
and
you
can
simply
add
your
own
MD
file
and
put
Rico.
Let's
create
something
against
it
and
yeah.
Definitely
we
can
discussed
over
that
or
just.
E
E
E
D
But
for
now,
if
I
were
you,
I
would
also
add
like
to
the
definition
of
done
to
deliver
documentation.
You
know
to
make
use
case
scenarios
to
deliver
use
case
scenarios
for
what
you
deliver.
Oh
everybody.
You
know
you
give
a
demo
to
your
crew.
You
give
the
tools
for
other
people
to
give
the
same
demo
so
so
far,
I.
Finally,
your
stacked,
just
a
description
of
what
you've
been
doing
mostly
there's
no
real
use
case,
take
me
by
the
hand
documentation
which
very
mature
packages
do
offer.
Of
course,
it's
always
goes
like
that.
E
E
D
E
Right
now,
contributors
are
100%,
SCP
employees
and
yeah.
We
are
paid
for
that,
and
but
this
is
the
strategy
of
of
the
company.
Yes,
so
we
won't
be
present
in
this
community
and
this
is
real
open
source
project.
So
we
have
some
first
contribution
from
external,
a
companies,
many
S's
or
perhaps
other
people's
interested
and,
of
course,
yeah.
We
encourage
people
to
contribute
and
and
keep
it
open.
So
that's
why
we
have
this
meeting,
for
example,
which
is
open
and
fair,
and
we
listen
to
the
community.
We
want
to
engage
with
community.
E
D
E
One
zero
is
actually
actually
in
two
weeks,
but
this
is
more
thing
that
that
we
establish
have
more
stable
version
of
of
0.9
we've
complete
features
and
complete
security
and
then
in
one
oneness,
yeah,
TLS
and
and
logging
supports.
So
this
is
missing
a
couple
of
bucks
and
then
one
one
we
have
these
these
add-ons,
but
one
one
is
13
of
May.
So
in
maybe.
E
D
E
This
is
open-source
yes,
so
you
can.
You
can
use
that
also
in
on-premise
yeah,
of
course,
yeah.
We,
you
need
to
come
to
the
configuration
that
will
work
in
in
that
because
we
test
that
only
on
GAE
and
IKS
right
now,
so
we
never
run
it
for
for
any
bare
metal
configuration.
So
I
do
not
it's
yeah,
so
there
is
nothing
special
in
there
installation
but
yeah
yeah.
There
is
always
some
corner
case
that
can
happen,
but
it's
open
source,
so
we
can
fix
it.
Yes,
yeah.
A
D
A
If
you
have
no
more
questions,
let's
jump
into
our
generic
secure
boards,
where
right
now,
usually
for
the
new
ones,
we
sometimes
have
proposals,
as
I
showed
you
for
to
be
discussed
during
our
production
sig
or
during
the
normal
working
days,
and
these
lands
with
issues
that
have
a
label,
sick
or
and
right.
Now
there
is
an
issue:
that's
in
progress
about
providing
a
proposal
that
allow
us
to
release
our
components
in
a
single
pull
requests,
because
right
now
the
process
is
that
for
a
new
component,
akima
core
component,
like
I,
mean
a
component.
A
That's
already
part
of
kima
repository,
you
push
some
code
and
then
that
needs
a
pull
request
that
will
be
merged
in
the
component
code
because
we
have
a
monorail
right
now,
then
you
need
a
second
pull
request
that
will
update
the
home
chart
to
use
they're
released
block
our
versions
that
were
built
from
that,
and
we
are
trying
to
figure
out
if
we
can
actually
have
both
of
them
in
a
single
request
which
will
improve
our
building
stability
and
speed,
also
quite
significantly.
So.
A
This
is
something
that's
going
in
progress
if
you
would
also
be
interested
in
the
proposal
that
will
be
coming
out
of
that,
then
probably
we
were
expecting
this
to
come
within
this
release.
So
probably
at
one
point
we
will
discuss
this
proposal
once
it's
ready.
If
we
found
out
that
this
is,
there
is
a
efficient
way
of
doing
and
improving
that.