►
From YouTube: Kyma Core SIG community meeting 20190313
Description
Meeting notes: https://docs.google.com/document/d/1vWleTon7sJIk0teee4SoVezS4mR3K8TlkvXkgLJwRD8/edit#heading=h.ds9remfq6dch
A
Our
meeting
okay,
so
let
me
welcome
you
on
our
next
Korra
meeting
and
today.
I
will
be
your
moderator
and
look
at
greens.
Can
we
take
the
notes
on
our
agenda
right
today?
We
have
two
top
piece
from
putable
in
ski
poof.
Of
them
will
be
about
release,
one
is
about
announcing
current
release
0.8
and
next
one,
the
Retiro
dot,
nine
scope.
After
that,
Adam
shuts
off
Cara
will
present
the
new
octopus
framework,
which
is
basically
the
handless
on
straight
and
after
all,
I
will
do
the
triage
board
from
sick
core
project.
A
B
B
B
B
So
I
will
go
maybe
directly
to
the
release
notes.
We
have
released
notes
as
a
pull
request
in
the
website
repository.
You
can
check
that
this
will
be
a
little
bit
adjusted
to
the
scope
we
have
in
the
release,
but
the
highlights
are
already
there.
We
work
on
us
promised
on
the
Canadia.
It
is
now
full
support,
4k
native,
but
still
it
is
not
default
event
back
back-end
for
in
keema
the
Canadia.
You
still
have
to
enable
it
with
future
togo.
We
enabled
backup
and
we
use
happy
to
arc
which
is
renamed
to
Valera.
B
Maybe
it's
the
name
you
you
know
it
under
and
we
introduce
the
backup
for
our
resources
and
also
for
customer
workload.
We
have
to
set
of
backups
that
you
can
enable
in
keema.
The
first
point
actually
is
removed
from
the
scope
will
be
part
of
the
next
release,
because
we
did
it
last
minute
and
it's
not
in
the
release
candidate.
B
We
removed
all
so
many
cube
constraints
right
now.
You
can
use
any
mini
cube
version
above
the
current
supported
one.
It's
not
complaining
that
the
Installer
local
installer.
We
have
also
a
lot
of
improvements
in
application
connector.
We
added
certificate,
revocation
and
renewal
to
to
the
application
connector.
So
it's
we
also
improve
security.
We
replace
also
ok
lock,
with
locking
the
first
iteration
of
available.
So
this
is
the
first
reason
a
lock
is
available.
It's
not
full
access,
full
integration.
We
do
not.
B
B
We
also
introduce
some
thing
to
improve
security
in
the
cluster
right
now
we
have
built
in
a
command
developer
role,
which
is
meant
to
be
given
to
two
developers
that
can
work
on
single
namespace.
So
so
far
we
had
admins
and
viewers
admin
could
change
whatever
settings
in
the
entire
cluster
has
cluster
administrator
role?
Kamat
developer
is
limited
to
the
namespace,
so
we
can
play
with
with
the
permissions
in
the
way.
That's
one
developer
can
do
something
in
the
stage
namespace
and
another.
B
In
in
the
production
namespace,
which
is
something
that
you
would
like
to
work
on
in
production
and
we
had
several
improvements
in
the
UI
and
all
the
views
are
not
now,
as
if
you're
in
compliance,
we
have
the
same,
look
and
feel
details
are
there.
We
also
worked
on
an
asset
store,
but
we
will
see
more
about
assets
or
appliance
in
the
next
release.
B
B
I
will
show
it
here
so
in
0.9,
which
will
be
in
four
weeks
available.
We
will
have
clay
native
integration
with
an
eventing
implemented
on
on
top
of
key
native
as
a
default,
so
no
longer
feature
toggle.
We
will
also
migrate
to
East
e11
and
we
would
enable
mutual
TLS
by
default.
This
is
a
really
big
change.
