►
From YouTube: Keptn Community Meeting - June 28, 2021
Description
Meeting notes: https://docs.google.com/document/d/1y7a6uaN8fwFJ7IRnvtxSfgz-OGFq6u7bKN6F7NDxKPg/edit
Learn more: https://keptn.sh
Get started with tutorials: https://tutorials.keptn.sh
Join us in Slack: https://slack.keptn.sh
Star us on Github: https://github.com/keptn/keptn
Follow us on Twitter: https://twitter.com/keptnProject
Sign up to our newsletter: https://bit.ly/KeptnNews
A
Okay
good
morning
so
today
is
july
28th
and
we
have
a
developer
music.
So
germany,
you
wanted
to
discuss
duplication
of
services
ahead
of
other
things
right.
B
Yes,
oh
the
first
announced
that
by
end
of
today,
hopefully
our
latest
tomorrow,
captain
a
team
will
be
released.
The
latest
and
greatest
version
of
captain
so
stay
tuned
and
please
try
that
out
and
if
you
find
any
issue
just
report,
some
bugs
hopefully
zero.
But
you
never
know
on.
A
B
More
serious
topic:
instead,
we
planning
to
get
rid
of
help
and
jim
meter
service
from
the
captain
core
repository
and
instead
move
it
to
the
captain
contrib.
The
reasoning
for
this
is
helm
and
geometer
are
an
integration
of
captain,
so
it's
to
provide
battery
included
into
the
captain
project.
But
officially
there
are
an
integration,
they're,
not
part
of
the
orchestration
layer,
and
we
want
to
focus
the
captain
core
repository
to
be
just
focusing
on
the
orchestration
layer
and
the
integrations
can
be
found.
Then
in
captain
contrib.
B
B
C
A
So
currently,
these
services
are
managed
by
helm
already
yep.
So
I
assume
that
basically,
we
will
update
the
home
chart
servicing
new
location,
so
users
shouldn't
notice
the
difference
at
all.
A
B
D
D
I
hope
my
screen
is
visible.
Apparently,
baby
doesn't
allow
you
to
log
in
with
the
same
account
twice
so
I'm
on
two
different
machines
with
two
different
accounts.
So
you
should
see
two
of
me
in
the
meeting
a
quick
overview
of
data
doc,
so
data
dog
is
very
similar
to
grip
and
dynasty.
Actually,
it
is
a
competitive
dynamics
and
sumo
logic
and
signal
that
essentially
have
something
like
dashboards
and
check
out
all
the
charts,
and
it
has.
D
Different
formatting
for
the
matrix
it
uses
its
own,
so
it
looks
something
like
this
system
cpu.user
and
you
see
the
value
so
the
databox.
D
Integrate
the
light
provider
for
captain,
I'm
directly
going
to
go
into
the
repo
and
show
you
how
it
works,
so
here's
the
readme
and
switching
to
and
here's
the
repository
with
me
so
I've
my
captain
instance.
It
is
at
captain
0.15.0.
D
We
have
these
mini
ports
installed
the
standard
setup
except
I
have
been
the
extra
stuff
that
I've
installed
is
theta
dog
and
data
doc
service,
which
is
the
actual
sli
integration.
I
have
a
script
called
k
up,
dot
s
service
repository
where
you
can
go
and
see
what
I
have
installed.
I've
just
run
this
script,
so
it
just
installs
zero
geometer
service,
bunch
of
other
things.
D
I
have
an
example:
helm
chart
that
I'm
using,
which
sets
an
example,
is
quick
start
and
you
can
find
it
in
the
official
repository
here
as
well.
I'll
promptly
show
you
how
it
looks
in
caption,
but
essentially
I'm
adding
the
hello
service
m
chart
endpoints
the
jmeter
configurations
of
this
thing.
Then
I
come
down.
I
have
datadog
and
then
I
come
down
and
then
I
install
the
debug
status.
You
need
a
data
api
key
and
app
key
for
that
and
at
the
very
end
I
trigger
a
directory.
