►
From YouTube: Keptn Developer Meeting - May 05, 2022
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
So
hi
everyone
welcome
to
the
captain
developers.
Meeting
today
is
may
5th,
so
just
in
case
may
the
fourth
be
with
you
and
I
hope
everything
is
going
fine.
A
A
A
Okay,
so
what
do
we
have
on
our
agenda?
We
have
a
few
presentations
today
and
I
guess
that's
it
if
you
would
like
to
add.
C
C
Okay,
I
have
two
things
on
my
agenda.
First
of
all,
a
quick
announcement.
So
yesterday
the
team
has
released
job
executive
service
version,
0.2.0,
here's
a
link
to
the
release
page
as
well.
Please
go
through
the
release,
notes
if
you're
interested,
we
did
a
lot
of
security
hardening
and
therefore
we
have
some
breaking
changes.
In
case
you
were
using
the
enable
kubernetes
api
access
flag
before
so
use.
Cases
like
helm
deployment
or
cube
ctl
access
within
job
executive
service
are
now
a
little
bit
different.
C
There
is
now
also
a
dedicated
service
account
field
which
you
can
use
to
define
a
service
account
for
accessing
certain
namespaces
accessing
ports
deployments
whatever
you
need,
so
we're
a
lot
more
customizable,
but
we're
also
a
lot
stricter
with
security
here
with
that
release.
If
you
have
any
questions
there
is
the
help
integrations
channel
on
the
captain's
lag.
Please
come
here.
C
I
and
some
members
of
our
team
are
watching
that
channel
and
we're
happy
to
help
you
and
the
second
one
is
more
of
a
discussion
item.
It
is
about,
should
we
or
can
we
move
integration
issues
that
we
have
in
the
captain?
Captain
repository
to
a
separate,
a
dedicated
repository,
so
we
do
now
have
like,
in
the
captain
captain
repository
a
lot
of
github
issues
that
are
marked
with
integrations.
C
Those
are
ideas,
sometimes
they're
requests
coming
from
different
people
and
yeah.
I
mean
it's
like
26
issues
that
are
nice
to
have,
but
that
are
polluting.
The
captain
captain
issue
tracker
a
little
bit
and
yeah
as
a
cleanup.
I
would
suggest
that
we
move
them
into
a
different
repository,
like
captain
slash,
integrations
or
even
captain
contrib
slash
integrations
or
if
you
have
any
other
idea,
please
speak
up
now,
yeah.
So
what
I.
A
A
What
we
could
do
is
probably
creating
a
new
repository
proxy
for
captain
can
trip
and
captain
sandbox,
because
our
main
issue
that
we
cannot
connect
issues
located
in
other
github
organizations,
neither
connect
to
the
roadmap,
no
connect
to
the
rest
of
the
system.
A
Github
is
still
pretty
bad
about
it,
but
if
you
have
another
repository
and
if
you
have
just
one
pinned
issue
for
integrations,
so
if
you
wonder
about
integrations,
please
go
to
this
repository,
I
can
also
add
metadata
to
a
new
screen
so
that
when
you
open
it,
if
you
want
to
talk
about
integration,
it
can
send
you
over
to
another
repository,
and
maybe
it
will
provide
good
enough
experience.
C
Okay,
so
another
rapper
is
fine,
I'm
happy
if
putting
it
into
the
captain
organization,
because
that's
like
the
head
organization,
anyway,
just
captain
contrib
and
sandbox
in
the
end
should
be
I
I
would
say
more
source
code
focused.
The
captain
organization
also
contains
the
community,
contains
docks,
etc.
So
I
think
there
is
a
good
place
and
then
at
least
you
should
also
be
able
to
connect
it
with
your
roadmap,
because
I
think
the
roadmap
is
in
the
captain
organization
as
well
yeah.
So
there
are
some
quirks
loving
to
connect.
A
C
Your
son
agrees.
Okay,
my
phone
has
a
post
bus.
Okay,
I
will,
I
will
see
I
will
contact
you
offline
and
afterwards
for
the
repository
and
for
moving
the
issues
yeah.
Well,
I
can
actually
do
this.
Synchronously.
A
C
Yeah
is
suraj
here
I
think
he
mentioned
in
the
captain's
like
that
he
will
not
be
able
to
join,
but
he
mentioned
that
he
has
like
the
pull
request.
I
think
that's
also
what
it
says
on
the
agenda
here
in
in
the
document
that
he
still
needs
somebody
to
review
his
his
pull
request.
