►
From YouTube: Keptn Community & Developer Meeting - Mar 10, 2022
Description
New feature demos by the Keptn team, GSoogle Summer of Code discussions.
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
Hi
all
welcome
to
the
captain
developer
meeting
today
is
march
10th
and
thanks
a
lot
to
everyone
who
joined
the
call.
I
guess
this
is
the
last
meeting
before
we
have
a
time
zone
change
hell.
So
please
it's
actually
one
topic.
I
wanted
to
discuss
what
time
we
keep
ugc
or
central
central
european
time
for
this
meeting,
because
yeah,
I
believe
everyone
expects
central
european
time
but
yeah.
I
just
wanted
to
confirm
it
explicitly.
A
So
basically,
it
impacts
just
one
meeting
unless
you
are
based
in
countries
that
do
not
change
time
zones
like
russia,
some
us
or
and
australian
states.
If
you
recall
correctly,
so
I
think
that
we
can
just
keep
central
european
time
taking
the
majority
of
contributors
so
rush
megrash.
Is
it
fine
with
you.
C
D
B
Yes,
so
we
just
released
two
patch
versions,
one
for
12
and
13,
but
unfortunately
we
discover
another
small
problem
to
to
our
system
to
be
distributed.
You
know
there
is
always
some
hidden
problem
about
communication
timing
between
different
messages
and
therefore
we
will
also
follow
up
with
another
patch
release
for
13
and
12,
to
also
fix
this
small
problems,
with
the
distributor
that
they
can
get
out
of
sync
with
nets.
B
A
So
at
the
moment
we
want
to
announce
this
patch
releases
right
until
it's
fully
fixed
yeah.
B
Yes,
so
directors
donated
to
the
captain
community,
one
environment
that
we
can
use
in
order
to
monitor
our
integration
tests,
so
I
will
invite
all
the
maintainers
of
the
project
to
this
environment.
So
whenever
they
trigger
any
integration
test
run,
then
they
can
see
the
results
of
the
monitoring
in
this
director's
environment
to
help
debugging
issues.
B
A
So
speaking
of
that
yeah,
I
created
a
folder
for
project
infrastructure,
so
maybe,
when
we
create
new
infrastructure
components,
it
makes
sense
to
just
put
some
really
small
right
up
here,
so
that
people
know
that
at
least
it
exists.
B
A
B
A
A
A
B
A
Okay
next
captain
is
accepted
to
google
summer
of
code,
so
now
fully
official.
So
we
participate.
We
are
listed
on
the
website
and
basically,
we
already
have
student
potential
contributors
reaching
out
so
this
year,
it's
not
about
students,
so
this
year
even
professionals
can
participate
and
yeah
so
we're
here.
All
the
documentation
is
in
place.
We
gradually
extend
it
based
on
the
feedback.
I
will
be
submitting
another
pull
request,
but
what
is
important
is
222
project
ids.
A
So
this
is
what
we
have
on
the
table
now:
seven
ideas
and
for
every
ide
we
still
look
for
potential
mentors
and
they
still
time
frame
to
submit
a
new
project
ideas
if
you're
interested
to
do
so.
A
So
just
keep
it
in
mind
and
thanks
a
lot
to
everyone
who
participated
in
the
application
phase,
because
yeah
gsoc
is
a
competitive
program.
Less
than
50
percent
of
organizations
are
accepted
and
it's
great
to
have
captain
accepted
this
year.
A
I
think
we
should
at
least
discuss
that
yeah.
I'm
sharing
my
screen
again
so
here
there
is
a
project
idea
by
meha
already
created
so
basically
a
result
of
the
discussion
so
and
if
you
have
a
project
id,
if
you
have
potential
mentors,
then
I
think
we
should
apply
so
google
season
of
docs
is
a
counterpart
of
summer
of
code
which
focuses
on
documentation.
A
A
Okay,
so
if
anyone
has
a
well,
we
have
plenty
of
ideas
what
we
can
do
with
our
documentation.
Yeah
I
mean
there
are
various
opinions
from
radical
to
very
radical
ones.
So
yeah
any
ideas
are
welcome.
F
C
A
A
Okay,
so
let's
press
it
with
next
item,
so
there
was
a
topic
from
rajiv
about
captain
documentation,
site
revamp,
but
rajiv
is
not
on
the
call.