D
So
let's
go
to
captain
so
I
have
boot
forwarded.
This
boot
forwarded
my
api
engine
export
and
I
lost.
D
All
right,
so
this
is
this-
is
where
my
previous
experiments,
but
this
is
what
it
looks
like.
I
am
trying
to
attract,
release
zero
one
one,
hello
service.
I
will
quickly
go
over
what
I'm
going
to
do
as
part
of
demo
and
show
you
the
sli
and
slo
as
well.
So
this
is
the
healer
service.
Is
the
service
I
want
to
deploy
and
I'm
going
to
deploy
a
zero
one,
one
or
zero
one.
Two
version
of
the
service
the
sli
sits
here.
D
I
have
added
explanation
around
this.
A
lot
you
can
go
and
read
it
in
the
official
repository
it
essentially
looks
at
the
response
time
the
http
response
time
for
the
header
service,
the
hello
service
is
a
community
service
which
sits
here.
D
There's
one
peculiar
thing,
though,
where
in
datadog
you
need
to
add
special
annotations,
so
that
datadog
enables
an
integration
which
is
called
http
check
which
allows
you
to
capture
the
response
time
as
a
metric
in
detail.
This
is
the
only
difference.
Otherwise,
the
services
part
of
is
taken
from
our
examples
from
kept
mean
captain
repository,
so
the
sli
is
http
response
time.
It's
average
http
response
time
for
the
duration
of
the
test
and
the
slo
basically
says
if
the
response
time
is
less
than
100
milliseconds.
D
D
I'm
just
using
the
standard
meter,
config
thing
slightly
with
a
slight
difference:
it's
like
I've
changed
the
values
to
a
little
higher
values
so
that
you
can
see
the
test
run
for
a
minute
to
see
a
nice
sli
result
any
questions
until
now.
D
Okay,
I
will
trigger
the
delivery
and
to
show
how
it
looks
me
like
the
sample.
D
D
D
D
Can
you
can
you
see
that
one
more
time
any
version
restrictions.
A
For
datadog,
do
you
just
take
the
latest
spec
of
the
service.
D
Oh
yeah,
that's
that's
a
good
question
so
right
now
I
just
stick
the
latest
version,
but
I
should
definitely
be
billing
down
to
some
version.
I
recently
changed.
The
versioning
strategy
for
cap
deal
lock
service
from
like
zero
mapping
to
zero
one
understanding,
the
captain's
versioning
strategy,
but
I
think
we
definitely
need
a
mapping
to
say
what
words
of
support
and
I
think
that's
a
good
point.
A
D
I
think
I
use
the
I
believe
it's
the
video
version,
but
I
can
quickly
go
and
check
that
and
get
back
to
you,
but
I
believe
it's
the
latest
thing.
I
think
it's
the
v2
version,
because
it
is
that
I
think
you
can
quickly
go
and
check.
That's.
A
good
question
is
that
oh,
it's
v1,
okay
looks
like
I'm
wrong,
so
it's
seems
to
be
using
the
client
for
1.
A
D
A
I
was
just
wondering
because
data
docker
is
in
service
right
quite
rapidly,.
D
Okay,
yeah
so
yeah,
I
remember
now
so
we
do
is,
I
I
believe
it
supports
extra
stuff
and
b1
is
still
supported
if
I'm
not
wrong,
so
the
the
api
we
are
trying
to
access
for
matrix
is
a
part
of
v1
and
it
it
gives
you
like.
I
believe
the
metric
is
one
is
part
of
e1
and
anything
extra
goes
into
v2,
but
then
I
might
be
wrong,
so
I
can
get
back
here
but
well.
I
think
that.
D
Okay,
yep,
it
looks
like
a
pass,
so
I'm.
D
D
So
if
we
go
back
to
inbox
service
logs,
you
see
so
you
can
see.
The
query
was
this
everything
down
to
this,
so
we
are
looking
at.
D
This
is
a
query
that
is
sent
to
data
doc,
so
average
network
http
response
time
by
hello,
service
and
roll
up
to
the
amount
of
time
it
took
to
run
the
test
from
unix
epoch
this
to
your
next
process.
