►
From YouTube: Keptn Community & Developer Meeting - Jan 27, 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
Okay,
so
here
we
go
thanks
for
joining
the
captain
meeting
today
is
january
27th,
so
we
have
quite
a
lot
of
contributors
on
the
call.
So
thanks
everyone
for
joining
and
let's
take
a
look
at
the
today's
agenda,.
A
I
guess
I
just
kicked
it
off
today.
I
don't
have
many
items
from
myself,
so
just
quick
update
next
wednesday.
We
will
have
another
epac
emea
meeting.
So
currently
we
try
to
focus
on
user
experiences,
so
we
will
have
one
demo
for
integration
and
then
brett
mcqueen
will
also
talk
about
his
captaining
experiences,
so
this
tentative
agenda
and
please
feel
free
to
contribute,
and
what
do
you
have
today?
So
we
have
issues
and
contributions.
C
B
A
Well,
one
of
options
is
that
during
a
meeting
in
these
times
on
once
per
two
weeks
and
once
per
two
weeks
in
epact
time
zone,
so
basically
changing
every
other
week,
wouldn't
it
make
sense
for
you,
or
would
you
like
to
have
a
separate
meeting
on
a
weekly
basis
in
the
impact
time
zone.
B
A
This
thursday,
generally
finding
with
everyone,
would
you
also
like
to
have
another
day.
A
C
You
don't
yeah
great,
so
hi
everyone
from
my
agenda.
Actually,
there
are
two
main
topics
I
would
start
with
the
second
one.
The
second
one
is:
I
need
to
the
integration
test
developer
documentation.
C
Actually
we
didn't.
We
had
integration
test
developer
documentation
before,
but
actually
it
was
outdated.
So
we
created
a
new
one
which
describes
the
the
structure
of
the
integration
tests,
also
adding
how
to
add
a
new
integration
test,
running
integration,
tests
remotely
or
locally
it
can
it's
described
in
the
captain
docs,
and
also
how
to
set
up
your
local
environment
to
run
the
integration
test
locally.
C
Okay,
3d
is
more
recommended
from
our
site,
which
uses
the
k3d
version
5.,
and
therefore
you
can
run
the
installation
on
the
integration
installation
of
the
integration
tests
and
run
them
locally.
So
you
can
see
after
your
changes,
the
results
of
the
integration
test,
yeah
and
the
second
topic
is
introducing
a
new
waiting
state
in
the
sequence
endpoint.
C
I
have
a
quick
demo
for
this
currently
till
now,
when
one
sequence
was
blocked
by
another.
It's
sustained
in
the
trigger
state,
but
now
we
have
introduced
in
a
waiting
state.
So
when
one
second,
when
the
sequence
is
blocked,.
E
C
C
F
Andrei
could
you
reopen
the
bridge
please.
F
My
question
is
on
the
right
side:
when
you
scroll
up
is
the
state
of
the
sequence
also
reflected
in
the
top
line
at
the
top
and
the
header
delivery
succeeded?
Is
then,
when
the
when
delivery
is
in
a
waiting
state,
is
then
there
the
message?
Delivery
evading.
C
I
have
was
clicked
on
another
sequence,
so
therefore
it
was
misleading.
Maybe.
F
Okay,
yeah
just
wanted
to
ask
whether
the
label
on
on
the
the
headline
or
on
the
top
is
also
adjusting
cool.
B
B
So
now
there
is
a
message
that
saying
that,
if
you
are
creating,
you
have
created
a
project
without
an
upstream
then,
instead
of
presenting
the
information,
icon
will
present
a
warning
icon
and
next
to
that,
a
message
telling
you
that
soon
captain
will
only
work
with
upstreams.
So
please
add
the
upstream
to
that.
B
B
For
instance,
in
this
screenshot
from
a
mock
of
the
ui,
you
can
below
your
project
name.
You
can
configure
an
upstream
repository,
both
the
https
or
ssh.
Here
the
example
for
https,
where
you
can
present
the
standard
data
from
the
git
url,
a
user.
In
the
token,
you
can
also
then
specify
a
specific
certificate
if
you
use
any
soft
sign
authorization,
or
you
can
also
attach
it
as
a
text
code.
B
G
B
B
B
B
A
In
the
future,
it
would
be
nice
to
know
how
secrets
would
interact
with
cap
60.
I
mean
airbag
model,
so
whether
there
is
a
plan
to
somehow
restrict
secrets
access
within
the
instance
or
whether
they
would
be
shared
yeah.
I
guess
it's
something
to
cover
in
cap
60
now
yeah
but
yeah.
This
is
the
topic
we
will
need
to
discuss.