So
I
guess
we
can
either
skip
it
or
put
it
into
the
end
of
the
meeting
if
he
joins
yeah
and
meanwhile
welcome
welcome
if
I
pronounce
it
correctly
so
yeah
yeah.
G
Cisco
and
like
I
have
like
a
lot
of
interest
working
with
cloud
native
applications
and
tools,
so
therein
I
got
into
this
open
source
contribution.
I
have
been
contributing
to
this
community
since
two
years
now,
and
captain
is
one
of
the
application
that
have
worked
as
a
poc
a
couple
of
years
ago,
and
that
is
where
I
you
know.
I
was
introduced
to
captain
project
and
this
time
when
I
saw
that
they
are
going
to
create
something
from
the
scratch
the
integration
with
the
backstage.
G
F
A
Okay,
thanks
for
your
interest
and
yeah,
hopefully
we
can
deep
dive
in
the
project,
so
we
can
do
it
probably
later
today,
if
you're
fine
with
it.
So
we
can
run
through
the
agenda
and
then
just
do
deep
dive
for
this
discussion.
Would
you
believe
that
okay,
so
then
we
can
proceed
with
our
regular
demos.
D
Good
yeah,
so
the
first
two
are
bug
fixes.
The
first
is
we
exp
experience
the
segmentation
fault
in
his
helm
service,
where
actually
the
helm
service
tried
to
proceed.
We
had
actually
on
unresolved
error
where
the,
where
the
helm
charts
when,
when
the
helm
charts
had
loading
problems,
then
actually
the
code
just
moved
on
and
experienced
the
segmentation
fault.
So
this
was
found
out
by.
D
D
There
were
some
minor
problems
with
error
messages
which
were
doubled,
yeah
upon
that
we
actually
refactored
the
secret
service
and
also
the
shipyard
controller,
which
was
actually
uniformed
all
the
error
messages
and
the
handling
of
the
error
messages.
D
So
now
it
should
be
quite
quite
clean
and
easy
to
use
for
the
next
new
error
messages
or
new
contributors
yeah
and
the
future
thing.
Actually
from
now,
we
support
the
authentication
authentication
to
remote
git
repository
via
proxy
server.
D
Actually,
this
can
be
known
from
community
users
from,
for
example,
github
configurations.
So
as
you
can
configure
a
git
proxy
through
a
normal
git
command,
and
this
will
this
will
be
your
configuration.
D
So
the
all
the
traffic
from
your
local
repository
to
your
git,
remote
repository
will
be
rooted
for
a
proxy
server.
So
actually
we
support
this.
Also
in
captain
for
now
we
added
some.
This
is
an
example
of
the
using
the
cli.
So
when
creating
a
project,
you
add
the
parameters
that
get
user
git
token
with
the
remote
url
as
already
known.
D
It's
optional:
okay,
it's
not
mandatory!
It's
only
optional
when
you
want
to
use
the
git
proxy
and
root
your
traffic
through
the
proxy
server.
So
if
you
use
it
that
it's
up
to
you,
you
can
actually
need
to
use
it's
like
it's
implemented
for
the
github
repository,
for
example.
H
Okay,
I
have
two
pro
requests
today
that
I
want
to
show
you
so
the
first
one
is
in
the
api
for
the
uniform
subscription.
So
giovanni
already
mentioned
this,
so
we
had
a
bug
there.
C
H
In
some
cases,
the
subscriptions
for
an
already
is
that
uniform
might
have
been
lost.
So
basically,
let's
really
quickly
look
at
the
faulty
code
here
so
yeah,
as
you
can
see
here
on
the
left
side.
If
the
version
of
the
integration
that's
re,
restoring
itself
has
changed,
then
we
just
over
or
re-registered
the
integration
and
that
process
also
overwriting
any
subscriptions
that
might
have
been
set
up.
F
H
All
right,
if
not
so,
let's
continue
with
the
next
one,
so
this
is
about
the
shipyard
controller.
This
was
quite
a
big
pull
request,
so.
C
H
Here,
but
basically,
what
was
that
about?
I
think
I've
teased
it
in
one
of
the
recent
developer
meetings.