So
I
can
quickly
show
you
what
it
looks
like
indeed
block
just
for
cross
verification.
D
So
if
I
go
to
data
dog
matrix
explorer
and
I'm
going
to
switch
to
the
query
mode
I'll
advance
mode
and
say
this
is
my
query
and
the
from
end
to
right
here
so
right
now,
the
from
in
the
unix
seconds.
Is
this
I'll
go
and
paste
it
here
and
then
do
it
too
paste
here
and
the
value
we
should
have?
Is
this
so
this
this
is
one
second
y
is
showing
okay,
it's
showing
two
minutes.
Okay,
all
right!
D
So
if
you
see
the
value
is
24.35
milliseconds,
so
sometimes
dog
shows
this
range
slightly
longer
than
what
we
are.
So
the
value
we
actually
want
to
see
is
this.
So
if
you
look
at
the
this
is
the
value
we
want
to
see
and
you
can
see
it
in
0.024
seconds,
which
is
24.7
milliseconds
and
it
just
shows
it
has
milliseconds
here
it
closes
in
seconds
here.
D
I
think
that
is
pretty
much
it.
I
would
like
for
the
community
to
try
out
and
give
more
feedback
around,
and
that
sits
here.
You
can
just
go
and
run
the
script
in
kf.sh,
here's
the
list
of
open
issues
and
I'm
still
working
on
it,
and
I
want
to
support
other
newer
futures.
A
D
D
A
D
For
adoption,
so
I
I
the
current
status
of
the
service.
Where
is
it
right
now.
D
I
think,
from
my
end,
I
think
it's
in
a
usable
state
right
now.
What
I
really
like
is
alpha
users
and
give
me
more
feedback.
It's.
It's
certainly
adds
rough
edges
and
I'd
like
to
hammer
them
out,
but
what
I'm
really
looking
at
at
this
point
is:
I
want
more
users
to
try
it's
more
in
a
alpha
state
and
it
is
also
available
on
artifact
hub
right
now.
The
home
chart
is
not
available
yet,
but
the
package
itself
is
available
on
artifact
hub,
so
you
can
check
it
out.
D
D
There
you
go:
oh
it's
available
on
your
club,
but
if
you're
interested
you
can
check
that
out.
If
there
are
no
more
questions,
I
would
like
to
stop
here
and
we
can
go
to
the
next
update.
Okay,.
A
So
yeah
spend
some
time,
maybe
next
week
or
when
I'm
back
from
vacation
to
promote
data
talk
service.
So
maybe
we
could
drive
in
some
users.
A
D
Yep,
thank
you
so
much
I'm
going
to
stop
sharing
the
screen
and
yep.
That's
it!
Thank
you.
Everyone.
E
Yeah,
I
believe,
I'm
next
all
right.
You
should
see
my
browser
now
with
the
meeting
notes
open
so
today
I
want
to
show
you
a
new
endpoint
that
has
been
introduced
in
or
will
be
introduced
in
captain
018,
which
is
the
new
sequence
execution
api
endpoint.
So
why
has
this
been
introduced
so
up
until
now?
E
And
up
until
now
it
has
been
not
exactly
easy
to
actually
find
the
stalling
sequence,
which
is
causing
other
sequences
to
not
be
able
to
be
executed.
So.
E
And
where
pretty
much
all
the
information
that
gp
needs
to
orchestrate
them
is
in
there,
and
this
will
also
hopefully
help
you
as
a
user
to
to
better
identify,
what's
going
on
in
your
sequence
and
yeah,
so
we
have
a
mandatory
project,
parameter
clear
parameter.
E
You
can
see
the
the
documentation
of
the
endpoint
in
the
swagger
dock
of
the
control
plane
component
of
our
api
and
in
addition
to
the
project
you
can
filter
on
the
stage
the
service,
then
maybe
the
name
of
the
sequence
and
what's
really
important
the
the
status
of
the
sequence.