J
I'm
going
to
present
one
full
request:
I've
been
working
on
showing
the
an
error,
a
proper
error
message
when
having
problems
parsing
a
cheaper
example
and
in
particular
the
problem
here,
was
that
when
editing
or
adding
a
subscription,
we
load
the
shipyard
drama
to
parse
the
tasks
for
a
task
drop
down
for
the
drop
down
there
and
yeah.
J
If,
for
some
reason,
the
shipyard
demo
is
malformed-
and
this
might
happen
through
a
yeah
change
on
the
git
repository
yeah
previously,
the
page
was
stuck
here
and
yeah
that
is
now
fixed,
so
that
with
cheaper
demo,
that's
not
possible.
The
user
gets
a
message
could
not
pass
ship,
your
tml
and
yeah.
That
page
is
not
broken
and,
of
course,
you
can
retry
or
jump
back
to
the
uniform
view.
J
Yeah.
I'm
here
still
struggling
with
some
ui
tests
because
yeah
they
are
very
flaky
and
yeah
act
differently
on
github
than
locally
but
yeah.
That's
actually,
all
from
my.
B
Site,
I
have
a
question
on
that
in
the
screenshot
that
you
show
us,
you
have
two
times
there
or
one
on
the
top
one
on
the
bottom.
J
H
Yet
we
need
to
improve
the
error
messages
here,
because
we
are
intercepting
all
errors
and
showing
them,
and
then
we
have
different
screens
where
we
also
show
a
different
message.
G
Browser
are
you
using
chrome,
but
it's
on
mac
os
and
I
need
to
give
special
permissions
and
stuff
and.
G
Yeah,
so
I'm
just
gonna:
do
it
like
that
and
we
can
show
the
pull
requests,
maybe
a
little
bit
so
the
first
big
thing
that
I
worked
on
this
sprint
was
basically
switching
our
releases
of
of
the
health
charts
of
captain
away
from
having
a
gcp
bucket,
a
storage
bucket
that
serves
that
content,
and
I
switched
it
out
to
basically
have
two
separate
github
repositories
that
make
use
of
github
pages
to
basically
serve
our
our
home
charts
in
the
future.
G
As
basically
it's
kind
of
the
same
as
before.
Only
that
the
url
changed
you
can
see
it
in
the
help
chats
repository
in
the
readme.
Actually,
it's
now
just
charge.captain.h.
G
You
will
get
a
like
a
small
readme
welcome
page
there.
But
if
you
go
to
basically
any
any
sub
domain
there
to
the
github
pages
branch,
then
you
should
see
all
of
our
charts
there.
But
you
need
to
switch
to
the
github
pages
gh
pages
branch
they
round
in
there.
G
And
then
some
yeah
in
the
packages
folder
and
there's,
basically
all
our
charts
as
before,
just
that
they're
now
served
through
github
pages
and
this
repository
for
this
release
and
the
upcoming
release.
G
We
will
actually
support
both
hamchat
repositories,
so
so
the
new
the
new
release,
the
upcoming
one,
will
also
still
be
on
on
the
previous
charts
repo,
but
in
the
future
this
will
fully
be
switched
to
to
the
gita
pages
once
and
then
you
can
always
access
the
hand
repository
with
charge.captain.sh.
Basically,
a
second
thing
that
I
worked
on
was
some
verification
security
feature.
G
Basically,
I
added
a
feature
to
our
release
pipeline
to
also
generate
docker
image
digests
for
our
new
releases
and
those
will
actually
be
checked
against
our
online
docker
hub
images
nightly,
so
that
we
can
always
be
sure
that
that
our
images
are
still
exactly
the
ones
that
we
released.
Nothing
happened,
nobody
tampered
with
them
in
any.
C
G
So,
from
now
on
on
every
release,
there
will
be
this
digest.csv
file,
which
just
contains
all
our
artifacts
and
the
docker
sha-sums,
so
that
you
can
actually
verify
against
the
docker
images
you
download
yeah,
so
that
you
can
be
sure
that
you
have
the
right
images
and
everything
is
fine
with
them.
G
In
addition
to
that,
I
also
fixed
some
things
in
our
integration
test
pipeline.
We
have
basically
bot
build
branches.
G
So
basically,
I
fixed
that
by
by
filtering
out
those
runs
from
from
bots
and
only
taking
into
account
the
ones
that
were
actually
triggered
by
human
users.
Basically,
so
this
would
make
this
will
make
our
integration
test
pipeline
a
little
more
stable
again.
G
And
then
one
last
thing
where
the
pr
is
actually
already
open,
but
not
yet
reviewed
or
merged,
so
I'm
probably
gonna
present
it
again.
