►
From YouTube: Keptn Community Meeting - April 13, 2022
Description
Discussion of Keptn 0.13 to 0.14 upgradeability issues, making Helm charts more accessible to users, and Google Summer of Code Q&A.
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
A
So
what
we
have
on
our
list
is
basically
migrating
from
013
to
014
and
potential
compatibility
issues
and
making
home
charts
more
accessible
to
end
users,
and
we
will
use
whatever
time
remaining
for
gsoc
on
a
so
thanks
brad
for
these
topics
and
yeah.
Actually
we
can
start
from
migrating
because
this
something
many
captain
users
hid
already.
A
And
what
was
actually
surprising
change
is
the
distributor
service,
because
what
we
have
added
upgrade
guidelines
etc,
but
actually
what
we
experienced
before
you
that
yeah,
if
you
use
old
version
of
the
distributor
service
or
if
you
do
not
update,
then
you
have
an
issue
and
many
users
had
regressions
when
I'm
creating,
because
nuts
changed
so
many
integration
services
stopped
working
properly,
just
because
they
were
unable
to
connect
to
captain
and
communicate
to
other
companies.
A
So
many
instances
went
sideways
and
the
problem
that
you
need
to
update
the
distributor
service
and
basically
it
means
you
need
to
update
each
integration
right
now.
So,
at
least
for
me
it
was
a
surprise.
So
when
I
was
upgrading,
I
hit
this
issue
and
I
guess
anyone
who
uses
custom
services
would
also
hit
it
so
brett.
What
is
your
experience?
Did
you
also
hit
it
on
your
side.
B
Well,
yeah,
I
I
was
the
one
I
just
like
held
back
to.
I
came
to
this
meeting
and
asked
because
I
didn't
want
to
waste
time
troubleshooting,
but
if
all
we
have
to
do
is
update
the
distributor
container
image
the
tag,
then
that's
easy.
Yes,
I
believe.
That's
all.
We
need
to
do
right.
A
Yeah,
yes,
this
is
exactly
what
you
need
to
do
or
you
can
apply
work
around
by
changing
global
environment
variable
what
is
definitely
for
retrospective,
so
when
0.14.1
was
released,
even
if
it
was
explicitly
recommended
that
nasa
address
changed
from
either
standpoint
in
great
guidance,
etc,
it
wasn't
clear
what
is
the
potential
fallout
for
that
and
it
came
as
a
surprise
you
when,
on
my
test
instance,
which
is
really
primitive.
A
So
I
think
that
it's
something
we
need
to
work
better
on
basically
upgrade
guidelines
and
the
communication
when
you
have
braking
changes
and
actually
more
tracking
for
this
breaking
changes
to
start
piece.
So
when
we
plan
a
new
release,
etc,
we
should
clearly
see
what
are
the
braking
changes.
It
was
done
in
this
case.
The
initial
draft
included
three
changes,
but
what
was
missing
is
analyzing
impact
on
users
and
on
upgrade
paths.
A
So
this
is
something
we
should
invest
more,
maybe
at
the
explicit
item
or
to
the
checklist
to
verify
upgrade
guidelines
because
well
it
happens
with
any
open
source
project,
and
you
should
be
just
careful
about
that
and
having
additional
quality
git
in
terms
of
our
release
process.
Would
be
important,
I
think,
we're
just
adding
it
for
the
major
releases
but
yeah
this
captain
nuts
thing.
B
Yeah
I
feel
like
sometimes
if
it's
not
kept
in
core,
then
it's
not
really,
I
wouldn't
say
maintained,
but
it's
not
it's
not
updated
as
much
as
it
probably
should
be
in
the
major
releases.
For
some
of
the.
I
understand
that
it's,
like
you,
know
the
sandbox
area
and
things,
but
people
still
use
those
services.
Well,
it's.
A
So
basically
it
would
render
many
captain
instances
just
a
malfunction
after
this
upgrade,
and
it
raises
a
number
of
questions
so.
Firstly,
whether
we
do
upgradeability
testing
at
all
and
whether
we
have
test
automation
for
that
spoiler
alert.
We
don't-
and
this
is
something
we
need
to
talk
to
the
team.
