►
From YouTube: Keptn Refinement Meeting - Jun 27, 2023
Description
Meeting notes: https://docs.google.com/document/d/10Fig1eYFZ9iQFSYWkz0c4eTwzgJiPtQI5IsczbvLsuE/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
Recording
and
we
can
kick
it
off
one
second,
let
me
find
the
right
tab
to
share
here.
We
are
so
as
always.
Let
me
share
this
in
the
chat,
so
everyone
can
have
access
to
it.
A
Oki,
so
we
have
some
tickets
to
refine
foreign
from
Moritz
to
migrate
from
a
single
Helm
chart
to
an
umbrella
Helm
chart.
Currently
we
call
ourselves
lifecycle
toolkit,
but
we
cannot
install
klt
as
a
toolkit,
because
it's
one
single
hem
chart
that
installs
everything
for
you.
You
cannot
pick
and
choose
the
right
things
that
you
would
like
to
deploy.
For
instance,
if
you
want
only
to
scale
your
workloads,
basil
metrics,
you
cannot
just
deploy
the
metric
server.
A
A
How
now
said
this
would
be
handled,
should
we
bundle
the
difference
30
in
the
different
charts,
or
we
should
keep
them
on
the
umbrella
top
level
chart
when
they
are
sure
between
different
components,
we'll
copy
them
in
multiple
places
or
not?
A
A
B
B
A
A
And
always
it's
hard
to
have
yes
or
no
answers,
but
with
the
POC
we
can
also
then
see
yeah
the
better
part.
One
thing
that
really
is
unknown
to
me
at
the
moment
is
help
me
find.
A
A
C
The
question
the
current
docs
say
that
you
can
install
the
metric
operator
by
modifying
the
helm
chart
without
having
the
scheduler.
Is
that
true?
Is
it
useless
since
I
think
certain
people
are
going
around
to
all
the
conferences
saying
they
can
do
that?
The.
A
Issue
of
the
moment
is,
you
can
have
that
behavior,
but
all
the
configuration
around
those
components
is
installed
anyway.
Just
we
don't
consume
resources.
Basically,
you
install
the
scheduler,
but
then
you
scale
it
to
zero.
So
virtually
is
not
in
your
cluster,
but
all
the
config
configuration
around
the
scheduler
has
been
installed.
C
D
A
Then
we
have
two
very
similar
tickets,
which
is
a
follow-up
from
the
recent
PR
that
Moritz
worked
on,
which
is
configure
everything
with
a
mono
repo.
This
is
a
kind
of
a
prerequisite
for
the
umbrella
charts,
because
the
idea
is
that
we
want
to
decouple
the
release
process
of
each
component
of
Captain,
so
whenever
there
is
a,
for
instance,
a
security
vulnerability
in
the
metric
operator,
we
don't
need
to
release
the
lifecycle
operator,
the
metric
operator,
the
scheduler
all
together.
A
A
B
Yeah,
so
Minor
Adjustments
will
need
to
happen
that
are
not
done
yet,
which
is
right
now
we
have
just
two
components:
KRT
in
it
in
a
hole
and
the
third
measure,
and
the
pipeline
is
kind
of
hardcoded
for
those
two
components
and
you
need
to
basically
switch
it
to
be
able
to
have
a
more
parameter
specifically
have
less
hardcore
variables,
but
other
than
that.
The
search
manager
PR
is
merged
and
works
already.
B
So
it's
a
it's
a
working
example
position.
A
Separation
of
opinions-
here
maybe
Andre,
let's
start
from
the
m
y
and
M
in
your
opinion,.
E
I
mean
there
is
always
things
aren't
always
straightforward
when
the
pipelines,
the
workforce,
in
this
case.
A
Me
it's
so
this
I
agree
with
Moritz
mag
from
just
looking
at
this
PR
to
me
something
s.
What
let
me
bump
it
to
an
M.
Was
it's
not
exactly
like
this,
because
we
need
to
provide
more
input
parameters
and
this
might
increase
a
bit
the
complexity.
Where
is
the
releasable
somewhere?
Remember?
There
is
a
logic
here:
artifact.
B
Matrix
logic,
basically
I,
don't
need
to
be
adjusted
and
all
the
variables
at
the
top.
As
you
can
see,
if
you
scroll
up
that's
the
variables,
I
meant
yeah
up
here,
a
photo
version,
so
the
hard
code
for
now,
okay,
between
S
and
M,
anyways.