Next
sprint
review,
I'm
currently
or
I
just
finished
working
on
having
static
gke
clusters.
G
In
favor
of
recreating
them
from
scratch
every
time
on
every
run,
basically,
we
have
two
static
clusters
now,
and
every
integration
test
run
happens
in
a
separate,
unique
namespace
that
has
the
github
actions
run
number
appended
to
it,
so
that
it
is
unique,
and
this
this
should
actually
speed
up.
Our
our
integration
test
runs
because
you
don't
need
to
set
up
clusters
every
time
from
scratch.
G
Yeah
and
also
we
need
a
much
more
restricted,
gcp
token
now
on
github,
so
this
token
has
far
less
far
less
privileges
than
the
the
one
we
had
before
yeah
and
that's
already
it
for
me.
I
will
present
this
this
pr
in
a
little
more
detail
next
time
when
it's
merged
and
revealed
and
everything
so
yeah
heading
over
to
class,
then.
H
So
I
worked
on
the
the
new
notifications
because
previously
we
had
about
three
or
four
different
ways
on
how
to
display
a
notification,
and
now
we
have
one
to
rule
them
all
and
yeah,
just
like
for
the
notifications
that
a
new
version
is
available.
There's
now
a
sticky
one
and,
for
example,
if
a
creator
project
there's
another
one
which
it
has
a
timeout
so
for
the
great
project,
it's
about
10
seconds,
I
guess,
and
then
it
fades
out
and
yeah.
H
This
should
be
after
a
couple
of
time:
yeah
then
it
fades
out-
and
this
is
gone
just
to
mention-
if
you
hover
over
it
again,
it
is
with
opacity
to
full
opacity
and
the
timeout
will
start
again
and
yeah.
That's
basically
it
notification
and
each
of
the
error
that
could
happen
on
a
http
request
is
now
shown
with
this
notification
instead
of
something
else
with
the
toast
on
the
bottom
or
a
message
embedded
in
some
view.
So
it's
just
this
one
and
yeah
that's!
Basically
it
are
any
questions.
C
K
C
K
What's
noteworthy
here
is
to
not
break
anything.
The
default
behavior
if
this
is
not
set,
will
be
such
that
the
started
event
will
be
indeed
sent
by
default.
So
only
is
if
this
is
really
specified
explicitly
to
false,
then
the
start
at
the
event
will
not
be
sent.
So
this
way
we
won't
break
any
existing
web
pro
files
that
have
been
configured
with
previous
versions.
K
All
right,
if
not,
let's
continue
with
the
next
one,
so
this
is
about
the
shipyard
controller
and
its
interaction
with
the
resource
service.
K
The
controller
will
now
check
for
incoming
sequence.triggered
events
and
if
the
git
commit
id
is
specified
in
this
event,
it
will
take
that
and
propagate
it
throughout
the
sequence
via
the
git
commit
id
property
in
the
cloud
events
which
lives
next
to
other
properties,
such
as
the
context,
study
or
the
trigger
dd,
etc.
K
So
services
will
be
able
to
retrieve
it
that
way.
So
if
the
sequence,
the
triggered
event
contains
that
eddy,
it
will
propagate
that
one
and
otherwise,
if
it's
not
there,
it
will
fetch
the
shipyard
file.
K
B
K
Yes,
ana
did
research
on
that
anna.
Do
you
have
the
link
to
the.
E
K
So
it
will
see
what
the
last
commit
was
and
then,
together
with
the
resource,
it
will
return
this
commit
id,
and
then
this
id
can
be
used
throughout
this
particular
stage
and
then,
for
example,
exactly
yeah.
And
then,
if
another
sequence
is
triggered
in
the
next
stage,
then
it
will
do
again
for
this
particular
stage.
E
E
Hon
right,
if
there
is
a
transition
from
one
stage
to
the
next
stage
banner,
you
can
reuse
the
git
commit
or
not.
K
For
now
we
cannot
reuse
it
because
right
now
the
stages
are
stored
as
individual
branches,
with
different
histories
in
the
git
tree.
So
now
for
every
new
stage
of
a
sequence.
We
need
to
do
that
again,
but,
as
you
might
know,
we
are
working
on
the
new
kit
model
where
we.
C
K
You
all
right,
then,
the
next
one.
This
was
a
little
fix
for
the
distributor.
So
let's
just
have
a
look
at
the
issue,
so
basically,
unfortunately,
the
cluster
is
not
there.
C
K
But
what
happened
there
was
that,
whenever
a
distributor
for
a
execution
plane
service
was
registering
itself
at
the
uniform
api,
it
also
has
the
possibility
to
communicate
the
version
of
the
distributor
that
is
being
used,
and
apparently
we
didn't
have
that
set
properly
in
the
helm
chart.