So
yesterday
I
will
be
meeting
with
johannes,
and
this
is
one
of
the
topics
I
actually
wanted
to
bring
up-
that
we
should
do
some
kind
of
upgradeability
testing
easily
automated
one,
if
not
possible.
A
At
least
manual
once,
let's
say
have
a
number
of
captain
instances
and
verified
verified,
upgrade
flow
there
before
we
roll
out
the
preview
release.
A
Yeah
we
could
well
in
this
case
any
basic
realistic
captain
instance
would
have
caught
it,
and
this
is
something
that
concerns
me
because
yeah,
it's
not
like,
we
need
something
sophisticated,
just
whatever
smoke
test
instance
where
we
do
upgrade
stage.
The
new
release
see
what
happens
and
yes
for
that.
We
could
have
put
it
on
public
demand
instance,
but
public
domain
instance
is
shut
down
indefinitely
until
we
have
cap
60
already
so
or
until
we
have
another
way,
but
cup
six
see,
at
least
for
me,
is
a
table
stake.
A
So
without
permission
model.
We
can
post
public
captain
instance
right
now,
and
if
we
cannot
have
that,
then
we
should
have
at
least
internal
capital
instance.
Maybe
private
captain
on
captain
et
cetera,
where
we,
where
maintainers
and
key
contributors,
have
access
to
the
way
we
can
at
least
verify
key
functionality.
B
From
mine
I
was
just
going
to
quickly
mention
as
well.
Do
we
have
any
new
members
that
need
to
introduce
themselves
on
the
screen.
C
Participating
in
gsoc
this
year
so
yeah
I
mean
I
peaked
an
interest
in
this
k6
integration
project,
but
I
have
some
doubts
regarding
the
scope
of
that
project
and
how
we
are
I
mean
how
we
are
going
to
go
ahead
with
that
I
mean
so.
I
saw
a
post
on
k6
community,
where
adam
probably
mentioned
that
we
can
do
this
using
the
job
operator
service.
So
are
we
planning
to
write
a
new
service
for
this
or
using
that.
A
And
here
I
would
say
you
tell
us,
because
the
implementation
really
depends
on
the
use
cases
you
want
to
address,
because
if
you
talk
about
the
really
simple
service,
of
course,
job
executable
service
would
be
fine.
You
just
put
it
cli
there
trigger
it,
and
something
happens
whether
it's
enough
for
use
cases
we
want
to
address.
This
is
something
that
needs
to
be
discussed
with
the
community
and
actually
implication.
Phase
is
a
good
time
to
discuss
it.
A
So,
let's
prepare
with
potential
users,
maybe
with
users
of
gpmeter
integration
service,
because
we
already
have
g
meter,
which
is
well
key.
Six
integration
service
could
be
basically
the
same.
The
question
is
whether
you
need
it
and
actually
discussing
each
visa.
The
user
community
is
something
I
would
recommend.
C
No
not
yet
I
haven't
tried
it
yet,
but
yeah.
I
will
try.
A
A
And
let's
move
on
so
making
fun
charts
are
more
accessible,
so
brett.
Would
you
like
to
summarize.
D
B
Yeah,
sorry,
if
I
when
I,
when
normally
I
get
artifact
hub,
my
helm,
charts
from
artifact
hub,
you
know
when
I
use
rcd
to
deploy
them.
I
want
to
reference
the
the
url
and
the
chart
version
was
only
the
core
helm.
Charts
of
flipping
are
available,
even
if
it's
like
a
github
pages
branch
where
the
helm
indexer
generates
the
manifest
index.young
file
would
be
sufficient,
but
a
lot
of
the
like
argo
servers.
For
example.
B
So
I
feel
like
it's
very
minimal
effort
to
just
just
have
a
github
action
that
will
run
the
helm
indexer.
It
will
upload
it
and
then
it
can
publish
it
to
artifact
type
as
well.
B
Just
a
mechanism
to
be
able
to
to
specify
the
the
helm
chart
url.
So
you
know
if
I
want
to
do
homework,
for
example,
so
it
doesn't
necessarily
have
to
be
artifact
help.
That's
my
preference,
and
I
would
guess
that
a
lot
of
other
people
use
artifact
up
too,
but
it
could
be
as
simple
as
just
having
a
github
pages
branch
and
then
the
yeah.