A
A
Do
we
have
chances
for
pipeline
no,
but
it's
fine
and
I
will
keep
it
the
same
for
this
one.
Then.
A
Migration
strategy,
so
you
installed
klt
with
a
manifest.
Now
you
want
to
move
to
the
helm
chart.
How
can
you
do
that,
and
this
is
basically
providing
an
updated
version
of
the
documentation
inside
the
install
and
upgrade
filter
sections
of
the
documentation
before
exactly
delete?
That
is
this
a
good
explanation,
mag
I
think
so.
Yeah.
C
It's
I
need
more
I
need
a
little
bit
more
information.
Obviously,
for
more
but
and
part
of
this
is
what
do
we
do
now
and
what
do
we
wait
until
later?
You
do
so.
A
C
A
No
but
the
worst
worst
case
scenario:
the
upgrade
can
be
Cube
CTL
delete,
minus
F,
the
Manifest
that
used
to
install
and
then
run
Helm
install.
So
for
a
period
of
time
you
don't
have
klt
in
your
cluster,
so
you
need
to
shut
down
the
deployment
of
your
services
until
you
do
this
upgrade
of
from
manifest
to
install
from
help
to
sorry
from
manifest
to
helm.
And
then
you
can
resume
your
deployments
through
the
klt.
A
Then
the
second
one
is:
okay.
Now
that
we
have
an
umbrella
hum,
chart
it's
easier
to
install
only
different
components,
and
it's
good
that
we
explain.
Okay,
if
you
want
to
use
just
the
metric
use
case,
how
can
you
he
stole
that
or
if
you
want
to
have
your
own
search
manager?
How
can
you
do
that
with
a
new
umbrella
chart.
C
Right
actually,
I
wrote
this.
We
already
tell
them
how
to
install
their
own
cert
manager,
but
it's
also
the
man
of
to
the
Manifest
and
I
realize
that
it's
in
the
wrong
there's
a
couple
of
structural
issues
in
what
we've
got
there.
So
that's
what
I
had
actually
written
this
before
I
knew
the
mono
repo
and
all
of
that
was
coming
so
so
yeah.
C
I
say
we're
saying
we're
doing
it
right
now
and
we
tell
them.
We
tell
them
how
to
modify
the
helm
chart
and
that
we're
just
saying
you
know
that
we
should
give
them
an
example.
That
was
the
point
when
I
wrote
this
now,
the
question
is:
should
we
do
that
in
the
meantime,
or
shall
we
wait
for
by
the
way
this
the
stuff
that
you're
talking
about
doing?
Are
we
talking
about
this
for
09
or
o10
or
30.
B
D
B
But
in
the
meantime
there
we
don't
really
have
a
way
of
doing
that
through
Helm
I.
Think.
B
I'm
not
fully
sure
what
the
metrics
operator
talks
say.
B
C
C
It's
a
couple
of
oh
shoot,
I,
don't
know
it's
not
well
documented.
It's
a
mention.
B
C
A
Think
was
in
the
getting
started
guide,
maybe
custom
metrics.
C
A
A
C
Maybe
maybe
the
good
news
is
with
it
poorly
documented
like
it
is.
You
have
to
really
know
what
you're
doing
to
do
it.
So
you
know
we
wouldn't
get
some
naive
young
thing
in
here,
whereas
once
we
say
you
know
type,
you
know
replace
this
with
this.
Anybody
can
do
it
so
and
especially,
but
if
we're
talking
about
this
may
be
solved
in
the
next
release,
we
just
wait
and
see
what
you
have
and
then
we
can
update
that.
I
can
move
a
couple
things
around
I.
A
A
F
C
The
one
thing
that
I
think
I
will
do
that
I
realized
in
reading
it
is
we
have
the
chapter
for
install
or
those
page
installed
from
the
helm
chart
and
then
after
you
go
through
all
those
instructions,
it
says:
here's
how
you
modify
the
helm
chart
and
that's
so
you
know
if
I'm
going
through
and
following
stuff
sequentially
I've
just
installed
this
and
now
I
find
out
I
have
to
modify
the
helm
charge
before
I
install
it.
So
I
think
I
should
pull
that
out
as
an
optional
step
before
the
installation
page.
Actually.
C
So
that
will
be
where
the
bring
your
own
managerias
and
put
as
an
optional
and
we'll
say
you
know
you,
you
don't
need
to
install
the
helm
chart.