So
basically,
let's
imagine
the
scenario
where
we
want
to
identify
a
blocking
sequence
or
actually
a
sequence-
that's
currently
in
execution.
E
So
for
that
we
can
go
to
this
api
and
set
the
project
parameter
and
to
find
the
currently
active
sequence.
We
set
the
status
filter
to
start
it
and
if
you
reload
that
we
have
now
an
empty
list
because
no
sequence
is
running.
So
let's
change
that
and
trigger
a
delivery
sequence.
E
So
now
that
has
been
started.
So
we
are
currently
deploying
our
beloved
sock
shop,
and
now
we
see
that
this
endpoint
actually
returns.
Something
let
me
open
that
in
a
formatted,
json
format
and
yeah.
So
if
you
want
to
see,
for
example,
the
captain
context
of
that
sequence,
you
will
find
that
here
in
the
scope
property.
E
So
here
you
will
see
the
the
captain
context
and
using
this
id
you
can,
for
example,
navigate
to
the
to
the
to
this
sequence
within
the
bridge
by
using
the
id
within
the
deep
link
or
you
could,
for
example,
directly
use
that
id
to
actually
cancel
the
sequence
within
the
sequence
control
api.
So
in
that
case
you
could
send
a
request
for
again
the
project
then
paste
the
sequence
in
there
and
send
the
console
command,
and
if
that
was
such
a
solving
sequence,
that
was
preventing
other
sequences
from
being
executed.
E
This
would
then
free
up
the
queue
for
new
sequences
to
be
executed,
eventually
yeah.
So
that's
pretty
much
it
for
now.
Are
there
any
questions
about
this
new
endpoint.
G
E
Yeah,
we
actually
had
a
discussion
on
that,
so
initially
it
was
a
path
parameter,
but
since
from
a
semantic
point
of
view
it
is
actually
a
a
filter.
G
G
And
another
thing
so
in
this
endpoint,
only
the
running
sequences
are
in
the
right
so
running
once
that
is
being
executed.
Look
you
can.
E
Filter
for
you
can
filter
for
all
sequences
as
well.
Just
in
this
particular
example,
I've
set
the
status
to
start
it
because
that's
usually
the
way
to
identify
a
blocking
sequence.
But
if
you
just
set
the
project,
then
you
will
get
a
huge
list
of
basically
all
sequences
that
have
been
executed
and
the.
G
H
You
should
see
my
screen
now:
okay,
I'm
quickly,
gonna
show
what
I
did
this
sprint.
So,
first
of
all,
I
mostly
worked
on
pocs
actually,
and
I
managed
to
finish
the
poc
for
captain
images
in
the
arm
architecture.
H
And
they're,
like
the
most
important
thing,
is
basically
that
we
set
go
os
and
go
arch
to
linux
and
then
r64
and
then
basically
go
will
compile
an
arm
compatible,
go
binary
for
you,
and
basically
I
did
that
for
all
our
dockerfiles
with
some
caveats
that,
for
example,
you
cannot-
or
at
least
for
us
it
didn't
work
to
use
this
link
mode
external
mode,
which
basically
tells
go
to
dynamically
link
our
binaries
and
not
statically,
link
them,
but
yeah,
so
that
there
is
some
caveats.
H
But
basically
I
managed
to
get
a
full
set
of
incompatible
images
from
from
captain
and
it's
even
built
directly
in
the
pipeline
using
docker
build
x
and
the
cuimo
emulator.
And
then
you
basically
find
in
your
docker
build
push
action
which
builds
our
images.
You
define
the
platforms
and
there
we
have
now,
in
addition
to
amd64,
also
rm64.
H
H
I
did
actually
manage
to
to
run
this
on
an
on
an
arm.
Karates
cluster
with
arm
nodes
inside
basically-
and
there
again,
we
have
a
caveat
because
we're
using
vietnamese
mongodb,
which
doesn't
actually
support
arm.
So
the
only
way
to
run
captain
on
arm
would
be
with
an
externally
hosted
mongodb,
which
we
do
actually
support.
So
you
would
adjust
your
ham
values
accordingly.