So
basically
we
had
to
implement
a
new
data
model
for
the
shipyard
controller
or
for
the
for
representing
the
sequence
execution
states
in
order
to
be
to
have
a
thread
safe
model
which
in
turn
will
allow
us
to
have
multiple
replicas
of
debian
controller
and
yeah.
H
So
previously
the
state
was
more
or
less
spread
across
a
set
of
different
collections
in
mongodb,
but
now
we
managed
it
to
aggregate
all
the
information
that
we
need
for
execution
of
a
particular
sequence
into
a
document
that
lives
within
one
collection,
and
this
also
made
the
code
for
the
shipyard
controller
a
lot
easier
and
hopefully
also
in
turn
more
robust
and
more
maintainable.
H
I
And
one
note
from
my
thoughts:
we
have
that
readme
markdown
file
for
the
shipyard
controller,
where
we
currently
describing
the
old
structure
and
the
word
mongodb
collections
and
so
on.
I
think
we
should
also,
with
a
future
pull
request,
update
this
one
either
delete
it
if
we
don't
have
a
new
one
or
update
it
to
the
new
model,
or
did
you
already.
H
B
A
Yes,
yeah
one
thing
we
could
do
just
once.
We
finish
the
demos
part
we
can
just
deep
dive
in
jsoc
projects.
Would
it
be
fun
vision?
G
I
I
Yeah?
What
do
I
have
prepared
these
four
little
things
here?
First
of
all,
I
kind
of
restructured
the
packet
structure
of
the
distribute,
distribu
distributor.
Why?
Because
we
had
their
kind
of
very,
very
flat
structure,
which
actually
didn't
make
sense
at
all.
This
has
historical
reasons
and
so
on
what
I've
just
did
and
that
you
can
see
in
the
pictures
depicted
here
in
the
pr
and
just
made
the
yeah
the
package
just
way
more
smaller
and
tried
to
assemble
them
them
to
their
responsibilities.
I
I
In
general,
I
will
propose
to
take
a
little
bit
of
focus,
also
in
other
modules
of
captain
core,
to
think
about
restructuring
a
bit
our
backus
package
structure,
because
it
makes
certain
things
a
little
bit
cleaner.
You
don't
have
name
clashes
and
so
on
and
we
can
follow
more
or
less
more
best
practices
which
which
are
used
in
respect
to
that
in
the
go
universe.
I
I
would
say
I
also
asked
in
our
slack
channel
if
we
should,
if
we
could
have
a
discussion
on
what
are
the
current
best
practice
when
it
comes
to
package
design
and
go,
I
can,
if
you
want,
prepare
or
put
together
some
notes
there
and
we
can
discuss
what
is
applicable
to
our
captain
corey
point
what
we
should
or
don't
don't
want
to
do
yeah,
but
for
now
I
just
did
it
on
the
distributor
thing
and
it
all
already
had
a
benefit
when
for
a
testing
code,
where
we
can
now
write
the
test
function
without
having
clashes
in
other
tests
where
the
same
test
function
is
already
been
used
and
so
on.
I
I
This
was
finished
back
then,
and
one
thing
which
was
left
to
do
was
to
trigger
the
authorization
code
flow
automatically
when,
for
example,
the
refresh
token
is
expired.
So
what
already
did
work
was
refreshing
the
access
token,
when
this
one
expired.
So
this
was
already
taken
care
of
in
the
background,
but
when
you
are
in
the
situation
that
you
cannot
use
the
refresh
token
to
refresh
the
access
token
anymore,
you
had
to
yeah
basically
execute
the
captain
alt
command
manually
again
now,
with
this
pull
request,
this
happens
basically
automatically.
I
There's
one
I
would
say,
pick
up
implementation,
so
it
basically
works,
but
I
did
not
found
out
how
to
do
properly
and
it's
stated
here
in
a
command.
So
what
we
are
looking
here
is
actually
the
kind
of
factory
method
which
gives
us
back
ready
to
use
a
http
client
which
is
configured
to
use
oauth.
I
In
the
background
once,
however,
once
you
have
logged
in
using
oauth
and
or
open
id,
then
you
are
in
a
situation
where
you're
relying
on
a
valid
refresh
token.
If
that's
not
the
case,
I
didn't
found
a
way
to
verify
whether
the
refresh
token
is
actually
valid.
So
the
what
provides
providers
where
we
haven't
used,
don't
give
you
this
information.