The
release
would
be
done
for
that
home
chart
on
the
github
releases
and
then
the
github
pages
reference
that
fire
does,
that
make
sense.
A
It
totally
makes
sense,
though
we
might
have
to
have
a
two-stage
deployment,
because
one
of
the
problems
with
the
current
implementation
is
that
you
cannot
just
grant
permissions
for
artifact
hub
to
any
service,
because
we
have
exactly
zero
way
to
have
different
credentials
to
the
system
or
to
somehow
protect
it.
So
how
I
would
see
that
that
there
would
be
two
stage
deployments
so.
Firstly,
for
example,
github
action
or
whatever
uses
github
packages
to
prepare
hand
charged
or
gaining
github
pages,
but
in
local
repository
with
local
github
tokens.
A
Basically,
just
whatever
instance,
that
would
take
all
these
deployments
and
deploy
to
artifact
hub.
So
it
could
be
a
separate
github
action
or
whatever
other
automation
solution,
but
it
could
be
a
separate
pipeline.
B
Yeah
because,
without
effect
up
you
don't
necessarily
need
to
all
you
need
to
do,
is
do
it
once
and
then
you
point
it
to
that
repo
and
then
every
hour
it
will
pull
the
repair
and
grab
the
latest
metadata,
but
anyway
like
even
if
it
was
in
the
in
the
get
reaper
itself.
I
wouldn't
care
like
just
something
better.
D
B
Having
to
you
know,
look
I
I
don't
think
it's
acceptable
to
have
to
clone
the
reproductive
the
folder
in
the
chart.
You
know
a
lot
of
people
want
to
use,
get
ops
and
they
just
it's
just
easier
like
I
don't
want
to
have
to
waste
my
time
doing
all
that.
Otherwise
I
can
just
fork
it
and
just
do
it
myself.
B
Be
a
good
yeah
yeah,
but
it
could
be
a
good.
It
could
be
a
good
little
project
for
a
student
or
yeah.
A
Yeah,
so
it
seems
to
be
quite
straightforward:
yeah
we
already
have
a
lot
of
artifact
automation
in
the
artifact
hub
repository.
So
basically
we
generate
the
entirety
function
based
on
the
metadata,
so
one
of
the
ways
would
be
to
just
when
we
run
these
scripts
just
to
check
how
charts
in
known
places
and
then
to
deploy
them
within
the
same
script,
which
already
has
all
the
permissions
and
access
to
artifact
hub
api.
A
So
I
think
it
would
be
generally
legit.
It's
definitely
a
nice,
let's
say
hackathon
project.
A
And
also,
we
still
have
services
that
don't
have
hound
checks
at
all.
A
A
Really
much
clearer,
it
would
be
a
nice
project,
I'm
not
sure
whether
potential
mentees
would
be
interested
in
hacking
a
bit.
Actually,
it's
a
good
question
how
we
organize
these
projects
because
we
used
to
do
let's
say
online
hackathons,
slash,
hacker,
gardens
and
jenkins,
so
basically,
when
we
team
up
together
and
do
a
kind
of
work
and
learn
project.
A
Have
a
bug
bush,
starting
in
the
beginning
of
may
so,
we've
been
accepted
to
keep
on
backwash
it's
from
may
second
to
my
thirteenth,
so
they
will
be
looking
actually
for
a
lot
of
good
first
issues.
So
if
you
could
at
least
separate
something
so,
for
example,
create
a
standard
flow
for,
let's
say
home,
charged
in
a
service,
so
I
guess
many
services
could
get
a
hand
charged
well
in
a
quite
straightforward
way.
A
A
So
yeah,
if
you
would
like
to
write
up
a
small
developer
guide
for
that,
basically
yeah
convert
the.
B
B
Our
the
otilia
service
and
the
the
test
cube
service
at
the
moment
as
well,
so
we
can
use
that
as
a
as
a
prototype
and
then
present
it
back
and
see
what
folks
think
about
it.
A
B
Yeah
yeah
for
sure
yeah
with
the
documentation
is
that
the
hugo
site
that
you
run
or
it's
something.