But
if
you're
going
to
make
modifications
to
your
configuration
you
have
to
well
so
now
we'll
say
you
have
to
modify
either
the
home
chart
or
the
Manifest
before
you
install
and
we'll
put
that
and
then
we'll
put
what
we
this
information
that
we've
got
here
and
then
we
can
augment
that
later
on.
C
C
Okay
well
I
know
in
the
captain,
V1
docs
there's
some
places
where
it
says
you
know,
using
the
command
line.
Options
is
not
the
greatest,
not
the
smartest
thing
to
do,
and
you
should
modify
the
values.yaml
file
and
other
places
where
it
just
talks
about.
You
know.
Yeah,
just
you
know,
don't
worry
about
values
that
yaml
just
use
those
things.
So
we're
not
real
clear
on
that.
So
I'm
just
gonna
leave
I'm
busy
with
migration
I'm
gonna.
Let
this
sit
till.
B
C
F
C
B
C
Right
right,
another
value
thing
is
the
still
the
question
you're
saying
that
for
klt
there's
no
need
to
ever
adjust
the
size
of
this
of
the
cluster
but
and
then
but
Adam
said
when
he
was
doing
that
little
tape
to
help
the
g-suckers
or
his
little
video
on
his
laptop.
He
had
to
reduce
the
size,
or
else
he
couldn't
get
it
installed
on
his
laptop
there
was
that
and
I
don't
know
if
there's
things
that
can
make
it
grow,
but
that's
a
another
question:
I
guess:
I!
Guess
as
we
go
through.
A
A
But
if
you
want
to
then
deploy
an
application,
which
is
very
big,
you
want
to
have
open
telemetries,
so
you
have
an
open,
Telemetry
collector.
You
have
a
Prometheus
instance
to
scrape
Matrix
a
Jaeger
installation
to
see
the
traces
grafana
to
present
some
dashboards.
This
is
outside
of
our
own
sizing
and.
D
A
We
solved
this
in
killer
coda
that
week
we
could
install
klt
no
issue,
but
then,
if
you
want
to
also
have
a
dashboard
in
Via
grafana
with
Prometheus
metric
script
through
the
Prometheus
operator,
it
was
not
possible.
This
was
too
many
resources
asked
by
Prometheus,
not
by
us,
but
I
know
that
you
need
to
have
that
part
of
Prometheus
in
order
to
scrape
the
metrics
of
klt.
Otherwise
it's
basically
useless
the
observability
part.
C
C
Yeah
I
think
I
plagiarized
you
for
this
too,
and
what
you
just
said
was
actually
very
good
and
but
along
with
that
and
then
explicitly
say
you
know,
if
you
need
to
change
your
size,
modify
this
and
this
not
not
tell
them
what
the
number
is.
B
C
Okay
and
and
also
the
other
one,
is
to
know
because
I
know
for
Captain
V1.
We
constantly
had
people
in
help
that
they
couldn't
get
all
their
pods
up,
because
they
didn't
have
enough
space.
D
C
D
C
B
B
C
B
C
B
C
D
F
C
A
Okay,
last
ticket
of
today,
for
the
refinement
is
to
move
and
update
the
information
about
the
cert
manager.
How
someone
can
use
their
own
set
manager-
and
this
already
put
it
this
morning-
that
we
need
to
depend
on
the
umbrella
chart
changes
before
we
can
adapt
the
documentation,
because
currently,
as
the
current
time
chart
is,
we
cannot
change
the
search
manager
because
we
don't
allow
to
change
annotations
and
label
through
the
Hem
charts,
and
we
need
to
adopt
to
the
new
umbrella
chart
to
have
this
possible.
C
And
if
we
would
have,
if
we,
this
isn't
ready
for,
if
none
of
this
umbrella
step
is
ready
for
zero,
nine
I
might
just
add
a
note
to
what,
because
what
we've
got,
except
for
the
fact
that
you
think
manifest
I
actually
like
I,
think
it's
fairly
well
done.
I
forget
who
gave
me
the
information,
but
you
did
good
if
we
would
have
to
go
to
zero
nine
zero
without
the
umbrella.
I
might
add
a
note
that
it's
not
currently
possible.
A
C
A
F
Yes,
I
had
one
question:
yes,
so
I
wanted
to
ask
that
what
does
the
code
Cove
patch
test
Do's
like?
Why
does
it
throw
as
errors
in
a
PR
I'm
not
really
being
able
to
troubleshoot
the
error?
Sweet
I'll
share
the
pr.
A
A
F
A
This
is
check
that
you
provide
enough
uni
test
for
your
code,
you're
changing.