We
also
want
to
refactor
gateway
for
for
application,
connector,
which
is
going
through
and
drinks
right
now
to
use
East
your
ingress
gateway.
So
we
will
have
probably
only
one
load
balancer
required
for
the
next
of
kima.
B
We
will
also
work
further
on
asset
store
and
for
documentation
and
application
registry
assets.
It
will
be
used
as
a
default
solution
and
assets
stories
is
one
of
the
places
where
we
plan
to
to
improve
a
lot
of
functionalities
about
upgrade
backup
and
restore.
As
I
mentioned
before
we
started.
The
recording
upgrades
is
already
possible,
so
you
can
upgrade
version
open,
7,
open
8.
We
will
describe
it.
It's
not
visible
right
now
in
the
release.
B
We
also
will
improve
our
tests,
so
each
components
will
have
some
tests
that
will
verify
how
it
behaved
after
backup.
So
when
you
you
make
a
backup,
then
you
restore
it
the
content
of
the
backup
and
we
will
check
how
different
components
behave
on
that
to
make
sure
that
after
upgrade
or
emergency
restore,
we
will
have
full
functional
cluster.
We
will
also
work
on
the
security.
B
For
example,
tiller
is,
will
be
secured,
so
highly
Maxis
with
will
be
limited
only
to
the
components
that
should
have
the
access
like
Helen,
broker
or
installer
will
also
increase.
Acuity
in
service
cattle
are
going
to
enable
TLS
and
connection
to
brokers,
and
we
will
secure
nuts
yeah
make
our
solution
more
secure.
B
Other
improvements
in
UI,
better
mobile
view-
and
we
will
have
also
my
model
micro
front
ends
as
mentioned
before
Loki-
will
be
fully
integrated.
So
we
will
you
we
can
have
some
views
in
the
graph
Anna
that
will
combine
metrics
and
logs
together
thanks
to
Loki,
and
we
will
also
improve
couple
of
features
of
application
connectivity.
That's
the
the
big
points.
I
will
produce
some
blog
post
soon
as
before,
and
you
will
see
details
in
the
blog
posts.
That's
just
a
teaser.
Thank
you.
That's
all
great.
A
D
So
I
would
like
to
give
you
a
short
introduction
to
the
octopus.
Octopus.
Is
the
test
runner
for
acceptance
test,
but
I
like
to
think
about
it
as
a
home
test
on
steroids?
As
matter
previous
dimension,
and
as
you
probably
know
you,
we
are
using
fan
tests
in
the
human,
but
it
has
some
limitations,
so
we'd
like
to
address
those
limitations
and
creating
octopus.
So
here
you
can
find
a
list
of
additional
features,
for
example
selective
testing.
It
will
be
very
useful
for
development
when
you
want
to
specify
which
tests
you
want
to
execute.
D
So
we
would
like
to
add
such
future
that
you
can
specify
that
you
accept,
for
example,
that
test
can
fail
one
time
and
then
retry
it,
and
next
future
is
running
test
multiple
times.
So,
for
example,
you
develop
a
new
test
and
you
want
to
check
if
it
is
stable,
stable,
so
we
do
should
be.
You
should
be
able
to
execute
it
very
easily,
octopus
and
last
but
not
least,
full
support
for
concurrent
testing.
So
currently
we
execute
our
test
one
after
another
and
in
octopus.
D
You
would
like
to
execute
them
concurrently
and
would
like
to
add
possibility
for
developers
to
specify
how
many
concurrent
tests
should
be
executed.
At
the
same
time,
in
the
octopus
we
have
already
defined
our
custom
resources
and
you
can
find
it
in
the
docs
here.
You
can
learn
that
octopus
is
working
on
to
talk
customer
service
definitions.
First,
one
is
the
diffusion
which
specified
which
pot
you
want
to
execute
the
test,
and
the
second
one
is
the
cluster
test
suit,
which
defined
a
suite
of
tests
to
execute
and
how
to
execute
them.