Basically,
and
then
you
get
a
full
set
of
of
working
captain
on
arm.
H
I
did
actually
also
document
all
my
findings
in
here
directly
in
the
in
the
pull
request.
If
you
want
to
check
that
out,
there's
also
some
nice
references
on
on
how
to
how
to
do
what
I
did
there.
Basically,
and
luckily
the
integration
tests
are
still
green.
So
that's
all
good,
even
though
those
run
on
x86,
of
course,
for
now
still
that's
the
first
thing.
Any
questions
about
that.
A
So,
for
demand
purposes
you
use
only
services
which
are
part
of
capital
right.
Yes,
it's
only
kept
the
course
and
they
usually,
we
would
need
to
update
other
services
to
run
them,
but
yeah.
A
You
want
to
support
yeah
on
which
container
runtime
did
you
test
it
by
the
way?
Sorry
on
the
beach
continue
runtime.
Did
you
test
it
because
last
time
I
tried
this
images
there
were
some
wishes
is
creo,
for
example.
H
This
was
so,
I
tested
it
on
a
cluster
on
gcp,
which
uses
container
d
in
the
background,
and
I
did
also
ask
a
friend
who
has
an
m1
macbook
with
k3s
directly
installed
on
the
machine
and
that
also
worked,
which
is
cool.
That's
pretty
cool
yeah.
F
A
But
yeah
it's
actually
nice
and
moreover,
we
should
have
no
problem
with
having
hybrid
captain
instances.
So,
let's
say
orchestration
playing
on
arm.
H
A
H
Yeah
exactly
I
left
the
jmeter,
but
I
did
include
the
helm
service
for
now.
So
I
haven't
got
this.
Oh
hamster
has
got
the
same
treatment.
H
J
meter
yeah.
I
just
did
some
minor
adjustments,
basically
some
refactoring,
but
I
didn't
manage
to
actually
get
it
running
or
not.
H
H
A
H
Okay,
then
I'm
gonna
go
to
my
next
plc.
I
did
a
poc
to
introduce
sneak
a
security
tool.
I
already
showed
last
last
developer
meeting
that
I
introduced
a
pipeline
with
some
security
scans.
H
It's
down
here
somewhere
and
all
is
still
red
and
we
just
need
to
still
work
out
some
stuff
but
yeah,
and
I
wanted
to
introduce
another
tool
there,
basically
to
have
more
broader
coverage
of
testing,
and
I
did
introduce
sneak
here.
This
is
just
the
issue
to
document
my
research
findings
basically
and
the
request
you
can
find
here.
H
H
So
we
should
see
security
results
directly
in
the
dashboard
inside
our
github
wrapper,
which
is
pretty
cool
but
yeah.
That's
yet
another
tool,
basically
in
our
arsenal
and
then
one
more
thing
actually
resulting
from
this.
I
already
worked
on
some
security
improvements,
this
time
for
nuts
in
our
hem
chart,
and
I
basically
added
some
best
practice
security
settings
here.
H
So
we
run
as
non-root
now
with
different
user
id.
I
added
some
resource
limits
and
requests.
H
Those
basically
come
from
integration,
test,
monitoring
and-
and
I
added
some
generous
amount
of
resources
on
top,
so
that
this
should
be
fine,
where
you
could
maybe
fine
tune
this
in
the
future
yeah
and
then
I
also
added,
like
common
security
settings
run,
is
run
non-root
no
prevesk
read
on
the
root
file
system,
drop-down
capabilities
and
stuff,
like
that.
I
also
disabled
this
exported.
This
is
the
what's
it
called
not
prometheus
exporter,
we're
not
using
it
actively
in
captain
core
so
for
default.
This
is
disabled
now,
but
you
can
re-enable.
H
Just
by
setting
the
handbag
basically
yeah-
and
this
is
just
some
of
our
testing
values-
any
questions
about
this
stuff.
H
Cool
actually
one
more
thing:
I
still
have
it
open
here.
This
is
still
for
the
arm
images.