A
C
D
Yeah,
don't
worry
about
if
you
don't
want
to
screw
around
with
hugo
and
all
that
I'm
happy
to
do
that
part
of
it,
but
because
then,
okay,
well,
what
I'm
wondering
too,
is
I'm
still
new
here
when
I
read
what
they
said,
what
the
documentation
is
about
the
breaking
changes.
I
still
didn't
quite
understand
how
major
what
how
breaking
they
were
until
something
else
I
mean
is,
is
what
the
documentation
we're
proving
now
adequate
is
just
a
question.
B
I
I
think,
like
based
on
my
feedback
for
the
first
item,
about
that
it
was
a
big
break
and
change,
but
maybe
it
could
have
been,
it
could
have
been,
maybe
portrayed
more,
that
it's
sort
of
minimal
effort
to
actually
fix
it,
because
all
you
really
had
to
do
is
upgrade
the
distributor
container.
B
Probably
it's
just
it's
just
what
your
service
talks
to,
so
I
don't
really
see
a
problem
with
upgrading
that
being
an
issue.
So
maybe
just
saying
like
this
is
all
you
need
to
do
so.
Maybe
a
guide
on
how
you
do
that
might
have
been
just
a
little
bit
like.
Oh,
it's
broken,
it's
not
compatible
full
stop,
whereas
just
a
little
bit
more
detail
when
it
would
have
been
nicer.
I
think.
A
For
this
particular
change,
I
would
argue
that
captain
upgrade
could
have
been
also
updated,
because
this
is
a
change
that
can
be
done
automatically
or
at
least
in
a
guided
way.
So
because,
when
you
prepare
to
the
upgrade,
you
can
get
all
the
information.
What
you
need
to
do,
because,
even
if
you
put
it
in
the
documentation,
half
of
the
people
won't
read
it.
D
B
A
I
also
read
the
documentation
only
after
I
hit
issues
with
separate,
I
just
browsed
through
change
log.
None
of
that
seemed
to
be
particularly
concerning
for
my
keys
but
well.
Finally,
he
did
it's
just
a
matter
of
some
recommendations
should
be
detailed
enough
so
that
it
catches
ie
for
user,
and
some
could
be
automated
so
that
you
cannot
make
a
mistake.
A
A
F
I
I
just
had
a
couple
quick
questions
actually,
with
regards
to
summer
code,
I
guess
the
first
one
has
to
do
with,
like
I
noticed
on
the
timeline,
there's
sort
of,
I
guess
like
an
initial
deadline
of
mid-september
and
then
an
extended
deadline
that
goes
to
like
the
end
of
november.
F
So
I
was
wondering
like
if
that
is
something
that
we
can
kind
of
take
into
account
in
a
proposal
like
specifically
that
period
between
like
the
end
of
september,
or,
I
guess,
like
middle
of
september,
to
the
end
of
november
or
if
that
should
just
purely
be
seen
as
like
a
buffer
in
case.
Not
everything
gets
done.
B
F
Yeah
well,
actually,
I'm
specifically
asking
with
regards
to
like
on
the
I
guess
the
program
was
website.
I
think
it's
like
developers.google
there's
like
there's
a
timeline
where
the
project
ends
sort
of
mid-september
and
then
there's
like
an
extended
deadline
where
it
goes
sort
of
to
the
end
of
november.
F
So
I
was
wondering
if
that
period
between
the
middle
of
september
to
the
end
of
november
should
like
be
purely
seen
as
as
a
buffer.
In
case
things
don't
get
finished
or
if
we
can
kind
of
like
use
that
in
our
planning,
as
well
like.
A
For
example,
so,
basically,
when
you
make
a
proposal,
you
define
what
would
be
the
timeline
and
my
recommendation
that
if
you
want
to
target
this
additional
time
to
have
let's
say
less
relaxed
time
during
the
implementing
sort
of
more
relaxed
time
during
the
summer
because
of
other
activities
or
if
you
want
to
just
want
to
do
more,
then
it's
something
you
need
to
discuss
with
mentors.
A
So
it's
not
the
first
time
where
if
something
goes
wrong
you
can
extend
well,
it
can
be
done
in
each
cases,
but
ultimately
for
the
most
of
the
projects.
You
should
plan
it
beforehand.