D
So
let's
have
a
look
on
that
series,
so
we
here
we
have
the
sample
test
definition
and
the
most
important
thing
is
which
image
you
want
to
execute
and
which
command
you
want
to
execute.
So
here
you
can
see
that
it
is
very
simple
and
below
that
you
can
see
other
parameters
that
you
can
specify
some
of
them.
It's
not.
Some
of
that
features
is
not
yet
implemented.
So,
for
example,
in
the
API
you
can
find
the
skip
and
slack,
but
this
feature
is
not
yet
implemented.
D
D
What
is
the
concurrency
lever
and
so
on,
and
we
also
have
in
the
documentation
very
simple
tutorial,
will
show
you
how
this
to
CRD
say,
cooperate
together
and
currently
I
would
like
to
show
you
on
the
life
that
did
the
same
scenario
so
at
the
beginning,
I
have
to
create
the
these
two
custom
resources
of
this
definition
and
test
suite
so
correctly,
I'm
in
the
octopus
directory,
and
we
have
convenient
a
command
making
start.
That
applies
sample
configuration
and
we
defined
here
test
definition
and
also
test
suit.
D
Here
you
can
see
our
tests
show
the
street
nothing
very
fancy.
Right
now
and
in
a
moment,
I'm,
execute
start
I,
run
octopus
and
what
you
should
expect.
So
we
are
expecting
that
the
status
of
the
suit
and
will
be
modified,
and
here
you
probably
will
see
the
testing
pots
that
are
created.
So,
let's
have
a
look.
What
will
happen.
D
Here
we
just
execute
and
go
run.
As
you
can
see,
testing
pot
was
created.
It
was
a
very
easily
beaten
because
we
just
print
working
directory
in
that
in
that
pod.
So
it
completed
very
quickly,
and
here
you
can
see
the
status
of
the
suit
we
can
see.
The
the
whole
suit
has
succeeded,
stack
status
and
also
we
can
learn
about
the
details
for
every
test
execution.
So
we
know
that
test
example
was
executed
in
the
namespace
default.
D
It's
the
state
which
also
succeeded,
and
we
have
also
an
information
about
the
concrete
execution
which
port
was
created.
What
is
the
current
phase
of
the
to
the
port?
We
have
also
starting
time,
and
we
also
can
learn
how
how
much
time
it
takes
to
execute
the
whole
suit.
On
the
left
hand,
side
you
can
see
the
output
from
the
our
octopus
controller,
so
at
the
beginning
it
in
July
suit,
then
it
let
us
know
that
I
think
what
was
created
and,
at
the
end
it's
just
print
that
do
nothing.
D
Suit
is
finished
and
what's
next
currently,
we
have
implemented
the
very
simple
scenario,
so
we
execute
all
tests
and
we
execute
all
of
them
at
the
same
time,
because
he
was
the
single
solution,
but
of
course
we
planned
to
improve
that.
So
on
the
octopus,
you
can
find
some
issues
which
we
want
to
plant
in
a
new
future.
For
example,
we
want
to
implement
timeouts,
we
can
define
on
the
test
or
on
the
whole
suit,
the
timeout
and
after
that
it
should
be
interrupted,
and
we
want
to
add
validation
for
series.
D
We
want
to
add
possibility
to
mark
this
escape,
so
you
do
not
have
to
edit
your
docker
image
and
skip
it
inside,
but
it
would
be
possible
to
do
it
on
the
custom
resource
lever
and
the
most
important
thing
is
here,
so
improving
scheduling
so
take
into
account
the
concurrency
level
as
also
disabling
concurrency,
and
next
one
is
to
allowed
to
specify
max
retries.
So
this
is
the
remedy
for
the
flaky
test
and
implement
selective
testing
that
you
can
specify
by
providing
test
name
or
labels
for
the
before
the
test.
A
A
A
A
Second
one
in
the
review
column
is
the
proposal
about
the
future
introduction
to
our
keema
I
tried
to
reach
Ahmed
about
the
stages
of
this
proposal.