You
will
see
that,
for
my
pull
request,
images
that
I
have
right
now,
docker
half
those
are
actually
now
built
on
md64
and
r64,
just
to
show
it
off
here
I'll
download,
but
that's
it
so
I
am
gonna
hand
over
to
battle.
I
I'm
gonna
show
you
three
little
things
one.
The
first
thing
is
has
to
do
with
the
cli.
I
As
you
know,
the
recent
version
of
captain
actually
yeah
requires
you
to
set
up
a
git
upstream
and
you
can
create
project
using
the
cli
and
when
you
did
not
require
did
not
pass
the
information
for
a
git
repository.
It
actually
said
to
you
that.
I
Actually,
this
is
not.
This
will
not
be
possible
in
the
future
anymore,
so
you
got
a
warning
actually
and
with
this
ticket
we
try
to
remove
that
warning.
So
with
the
most
recent
version
of
captain,
then,
actually
you
need
to
pass
that
information
to
do
a
captain
create
project
using
the
captaincli
here
just
showed
how
this
looked
like
how
this
looks
like
when
you're
using
the
cli
cap
and
create
project.
It
just
gives
you
the
error
that
you
must
give
a
remote
url
to
a
git
repository.
I
So
the
feature
that
is
actually,
in
the
background,
provisioning
negative
repository
for
you
automatically
whether
this
feature
is
enabled
or
not,
and
depending
on
that
will
give
you
a
warning
or
not.
So
if
it
is
enabled,
then
it's
still
of
course
possible
to
just
use
a
captain
create
project
block
command
because
in
the
back,
captain
will
take
care
of
setting
up
the
required
git
repository.
I
Basically
was
a
small
issue,
but
I
took
care
that
this
functionality
is
actually
well
tested.
It
was
not.
It
was
not
really
testable
at
the
first
place,
as
so
many
places
of
the
cli.
Unfortunately,
so
I
just
reflected
a
little
bit
and
basically
moved
out
the
functionality
between
its
into
an
own
component,
which
is
leaving
the
internal
project
creator
package.
It's
called
project
crater
and
this
one
is
very
well
tested
and
should
cover
or
yeah
okay,
this
that
could
could
appear
in
this
context,
good.
I
C
Problem
we
continue
to
see
is
that
people
initialize
that
github
repo
and
I
you
know
you
go
to
the
github
page
and
it
makes
you
feel
really
guilty.
If
you
don't
initialize
it
and
put
a
read
me
in,
and
people
are
succumbing,
they
get
errors
but
they're
bizarre
errors,
they're,
showing
up
on
the
help
channel.
What
happened
and
somebody
says,
did
you
initialize
that
thing
you
can't
do
that?
Is
there
any
possibility
of
having
an
explicit
warning
that
comes
out
that
says
this
repo
has
been
initialized
and
you
can't
do
that
or
something.
I
Depends
a
little
bit
what
we
get
back
from
the
captain
api,
but
the
thing
that
would
be
durable,
yes,
and
I
think
that
would
be
also
very
valuable.
Yeah
yeah.
G
I
I
Christian
actually
saw
that
now
we
have
a
experimental
go
sdk
in
place
which
can
be
used
for
creating
simple
capital
integrations,
and
he
just
mentioned
that.
Actually,
some
parts
of
this
implementation
of
the
go
sdk
could.
I
Could
be
enhanced
with
respect
to
test
coverage
and
I've
just
grabbed
that
issue
and
increased
the
coverage
of
the
go
sdk
we
didn't
saw
that
immediately
because
goes
to
key
is
kind
of
experimental.
It's
always
changing
and
it's
also
not
advertised
to
be
stable
and
so
on,
and
we
didn't
have
code
coverage
reported
for
the
whole.
Go
utils
go
utils
repository,
so
two
changes.
I
did
here.
I
Of
course,
I've
made
some
refactorings
and
and
so
on
and
so
forth,
to
increase
the
parts
of
the
go
sdk
that
are
really
not
where
covered
by
unit
tests
and
I've
also.