F
Okay,
that
sounds
good.
So
as
long
as
it's
sort
of
like
clear
what
the
timeline
is,
we
can
use
the
full.
I
guess
like
even
to
the
end
of
november
this
one,
what
I'm
getting
yep.
F
A
F
Perfect
yeah,
so
actually
I
did
have
one
other
question
as
well,
and
this
is
kind
of
more
specific
to
the
get
ops
project
which
is
kind
of
what
I'm
interested
in
like.
I
noticed
that
so,
like
thomas,
has
obviously
done
a
lot
of
work
on
this
already,
where
there's
like
a
fully
working
poc
so
like
it's
not
like
something
I'd
necessarily
be
building
from
scratch.
F
If
I
were
to
to
contribute
like
if
I
were
to,
I
guess,
like
suggest
enhancements
as
part
of
my
proposal
like,
I
was
wondering
if
there
was
because
I'm
looking
at,
I
think
it's
kept.
67
and
69
are
the.
F
That
are
currently
already
present
and
then
there's
like
issues
on
the
repo,
though
like
I
was
wondering
if
I'm
missing
anything
in
terms
of
like,
what's
already
planned
for
other
than
what
I
just
mentioned
in
case
like
I
don't
want
to
like
suggest
something
that's
already
being
worked
on
by
someone
else
or
by
thomas.
A
A
So,
basically,
looking
at
this
information,
you
can
take
a
slice
of
work
or
you
would
like
to
focus
on
and
then
discuss
it.
Let's
say
in
the
slack
chat.
So
basically,
once
you
have
an
idea
what
you
would
like
to
work
on
as
a
part
of
this
project,
it's
definitely
something
you
could
discuss
so
that,
firstly,
thomas
me
and
other
potential
mentors
know
what
you
want
to
work
on.
A
So
we
can
see
how
we
easily
this
work
so
that
it
can
become
a
g-sub-project
set
according
to
the
requirements,
and
we
can
also
see
what
how
to
actually
adjust
so,
for
example,
what
but
not
to
touch,
because
somebody
is
planning
to
work
on
that
already
or
maybe
other
topics.
So
in
this
case,
it's
important
to
have
a
discussion
in
the
community
so
that,
basically,
you
coordinate
your
work
with
whatever
happens,
and
we
still
have
a
few
weeks
to
do
that
before
the
application
deadline.
F
Awesome
so
I
guess,
like
I
guess,
in
terms
of
like
what's
happening
right
now
with
get
ops.
Is
it
mostly
thomas
that
I
should
be
talking
to
you.
A
Yeah
thomas
is
basically
a
champion
for
this
project,
so
he's
doing
the
most
of
architecture
work.
The
most
of
the
coordination,
so
thomas
would
be
definitely
the
key
source
of
information
for
whatever
happens
in
this
project.
A
B
Awesome
and
eric,
if
you
haven't,
if
you
haven't,
met
him,
he's
super
friendly,
so
really
approachable
he's
always
you
know
happy
to
talk
so
yeah
definitely
reach
out
to
him.
F
Awesome
yeah
he's
like
I've
already
gotten
some
feedback
from
thomas
and
he's
been
really
really
gracious.
I
think
with
with
some
of
my
dumb
questions,
so
this
is
a
dumb
question
right
right,
awesome,
yeah!
Those
are
my
questions
thanks.
So
much
guys
welcome.
G
Did
we
already
discuss
the
kept
70?
I
think
it
is
adding
state
on
just
on
the
topic
of
the
githubs,
adding
and
removing
stages.
Maybe
that's
a
one
for
a
separate
meeting.
A
Well
for
this
topic,
it
would
be
really
nice
to
have
johannes
on
the
call,
because
that's.
A
This
effort
yeah
just
to
show
the
context
it's
the
new
one.
A
Yesterday,
there
was
also
a
question
in
slack
asking
about
how
to
actually
reimport
the
projects
from
the
repository
and
all
of
that
actually
hits
this
area
as
a
problematic
thing,
because
once
you
have
the
structure
of
stages,
you
can't
do
anything
about
it.
Unless
you
recreate
the
project
and
if
you
recreate
the
project,
I
believe
you
also
lose
history,
which
is
not
unacceptable
in
some
cases.