Right
now
we
have
the
pre
request,
but
it's
not
merge
and
nothing.
It's
on
this
pre
quest,
so
I
will
try
to
get
more
information
about
it
and
close
it
or
or
ping
someone
we
need
to
work
on.
It
may
be.
Someone
on
the
Cordy
knows
something
on
this.
A
Deal
with
that
right
now,
it's
quite
outdated
I
think
so,
so
let
me
leave
it
as
it
is
in
the
review
column.
Okay,
the
next
one
is
about
the
refactor
release
process
improvements
propose
this
one
was
created
by
Christian,
I
already
talked
with
him
and
right
now
it's
almost
done
the
decision
itself,
but
it
need
to
wait
for
some
Khanh
proposal.
I
will
show
you
the
ticket
here
is
providing
proposal
that
allows
editing
components
and
it
slash
in
the
chart
the
same
for
request
and
those
two
proposals
are
somehow
contradicts.
F
A
A
So
because
of
that,
I
just
moved
it
to
the
blocked
section.
Next,
one
is
about
provide
specially
for
updating
pro
cluster
right
now.
I
saw
that
is
in
the
CIC,
be
backlog.
I
talked
with
the
Daniel
and
I
think
that
it
shouldn't
have
the
seeker
level
right
now.
It's
basically
implementation,
and
it's
only
about
the
area
CI.
F
A
I
just
remove
it
from
here,
so
you
not
track
it
and
same
question.
I
have
for
our
tasks
in
the
TV
columns,
because
we
already
can
see
that
a
lot
of
tasks
from
CI
CD
backlog
and
more
or
less
they
are
not
about
the
proposal.
But
implementation
stuff,
like
you
know
the
Quaker
donors,
and
to
check
for
their
branches.
F
A
A
A
B
A
F
A
A
B
F
A
A
F
B
F
B
B
A
B
Have
maybe
some
some
proposal
should
we
maybe
have
some
discussion
next
time,
because
our
meetings
are
pretty
short
and
usually
we
finish
in
half
an
hour
right
now
we
clean
up
the
the
board,
so
it
will
be
even
shorter.
I
would
like
to
have
some
discussions
like,
for
example,
these
proposals.
We
discussed
about
the
consistency
this
the
same
touring,
maybe
some
some
other
ideas.
Maybe
we
should
use
that
time
to
really
discuss
in
the
ground
that
people
interested
in
that
this
topic
I'm
just
asking.
What
do
you
think
about
that
here?
F
B
B
About
the
preparation
for
for
such
proposal-
yes,
so
maybe
this
would
be
nice
to
have
to
gather
some
ideas
what
we
need
to
prepare
so
right
now
this
this
proposal
about
having
one
pull
requests
against,
having
separate
repository
for
charts,
how
we
can
improve
our
developer
experience.
Maybe
there
we
should
we
have
some
discussion.
Maybe
there
will
be
first
proposal,
maybe
a
third
idea
how
to
really
make
developers
like
easier
and
shortened
time
of
reviews
or
yes,.
F
And
besides
that,
PPI
below
that
I've
just
realized
that
we
somehow
missing
the
documentation
how
the
development
process
looks
like
in
chemo.
But
we
know
that
that,
for
example,
we
have
to
request
way
or
one.
Maybe
we
should
describe
it
somewhere
as
well,
because
I
just
it's
when
we'll
decide,
let's
say
to
improve
that.
Maybe
we
should
choose
or
document
that
I
mean
not
only
about
how
to
run
came
about
how
to
develop
in
chemo
with
some
insights
rights
that
what
you
need
to
take
care.
If
you
upgrade
the
component.
B
B
B
A
B
A
B
F
B
Releasing
how
we
will
see
backporting,
maybe
some
box-
this
is
a
broader
topic,
and
maybe
we
need
just
to
discuss
that
the
full
picture
and
different
aspects
for
me.