I've
also
included
code
cuff
for
the
go
utils
project
which
yeah
is
now
also
merged,
and
we
have
here
when
we
look
at
codecof.io,
also
go
utils
in
there
and
see,
for
example,
that,
as
the
currently
sdk
currently
has
a
coverage
of
73
percent,
we
also
have
very
low
coverage
for
other
parts.
But
that
has
other
reasons.
I
For
example,
we
have
a
lot
of
models
drugs
in
there
which
are
not
really
tested,
but
there
are
lots
of
lots
of
lines
of
codes
and
so
on,
but
basically
we
have
an
hour
coverage
thing
here.
You
also
see
here
the
sap
connector,
for
example
it's
currently
at
80
percent
and
so
on.
So
now
we
have
a
way
to
track
this
and
improve
on
that
for
this
repository
as
well.
I
What
do
we
have
here
here?
All
of
the
sudden,
the
control
plane
was
not
reachable
anymore.
The
library
could
not
ping
that
component.
It
would
retry
a
couple
of
times
and
give
you
then
a
warning
that
it
reached
the
maximum
number
of
attempts
to
ping
the
control
plane,
and
then
it
will
basically
yeah
start
the
shutting
shutdown
sequence
and,
as
you
can
see,
it's
also
gracefully
shut
down
all
the
components
which
are
running
in
the
background
and
in
the
end,
stopping
the
subscription
source,
stopping
event
source
and
then
exiting.
I
Showed
the
new
read
me
of
the
goya
deals
repository.
I
B
I
G
Okay,
so
near
bridgeside
we
have
some
performance
improvements
because
we
had
the
project
bowling
in
the
bridge
for
pretty
much
every
page.
So
if
we
are
on
the
project
selection
of
their
project
board,
then
the
project
building
happened
in
the
environment
screen
in
the
service
screen,
etc.
So
for
oil
screens,
this
is
now
has
now
been
moved
to
the
environment
screen.
Previously
it
was
global
because
there
were
some
dependencies
in
other
screens
and
yeah
that
they're
now
eliminated
and
project
boiling
only
happens
now
in
the
environment
screen.
G
G
Then
the
next
issue-
yeah
there
was
a
bug
for
the
evaluations,
so
there
was
some
wrapping
because
we
used
incorrectly
used
aristo
component
for
the
evaluation
patch
and
yeah.
Now
we
wrote
our
own
component
and
now
there
is
no
wrapping
anymore
or
something
that
show
moi
is
displayed
instead
of
the
whole
relation
yeah,
and
this
is
now
also
fixed
yeah
and
that's
pretty
much
from
my
set.
So
are
any
questions.
F
Thank
you
very
much
from
my
site,
also
just
some
small
improvements
on
the
bridge
site.
F
We
did
some
ref
refactoring
in
the
project.
Settings
rxifying
it
and
yeah.
Basically,
nothing
changed
there
in
the
ui.
It's
just
about
maintainability
of
the
code,
so
yeah,
basically
that
the
state
in
the
page
is
yeah
in
one
object.
Some
loading
indicators
were
combined
there
and
yeah
just
to
make
the
code
more
maintainable
and
two
bug
fixes
from
my
site.
One
was
in
the
evaluation
port.
If,
in
the
configuration
change
in
the
payload,
the
image
was
missing,
then
the
evaluation
board
was
not
rendering
at
all.
F
So
this
is
fixed
now
yeah,
I
don't
have
a
screenshot
here,
but
yeah.
This
was
just
because
we
access
this
property
and
we
had
yeah
no
null
check
there
for
that.
One
and
another
bug
fix
that
is
in
the
pipeline,
so
not
yet
merged,
but
will
be
in
the
release.
That's
upcoming
is
the
detection
of
pending
changes
when
automating
provision
automatic
provisioning
is
activated.
F
This
was
found
by
christian
yeah
when
you
visit
the
project
settings
page
and
you
have
automatic
provisioning
enabled
then
previously
it
would
just
show
this
two
buttons,
https
and
ssh,
and
when
you
try
to
move
to
another
page,
although
you
did
not
change
any
settings
there,
it
would.