So
this
is
sorry.
Let's
have
a
look
at
the
name,
it's
called
diff
or
patch,
and
you
can
see
it's
called
diff
hit.
A
That's
why
it's
failing,
and
if
you
cook
sorry,
if
you
click
on
details,
you
then
can
check
on
code
curve
why
it
fails
and
provides
you
with
info
about
what
are
the
lines
that
are
not
covered
green
means.
There
are
uni
tests
that
covered
its
piece
of
code
and
yellow
means
that
not
all
code
or
not
all
branches
are
covered
and
red
means
no
test
coverage
so
test
never
execute
this
piece
of
code.
F
A
C
A
A
C
Mean
it
well,
it
just
struck
me
something
that
I
was
thinking
when
we
tell
them
to
set
up
their
clubs,
their
kubernetes
cluster.
How
many
do
they
need
like
if
I've
got
a
shipyard,
that's
deploying
like
a
Dev
and
hardening
and
production
and
they've
each
got
a
deployment.
A
A
C
C
D
C
Right
so
be
nice
to
say
this
is
you
know,
you're
gaining
something
here,
because
along
the
way,
we're
having
to
say
hey,
you
can't
quite
do
this
anymore.
So,
but
it's
it's
paradigm
shift.
So
anyhow,
I
did
a
very
rough
and
I.
Oh
some
of
the
stuff
I
couldn't
even
guess
at
and
a
significant
portion
of
my
guesses
are
wrong,
but
I,
throw
it
out
and
just
for
comment
I
when
we
Whatever
Gets
written
up
and
I
didn't
even
bother
with
the
xrefs
to
other
stuff.
It's
like
this
is
just
too
volatile.
C
A
C
Can
brought
you
some
feedback,
yeah
I'm
good
at
that
yeah,
if
I'm
good
at
bad
terminology,
yes
and
and
to
see
it
and
then
because
then
I'm
also
thinking
what
order
like
if
I've
got
Integrations
in
my
J.
Well,
if
I
got
Integrations
in
my
Jes,
I,
probably
don't
need
them
like
in
integration
for
dynatrace.
I!
No
longer
need
that,
because
I'm
going
to
have
my
captain
metrics
provider
and
that
will
Define
that
it's
not
an
integration
but.
D
A
C
A
minute
if
I'm
doing
Jamie,
no,
we
have
too
few
terms
and
we
use
them
for
too
many
each
term.
For
too
many
things,
computer
science
people
need
bigger.
Vocabularies
is
my
theory,
but
so
what
did
you
just
say
like
for?
If
I
want
to
run,
say
J
meter?
C
A
C
B
C
C
E
D
B
E
C
Right,
okay,
we'll
keep
working
at
this
I'm
sure
to
get
a
lot
before,
but
I
just
started:
I'm
real
I
thought
without
having
that
I'm
starting
to
write
all
these
other
pieces,
and
it's
like
no,
it's
not
going
well
because
I
don't
have
you
know.
I
I
was
starting
to
write
something
about
how
you
start
from
scratch
to
implement
klt
and
it's
like.
C
No,
this
is
I've
got
something
that's
working
I
want
to
make
it
work
under
klt,
so
I
need
to
see
what
I
can
map
and
what
I
can't
and
okay,
well
I'll
put
that
out
and
feel
free
to
knit
and
pick
and
whatever
edit.
It's
not
something
that
I'm
proposing
as
let's
approve
this
and
publish
it
sure.
So
it's
part
of
the
process
as
they
say.
Thank.
C
A
So,
tomorrow,
community
meeting
I
want
to
announce
something
ready
here.
We
have
quite
a
lot
of
meetings
in
the
captain
project
more
than
12
meetings
every
month.
If
we
would
like
to
get
down
a
bit
less
meetings.
Otherwise
you
know,
as
a
maintainer
contributes
it's
hard
to
get
into
the
community.
There
are
a
lot
of
time
wasted
between
quotation
mark
on
a
zoom
call,
so
we
might
rework
how
we
structure
this
refinement
meeting.
A
It
might
be
less
frequent,
so
every
two
weeks
instead
of
every
week-
or
maybe
we
combine
with
other
meetings
undecided
yet
but
we'll
let
you
know
as
soon
as
we
have
more
information
about
how
we
will
change
the
the
thing
and
looking
forward
to
tomorrow
committee
meetings,
because
worksheet
will
showcase
a
bit
of
the
gsoc
progress
that
he
did
so
far.