It's
really
essential
to
make
up
our
heads
how
we
want
to
work
in
a
year,
for
example,
so
this
is
more
not
about
solving
temporary
problems
with
someone
we
have
just
forgot
to
bump
the
image
of
the
of
the
component.
B
This
is
and
and
release
is
broken
before
because
of
that,
this
is
a
temporary
problem,
but
how
we
will
see
the
process
of
development
releasing
in
the
future
in
a
year,
do
we
want
to
still
have
mono
repo
or
not?
What
is
good?
What
is
what
is
bad?
What
what
are
the
pain
points
and
how
we
want
to
solve
it
in
the
long
term,.
E
I
just
noticed
that
Chris
mentioned
that
he's
on
the
call,
but
only
until
those
three
Authority
and
I
know
that
he's
from
the
community-
and
maybe
he
had
some
questions,
and
we
were
not
here
to
answer
them
because
we
were
talking
about
other
items
but
at
the
end,
I
think.
The
question
and
feedback
is
the
most
important.
E
B
G
B
G
Okay,
I'm
also
with
Chris
working
on
economies
of
we
are
more
now
and
on
the
the
user
side
of
keema
and
we
are
currently
trying
to
get
it
productive.
So
one
of
our
customers
and
I
have
now
two
questions.
So
the
first
one
is
I'm,
really
looking
forward
to
test
the
upgrade
ability
from
from
oh
seven
208.
G
B
F
It's
worth
to
mention
that
it's
the
thing
between
seven
and
eight.
Only
that
thing,
but
overall
all
of
the
components
are
SP,
be
set
package
by
helm
and
by
default
we
write
those
charts
in
the
way
that,
if
the
deployments
are
upgraded,
the
services
should
be
available.
That
means
zero,
downtime
right,
but
all
of
the
pots
will
be
restarted
and
and
that
that's
that
by
design
should
be
like
that.
We
still
need
to
test
it
a
bit
and
and
adjust
some
of
the
charts.
B
G
B
G
B
I
did
some
test
manual
tests
and
and
some
some
automation
this
it
looks
like
at
least
my
cluster.
It
it
works.
Stable
I
had
issue
with
application,
connector
and
the
gateways
Oh
gateway
is
refactored
and
I
got
new
IP
address
for
the
application,
connector
gateway.
So
you
need
to
change
the
NS
name.
So
it
is
time
downtime,
because
you
need
to
change
change
the
NS
and
to
change
also,
some
labels
on
the
application
gateway
components.
B
G
B
G
B
Yeah,
so
our
so,
this
is
the
first
version
of
kima
that
is
upgradeable
and
we
knew
that
it
will
not
be
perfect
in
the
future
releases.
We
want
to
avoid
such
situation.
So
that's
why,
in
the
in
the
release,
theme
for
for
the
0.9
is
more
upgraded
and
backup
tests,
so
we
will
learn
such
things
earlier
before
when
we
introduced
and
any
change
that
will
break
some
upgrade
abilities.
So
that's
that's
our
goal
and
neck.
B
G
Second,
one
is
so
we
with
it
some
POC
some
weeks
ago
and
where
we
also
did
let's
say
a
heavy
load
testing
on
our
team
Acosta
and
the
thing
is
we
which
are
used
more
or
less
the
configuration
of
EPO
and
and
females
else
is
provided
in
this
case.
But
the
thing
we
really
face
from
from
performance
issues.
So
then
the
use
case
was
more
or
less
to
just
have
a
bot
deployed
in
the
caster's
application
part
and
which
I've
tested
with
we
tested
several
things.
G
But
in
the
end
we
just
reduce
it
to
just
test
the
input
and
output
of
the
pod
itself.