A
F
You,
if
you
want
to
discard
your
changes
and
leave
the
page
or
so
this
the
change
detector
hits
there
and
yeah.
You
don't
see
the
configured
upstream,
so
this
is
now
fixed
if
there
is
upstream
configured,
although
you
have
automatic
provisioning,
it
would
just
show
this
configured
upstream
and
yeah.
F
When
you
create
the
project,
it
will
allow
you
to
select
no
upstream,
but
in
the
configuration
in
this
when
the
project
exists,
it
will
just
show
that
the
existing
configuration
here
yeah
if
there
are
no
questions,
that's
all
from
my
site
and
I'm
handing
over
to
andre.
J
J
J
So,
for
example,
if
the
admin
wants
to
forbidden
a
certain
url
for
creating
a
project,
for
example,
github
or
bigbucket,
he
just
needs
to
create
a
values
yaml
file
and
apply
it
via
minus
minus
values,
option
during
the
install
or
upgrade-
and
in
this
case
it
looks
like
that
that
when
the
project
is
created
or
updated,
this
denial
list
is
checked
and
if
the
updated
url
matches
via
the
regex,
some
of
the
denial
list,
some
of
the
items
in
the
dinner
list,
where
each
line
is
treated
as
a
one
regex
it
for
bits,
creating
or
updating
the
project
and
user
needs
to
use
another
url
I'll
jump
to
the
second
one,
which
is
introducing
a
new
web
config
migrator.
J
As
you
all
know,
we
recently
introduced
a
new
web
book.
Config
version,
which
is
the
v1
beta
1
version,
now
being
introduced,
a
migrator
which
will
migrate
all
the
web
configurations
from
v1
alpha
1
version
to
v1
data,
one
version
we
created
it
as
a
separate
new
cli
command,
which
will
basically
work
like
capture
migrate
web
hooks,
where
all
the
projects
are
retrieved
and
for
each
project.
The
web
hooks
are
retrieved
and
basically
migrated
to
the
if
they
are
in
the
bit
version.
Nothing
there's
no
changes.
If
they
are
in
the
alpha
version,
they.
C
J
We
also
provide
two
options
like
dry
run,
which
makes
actually
the
same
with
the
command,
but
it
does
not
update
the
web
configuration
in
the
git
repository
just
prints
out,
so
the
user
can
check
if
the
migration
is
the
migrated.
Webhook
configuration
is
looking
as
it
should
be,
and
also
the
project
option
where
the
user
can
specify
a
project
name
for
which
can
he
migrate,
for
which
he
wants
to
migrate,
wepo
configuration.
So
in
this
case
only
the
web
configurations
for
the
specific
projects
are
migrated.
J
J
And
when
retrieving
a
non-existing
project,
the
user
got
500
error.
This
was
caused
due
to
the
not
ideal
normalization
of
the
error
messages
as
database
was
responding
with
different
error
messages,
as
the
handler
was
expecting
it.
So
the
solution
was
actually
normalize.
The
error
messages
to
put
it
in
one
place
not
have
duplicates,
and
the
database
and
also
the
handler
studios,
are
now
using
only
one
single
source
of
truth
of
error
messages,
and
now
everything
works
as
expected.
J
The
second
bug
was
regarding
retrieving
the
in
the
resource
service
when
retrieving
a
single
resource.
This
endpoints
basically
didn't
work
when
a
user
tried
to
retrieve
a
resource
which
contained
slashes
in
its
resource
urine,
so
we
needed
to
just
escape
these
slashes,
so
it's
able
to
read
otherwise.
J
The
user
always
got
a
404,
so
resource
not
found,
and
some
chore
works
from
my
site
about
improving
the
nuts
configuration,
also
bumping
the
nuts
release
to
from
0.13.2
to
0.70.2,
and
also
adding
some
additional
parameters
setting
and
setting
some
things,
just
a
chore
work,
and
regarding
this
also,
we
baked
the
home
dependencies
directly
to
the
repository
charts.