Please.
So
that's
what
it
says
here:
yeah!
That's
about
the.
C
Yeah
so
I
think
then
next
one
would
be
moritz.
Yes
welcome.
Back
olek,
we
just
continued
with
suraj
and
we're
now
up
to
moritz.
C
A
F
A
So
yeah
I
will
handle
it
probably
today
for
the
lunch.
E
Okay,
I
have
two
changes
to
present
today.
The
first
one
is
about
our
help:
charts,
I'm
kind
of
in
the
process
of
of
refactoring
them.
This
is
just
the
first
step,
there's
actually
a
few
prs
following
this
in
the
next
sprints.
E
Probably
so
this
first
one
is
to
create
a
common
chart
for
duplicated
helper
functions
that
we
have
in
all
three
charts
that
we
maintain
right
now,
so
the
main
captain,
installer
and
also
ham
service
engine
meter
service,
there's
lots
of
helper
functions
that
that
are
just
duplicated,
and
for
that
basically,
I
created
a
common
chart
next
to
the
captain,
installer
so
yeah.
This
is
of
type
library,
of
course,
and
it-
and
it
really
only
has
helper
functions
that
we
can
use
in
all
the
different
charts.
E
So
this
this
pr
is
actually
just
introducing
on
the
chart
and
then
there's
gonna
be
a
follow
up
pr
which
actually
makes
use
of
it
to
make
to
make
it
easier
to
review
after
the
fact
so
yeah,
that's
the
first
one,
and
then
I
have
one
more.
I
made
a
small
fix
for
the
configuration
service
actually
for
the
installer
of
the
configuration
service.
E
We
we
had
a
regression
there
where
we
updated
the
the
update
strategy,
the
kuret's
update
strategy
to
rolling
update
for
the
config
service,
even
though
that's
actually
not
supported
by
it,
because
it
has
a
volume
which
is
read
right
once
so.
You
can't
really
join
create
a
second
part
of
the
conflict
service
because
you
can't
attach
the
volume
at
the
same
time.
E
So
I
just
changed
that
back
to
recreate
strategy,
and
everything
should
should
work
as
expected
with
that
now
yeah
and
then
the
the
resource
service,
that's
that's
going
to
be
fully
released
in
the
future,
is
still
on
update
strategy,
rolling
up
that
since
it
works
there.
E
So
yeah,
that's
already
it
for
me,
so
I'm
gonna
hand
over
to
flo.
If
that
works
now,.
D
Okay,
let's
have
a
look
so
yeah
three
items
for
today
for
the
first
one,
I'm
not
gonna
go
into
too
much
details.
Basically,
some
adjustment
needed
for
the
integration
tests
to
make
them
run
again.
D
That's
due
to
the
change
to
the
upgrade
strategies
that
moetz
also
briefly
talked
about
certain
things
like
restarting
etc.
Like
now,
a
little
bit
longer
and
there
we
had
some
basically
some
timeout
issues
in
some
of
the
tests,
and
this
has
been.
B
D
But
we
need
to
keep
an
eye
on
that
then,
let's
start
with
the
first
more
interesting
issue.
So
this
is
about
the
shipyard
controller.
So
when
we
have
sequence
being
executed
the
way
the
data.
B
D
If
it's
a
multi-stage
sequence,
the
data
from
previous
stages
is
also
propagated
into
the
next
stage
and
due
to
the
new
model
and
logic
for
the
shipyard
controller,
there
were
some
slight
changes
in
the
merchant
logic
which
yeah
messed
up.
D
Some
of
the
use
case
like,
for
example,
the
yeah
the
delivery
use
case,
because
the
order
of
your
eyes
that
were
reported
back
by
the
helm
service
was
different,
so
yeah,
thankfully
our
christian
has
detected
those
and
now
their
merge
logic
has
been
adapted
to
it
has
been
before,
but
yeah,
maybe
in
the
future.
We
should
maybe
also
consider
reworking
this
whole
concept,
because
right
now,
for
example,
not
clear
which
so
data
example
in
the
deployment
uri.
Since
that's
only
an
array,
that's
filled
across
the
stages
of
a
delivery
sequence.
D
For
example.
It's
not
exactly
clear
which
data
comes
from,
which
stage,
for
example,
so
hopefully
in
the
future,
we
can
maybe
rethink
this,
maybe
come
up
with
something
better
to
kind
of
have
a
clearer
separation
of
where
the
data
in
the
finished
events
and
then
the
triggered
events
comes
from
yeah.