So
basically
it
will
be
really
nice
if
you
were
able
to
modify
it
from
the
fight.
A
Do
I
capture
this
story
correctly,
adam.
G
I
know
sorry,
that's
a
different
kept,
never
mind.
Ignore
me
yeah
absolutely
so
this
to
me
is
when
I
was
playing
around
with
the
get
ops
operator.
This
is
the
missing
piece
of
the
puzzle,
because
we
can
do
everything
else
apart
from
stages
and
this
forces
us
into
pre-deploying
a
shipyard
file.
So
this
is
kind
of.
G
For
anything,
git
ops,
this
kept
would
be
the
focus
first
and
then,
and
then
that
enables
the
rest
of
it.
So.
A
Okay,
so
for
that,
I
guess
all
of
us
are
on
the
same
page
that
it's
quite
important.
What
is
currently
missing
is
timeline,
so
I'm
meeting
with
johannes
tomorrow,
I
will
actually
try
to
understand
how
the
captain
team
sees
it
in
terms
of
priorities
and
yeah.
I
think
that
is
something
we
can
start
preventing
grant
work
for,
because
if
you
can
do
that,
let's
say
in
the
next
three
months
or
so
then.
A
Then
you
should
deploy
shipyards,
because
if
you
define
deploy
everything
in
reverse
direction
at
some
point
there
might
be
some
validation
which
doesn't
pass
etc.
So
it
doesn't
seem
to
be
right
in
longer
term.
A
A
B
For
the
get
ops,
I
haven't
quite
looked
into
it
yet,
but
it
probably
would
be
nice
to
reference
your
own
secrets
as
well.
Instead
of
using
the
rsa
encryption
method,
does
it
so
I'm
going
to
look
into
it
first
to
see
if
it's
actually
possible
with
how
it
works?
But
it
would
be
nice
if
you
can
reference
an
external
secret
as
well
that
doesn't
have
to
be
baked
into
you
know
doing
it,
because
you're
still
imperative
one
way
or
another.
B
B
B
I
want
to
do
it
my
way,
not
your
way,
but
I
still
have
to
see
if
it
makes
sense,
because
I
haven't
looked
into
the
current
property
yet
so
maybe
I'll
talk
about
it
next
week,
based
on
my
findings,
otherwise
I
might
not
even
bring
it
off,
but
I
guess
it's
good
to
to
bring
up
like
that's
the
stuff
I
would
like
to
see,
but
whether
it's
possible
or
anyone
else
wants
it.
It's
a
different
story.
D
B
But
it's
just
good
to
bring
up
that
use
case
when
it's
sort
of
a
new
thing
and
not
doing
it.
You
know
when
we
have
a
thousand
people
using
it,
but
you
wouldn't
break
any
changes
anyway,
because
you
just
write
the
helm
chart
differently,
but
I
don't
know
I'll
bring
that
up
next
time,
based
on
my
findings,
I'll
have
a
look
into
it,
a
little
bit
more
of
how
it
works.
D
A
So
yeah
I'll
ask
johannes
for
having
let's
say
a
deep
life
discussion
on
thursday,
so
maybe
he
could
actually
present
what
we
currently
have
and
what
is
the
timeline
from
his
point
of
view,
I
could
also
try
inviting
thomas
just
to
see
what
do
you
think
would
it
be
possible
at
this
stage?
G
I
think
so
I
think
it's
gonna
be
heavily
utilized.
Obviously,
so
we
don't
want
to
get
into
a
situation
where
one
hasn't
considered
the
other.
I
I'm
imagining
it's
yes,
it'll
work,
I'm
imagining
that's
the
answer,
but
obviously
just
getting
just
making
sure
that
everybody's
on
the
same
page
is
always
a
good
thing.
A
A
A
A
Let's
keep
using
slide
channels
for
any
synchronous
discussion
and
if
needed,
we
can
set
up
additional
meetings
on
demand.
But
right
now
what
I
see
many
students
already
submit
proposal
drafts
and
for,
if
you're
a
mentor,
it's
a
good
time
to
actually
take
a
look
at
them.
If
you
provide
some
feedback,
because
this
is
the
most
important
part
of
the
application
phase
for
contributors.