So
more
or
less
the
number
of
requests
per
second
at
the
Paschal
candle,
the
thing
boss,
when
we,
when
we
fully
activate
it
easier
so
the
sidecar
and
also
going
over
the
the
ECU
impressed
and
they
really
had
performance
impact
like
before
it
was
1,000
requests
and
with
before
I
mean,
which
activated
is
your
side
carbs,
but
without.
G
If
you
increase
so
going
by
the
the
blink
of
an
it's
easier
to
the
bot,
we
had,
for
example,
thousand
requests
per
second,
but
going
with
fully
support.
It's
decreased
after
some
seconds
to
two
hundred
requests
and
all
the
the
response
times
itself
decrease
heavily.
So
we
invested
there
of
course,
some
time
because
the
post
is
the
application
bot
itself,
which
is
there,
was
no
no
load
on
it.
G
So
it's
pretty
ok
and-
and
we
identified
one
thing
that
that
there
is
a
restriction
on
the
east
side,
car
I'm,
a
restriction
regarding
the
virtual
CPU
power.
It's
1.0.1
I
think
so
it's
the
limit
of
the
fire-pot.
We
decrease
this
one
this.
This
helps
us
a
lot.
So
at
least
the
sidecar
wasn't
an
bottleneck
anymore,
but
we're
still
facing
issues.
When
we
go
over
the
the
ECU,
a
you
cross
itself
and
not
a
question
would
be:
do
you
do
any
kind
of
load
testing
we
have
any
references?
G
Are
you
I,
don't
know
if
they're
from
because
and
what?
What
I?
What
I
saw
that
needs
to
had
some
issues
in
the
past?
Regarding
performance,
there
were
some
some,
let's
say
some
some
issues
there,
but
it
shouldn't
be
guarantee
and
in
the
bah
1.0
version
of
EPO
and
also
the
impact
was
about
I,
don't
know
five
to
ten
percent.
Something
you
would
expect
from
the
from
the
whole
intervention.
F
G
F
Ok,
ok,
it's
not
the
case.
Iif
I
thought
because
in
the
version
of
history
we
have
right
now
it's
a
bit
buggy
about
memory,
consumption.
If
we
have
a
lot
of
site
cars,
for
example,
if
you
I
will
have
200
pots
with
side
cars
and
then
the
day
all
of
them
will
grow
in
memory,
consumption
and
as
I
know,
they
also
tackle
that
thing
in
the
upcoming
Aries
one
Oh
dot,
one
and
right
now
we
are
in
the
face
to
migrate.
To
that
is
your
version
then,
maybe,
and
and
about
the
CPU
value.
G
I
think
this
is
overwritten
in
the
index
criminal
context
somewhere,
because
I
check
the
default
Easter
settings
and
there
was
I,
think
I
think
there
was
no
restriction
in
it
and
I
just
saw.
Probably
it
was
some
optimization
for
us
for
a
for
a
miniature
contract.
I
don't
know,
but
I
found
that
we
think
they
take
I
think
there
was
some
kind
of
schema
if
your
pet
and
number.
C
F
B
G
B
G
B
G
Already
prickly
has
something
and
I
wanted
to
also
provide
some
detailed
data
on
the
richness
and
all
the
frequent
Santa
stuff
when
I
try
this
and
the
last
time
important.
That
said
anymore.
But
we
really
had
this
issue
because
we
had
back
then
with
a
couple
of
our
load
tests
and
tries
everything
around,
and
we
had
this
all
the
time.
But
the
last
time
when
I
wanted
to
get
all
the
data.
It
was
the
body
that
said
those
drink
that
there,
but.
B
If
you,
if
you
have
any
findings
here,
please
please
share
with
us
and
we
are
really
open
for
the
contribution
and
and
even
Bakke.
If
you
cannot
provide
the
solution,
because
this
is
yeah,
you
have
your
own
use
case.
We
probably
have
different
sets
of
tests
and
if
you
found
something
please
sure
we
are
yeah,
we
we
will
do
that
as
soon
as
possible.
Yes,
oh
okay,.