So
this
fix
is
part
of
the
recently
released.
D
0142
hotfix
release
and
will,
of
course
be
included
in
zero
fifteen
next
one,
yes,
also
about
the
shipyard
controller.
D
Give
you
a
little
bit
more
background
about
that
is
in
some
or
a
couple
of
versions
ago.
We
introduced,
we
used
a
mongodb
that
is
deployed
per
default
together
with
the
with
captain
core,
but
also
provide
a
connection
string
to
any
external
mongodb
that
you
might
want
to
connect
to,
and
one
of
those
will
also
be
amazon
documents
db.
B
D
D
It
might
be
that
that
we
had
those
because,
for
example,
it
also
stores
our
sequence
execution,
the
yeah
certain
data
that
was
sent
back
by
the
captain
integration
that
took
over
a
task
and
within
the
finished
events,
and
by
that
task
it
might
be
that
there
was
a
property
of
without
in
the
name.
And
so
basically,
what
has
been
done
in
the
spur
request
here.
H
H
Could
you
stop
sharing
your
screen
thanks
so
much
okay
for
today,
just
or
a
rather
big
topic,
more
or
less
for
the
bridge?
H
So,
for
example,
if
there
are
some
data
points
or
some
dates
that
are
duplicated,
it
adds
a
number
to
it
with
the
yeah.
The
duplicate
count,
for
example,
additionally,
is
supported
with
expanding
and
collapsing.
For
example,
if
there
are
more
than
10
sls,
then
you
can
then
show
or
sli
button
shows
up
and
yeah.
You
can
expand
and
collapse
this
and
next
year
also
support
the
case.
H
If
there
is
a
sri
with
yeah
with
a
length
over
yeah,
the
specified
one
also,
this
is
set
to
ellipsis
style
next
to
that
yeah
also,
the
selection,
so
the
primary
selection
and
the
secondary
selection
is
also
in
place.
H
Also,
if
you
click
on
an
empty
tile,
for
example,
if
there
are
no
results
for
the
slis
just
for
specific
ones,
then
this
is
empty,
but
you
can
still
click
and
select
them
next
to
that
the
filter
that
was
not
working
in
hatchers,
but
this
is
working
now.
So
if
you
press
on
this,
the
legend
below
the
tiles
are
disabled
and
also
the
tooltip
is
disabled
here
regarding
the
tooltip
yeah,
as
the
tooltip
is
supported,
also
for
two
cases
for
the
sls
and
for
the
score,
a
different
one.
H
Next
to
that
also
a
resize,
so,
for
example,
if
you
resize
the
page
also,
the
heat
map
is
adjusted
accordingly
and
also,
if
you
look
at
the
x-axis,
you
can
see
that
if
a
certain
point
is
reached,
the
yeah,
every
second
element
is
shown
instead
of
or
and
yeah
also
supports,
zooming
zooming
in
and
out
that
also
the
tooltip
is
adjusted
accordingly
and
also
if
it's
cruel,
the
tooltip
is
adjusted.
C
H
A
Yeah
thanks
just
to
understand
duplo
moving
out
of
high
charts
completely.
I
H
A
And
basically
yeah,
I'm
just
looking
from
the
community
standpoint,
so
yeah,
so
high
charts
is
a
huge
library,
so
for
us,
maybe
just
having
our
own
hit
map
implementation
is
rather
reasonable.
J
So
you
should
be
receiving
good
yeah.
So
in
the
last
weeks
I
concentrated
on
two
things:
the
first
one
was
introducing
a
new
captain,
the
book
conf
comfy
config
mac
with
the
user
defined
or
user
editable
deny
list.
J
The
point
is
that
until
now,
with
the
alpha
standards
there,
we
actually
had
some
list
of
hardcoded
urls,
which
couldn't
be
used
as
part
of
the
qr
curl
commands,
but
in
beta
version
this
list
isn't
hard
coded
anymore
and
it's
user
configurable,
it's
user
configurable
via
the
config
map,
which
can
look
like
this.
I
will
show
an
example.
J
Where
we
have
a
dna
list
with
the
editable
con
with
list
and
the
user
can
define
which
config
maps
we
want
to
deny
and
allow
yeah.
The
second
issue
was
actually
to
configuring.
The
termination
grace
period
pre-stop
hooks
and
upgrade
strategy
in
our
helm
chance.
This
is
actually
leading
to
zero
down
time.
J
This
direction.
We
are
heading
for
a
longer
time.