So
this
was
just
a
matter
of
required
distributor
version
environment
variable
in
the
helm
charts.
So
now
this
is
being
set
to
the
distributor
image
tag.
K
So
recently
we
had
some
permission
issues
on
certain
platforms
where
the
configuration
service
couldn't
properly
use
or
set
the
global
git
configuration,
and
that's
why
we
have
introduced
a
change
in
the
last
captain
version
where
the
git
config
is
set
individually
per
project
within
the
project
directories,
meaning
that
the
user
and
the
git
email
which
are
required
are
set
there.
But
then
we
had
a
backwards
compatibility
problem
where
in
projects
that
have
been
created
with
older
captain
versions
and
were
still
the
global
conflict
was
used.
K
For
a
stage
before
a
commit
is
staged
and
pushed
the
git
user
is
configured
properly
so
now.
Also,
projects
created
with
previous
captain
versions
will
then
continue
to
work
as,
as
usual,
any
questions
about
that.
K
All
right,
then,
the
next
one
is
really
small
one.
So
this
was
also
for
the
configuration
service.
So
here
we
had
the
bug
that
if
upstream
repository
was
updated
for
for
a
project
and
that
upstream
url
was
not
found,
we
got
a
500
error
and
yeah,
basically
that
all
that
needed
to
be
done
here
was
to
properly
set
the
response
code.
So,
instead
of
a
500,
we
now
return
a
404
in
case
the
upstream
repository
is
not
found
and
then
finally
something
useful
for
the
cli.
K
K
All
you
need
to
do
is
set
the
http
proxy
or
https
proxy
environment
variable
and
then
the
cli
will
route
all
outgoing
api
requests
through
this
proxy
that
is
specified
here.
So,
as
you
can
see,
you
can
either
set
an
ip
or
a
url,
or
if
you
have
some
sort
of
basic
http
authentication,
then
you
can
provide
the
url
in
this
format
and
then
the
cli
will
work
with
that
proxy.
F
Not
a
question
but
flow,
then
the
documentation
update
you
showed
us
is
for
captain
or
12
right.
K
F
No,
I
get
it,
there
were
no
changes
required
on
the
cli
side,
as
this
is
supported
out
of
the
box.
Am
I
right.
B
B
So
the
sso
client
id,
which
is
the
auth
client
id
basically
and
then
the
os
sso
discovery
url,
which
is
the
one
used
for
the
open
idc
to
know
all
the
different
url
that
are
needed
in
order
to
start
to
the
authorization,
authentication
flow,
sorry
and
yeah.
This
was
his
first
pr
any
question
for
this.
B
B
So
you
don't
you
don't
need
to
provide
it
as
a
part
of
the
ui,
but
the
authentication.
The
sso
also
then
grants
you
the
rights
to
automatically
pass
through.
So
dci
does
not
complain
anymore.
If
you
don't
specified
api
token,
but
you
need
to
make
sure
that
you
configure
your
system
in
order
to
inject
this
api
token
automatically
after
the
sso
authenticates
your
users,
we
will
follow
up
with
some
more
documentation
on
this.
How
to
prepare
such
an
environment.
B
B
F
All
like
had
to
had
to
leave
and
it's
the
reason
why
I'm
taking
it
over
and
unwrapping
it
up,
yeah
bring
me
the
question
to
the
audience.
Are
there
additional
questions
feedback
you
want
to
share
in
this
meeting.
K
Because
I-
and
I
forgot
about
it
when
I
was
presenting
so
anna
was
kind
enough
to
put
the
link
to
the
results
of
the
performance
evolution
for
the
configuration.
So
it's
in
that
chat.
F
Nice,
okay,
great,
what
you
can
see
here
is
actually
the
improvements
on
a
performance
level,
and
here
we
are
talking
about
1.9
seconds.
F
Oh
yeah,
the
old
one
and
the
new
one
200
milliseconds.
L
That's
the
new
one,
retrieving
always
with
the
pool
and
lower
there
is
with
the
git
commit
like
the
last
result.
It's
in
in
nanoseconds
yeah.
D
F
Wow
yeah,
that's
impressive
cool
thanks
for
sharing.
Let
me
also
add
the
link
here.
F
F
F
F
Okay,
yeah,
please
participate
in
this
doodle
to
also
represent
your
voice,
and
if
there
are
no
further
questions.
F
No
not
the
case,
then
I
will
close
the
meeting
for
today
thanks
for
joining.
If
you
want,
please
also
join
the
apec
email
meeting
next
wednesday
or
let's
come
together
next
thursday.
Then
have
a
nice
week
and
see
you
bye.