B
J
So
yeah,
it's
kind
of
all
of
the
values
for
the
pre-stop
hooks.
The
time-outs,
also
for
the
grace
grace
period
are,
are
configurable
via
helm
charts.
We
are
still
kind
of
testing
the
default
values,
but
it's
everything
should
be
configurable
via
home
charts
and
therefore
the
users
should
be
able
to
manage
it
themselves.
J
Yeah,
that's
everything
from
my
side.
Are
there
any
questions.
I
B
B
This
is
mostly
a
copy
paste
and
done
thanks
to
moritz
pipeline
on
integration
test.
The
only
big
difference
that
is
there
is
basically
the
pipeline
takes
two
arguments,
one
for
the
installation
of
the
captain
version
and
one
for
deciding
which
version
to
upgrade
and
plus.
B
It
just
run,
go
test
sum
and
basically
run
the
zero
down
time
suite,
which
is
done
in
the
next
step.
B
How
do
the
tests
look
like?
Basically,
there
is
a
main
task
suit,
which
is
provided
with
a
template
to
do
parallel
testing.
At
the
same
time,
you
will
run
a
helm
upgrade
with
the
past
environment
variables
with
the
two
captain
environment
and
there
will
be
a
suit
of
api
calls
that
ping,
the
captain
api,
and
while
these
two
are
running
at
the
same
time,
a
function
of
your
choosing
will
be
could
be
run
as
well.
B
In
the
in
our
case,
we
have
two
function
implemented,
one
is
for
the
sequences
and
one
is
for
the
web
book
for
the
sequences
test.
We
do
have
a
basic
evaluation
that
should
pass
with
hundred
delivery.
That
should
fail
because
no
no
found
task
sequence
and
then
evaluation
that
should
fail
because
of
the
missing
time
frame.
So
this
is
lighthouse
and
cheaper
controller,
mostly
and
for
the
web
book
we
are
recycling
the
web
book
basic
test.
B
So
we
create
a
web
book
projects
and
we
run
multiple
tests
with
it,
where
we
have
one
pass
with
the
task
and
the
other
seven
to
fail,
and
this,
as
you
can
see
in
both
of
them,
how
it
is
working,
is
basically,
we
are
looping
until
we
quit
from
the
from
the
rolling
upgrade,
so
the
rolling
upgrade
are
finished
and
we
keep
on
looping
on
this
test
for
the
api,
we
are
using
an
api
test
suit,
which
is
allowing
us
to
nicely
format
everything
I
will
show
you
from
here.
B
So
basically,
this
is
how
the
pipeline
look
like
it's
very
similar
to
the
integration
test.
There
is
the
row
format
of
the
test,
and
then
there
is
a
3d
format
where
you
can
see
all
of
the
steps
here.
For
instance,
we
have
both
the
sequences
and
the
web
book
and
we
have
a
summary
for
each
of
the
tests
where
we
have
the
count
of
the
failures
and
successes
in
sequences
and
api
probes.
B
You
can
open
each
of
the
tests
and
check
whether
or
not
they
passed.
This
is
nice
for
the
api
test,
because
then
you
can
see
the
some
of
the
data
we
could
later
on
use
like
the
duration
of
the
test
and
what
was
the
get
method
and
what
was
passed
and
received,
of
course,
filtering
out
information
like
the
token
and
whatnot
and
yeah.
You
can
clearly
see
if
some
of
the
tests
fails
and,
for
instance,
here
one
of
the
web
calls
fails
during
one
of
the
upgrade,
and
we
will
see
that
yeah.
B
B
Yeah,
I
think
that's
it.
Maybe
I
will
show
you
how
to
run
it.
What
are
the
parameters
you
can
use
so
basically
for
the
workload
flow.
For
now
this
is
not
completely
merged,
so
I
will
have
to
run
from
my
branch.
B
This
is
from
what
branch
you
start
with
the
first
installation
of
captain,
and
here
you
can
pass
what
helm
chart
you
want
to
use,
so
this
can
be
either
in
the
captain
or
in
the
captain,
dev
repo.
I
F
F
G
A
For
16.,
yes
yep,
so
yeah
thanks
everyone
and
again
just
to
follow
up
on
what
we
have
discussed
in
the
recent
weeks.
Yeah
on
my
site,
I
was
going
through
backlog,
et
cetera,
for
integrations
ideas.
We
had
on
the
backlog
tickets
and
actually
I
want
to
do
the
same
for
web
ui.