Actually,
it
gives
you
the
information
for
the
access
token,
but
not
for
the
refresh
token.
I
I've
used
the
client
to
execute
a
http
head,
call
to
the
token
endpoint
which
was
used
for
for
getting
the
access
token,
and
if
this
fails,
we
can
just
assume
in
this
ui
code
here
that
there
might
be
something
wrong
with
the
reverse
token,
if
that's
the
case,
I
would
just
hit
the
word
applicator
again
calling
the
auth
method
with
all
the
client
values,
and
what
will
this
trigger
then
in
the
end,
is
it
will
yeah
again
open
the
local
web
browser
asks
you
for
credentials
et
cetera,
et
cetera,
and
you
will
end
up
with
a
fresh
refresh
token
and
access
token,
and
you
can
continue
to
to
use
the
client
as
usual.
I
It
works.
It's
kind
of
I'm
guessing
what
or
I'm
just
assuming
that
this
was.
The
error
could
also
be
another
error.
I
didn't
find
another
way
yet
room
for
improvement.
If
somebody
somebody
has
any
idea
how
to
do
it
properly.
I
also
researched
a
little
bit
online
and
there
also
just
was
the
suggestion
here.
I
Then
question
to
that,
if
not,
I
will
continue.
J
Can
I
jump
in
for
a
second
jump,
because
I
have
a
question
not
related
to
the
pull
request,
rotary
feature
but
very
close
to
the
topic
itself,
because
yesterday
I
was
playing
around
with
the
captain
auth
command
and
they
are
realized
that
we
have
the
flag,
minus
minus,
o
minus
client
secret.
I
No,
we
don't
support
it,
so
what
I've
did
while
implementing
the
authorization
code
flow,
is
to
try
it
in
the
end
out
without
any
traces,
so
google,
what
provider
or
microsoft
and
so
on,
and
what
I've
seen
is
that,
while
we
are
not
required
to
give
a
client
secret
to
our,
for
example,
liner
trace
sso
provider.
We
are
required,
for
example,
for
google,
so
the
workflow
did
just
not
work
without
giving
also
a
client
secret.
I
J
I
J
I
It
was
failing
one
out
of
50
times
or
so
I
just
did
an
experiment,
if
that's
the
case
by
just
executing
the
the
tests-
and
indeed
we
have
here
on
the
38th
run,
just
the
test
failure,
another
failure
in
the
distributor
code,
but
just
the
hiccup
in
the
test
testing
code
and
also
here
at
the
87th
run.
We
also
had
the
same
issue
in
the
end.
I
don't
even
know
what
was
the
fix
to
the
some
time
ago.
I
I
don't
know
it
makes
sense
to
take
a
look
at
that,
but
there
was
something
wrong
with
comparing
events
out
of
her
out
of
a
map
or
slice
which
had
different
orderings
from
run
to
run.
So
we
are
just.
I
We
just
assumed
that
the
ordering
will
be
the
same,
each
one,
which
was
not
the
case
from
time
to
time,
and
I
just
fix
that
and,
as
you
can
see
with
the
fixed
code,
then
all
of
the
runs
until
the
100s
did
pass
through
without
the
problem,
so
that
should
again
make
our
pipeline
a
little
bit
more
stable
because
we
don't
have
flaky
tests,
then,
which
is
not
yet
merged,
but
already
approved
by
florian
and
andre
was
the
issue
with
this
cupboard.
I
Also
on
the
distributor
side,
where
it
was
the
case
that
actually,
if
you
have
running
your
integrations
on
the
on
the
control
plane
and
you
restart
the
nuts
broker,
it
could
be
the
case
and
it
was
the
case.
We
discovered
that
the
hard
way
that
actually
the
nuts
client
connection
used
by
the
distributor
was
in
the
state
where
it
was
not
properly
connected
anymore.
It
was
in
a
closed
state
already
because
it
did
exceed
the
number
of
retry
attempts,
which
is
correct
me
from
wrong
flow.
I
But
I
think
it's
20
or
something
like
that,
but
ready
for
the
library
we
are
using
is
trying
to
reconnect
20
times.
If
you
are
having
having
a
downtime
which
exceeding
the
I
think
20
times
two
seconds
or
something
like
that
of
nuts,
then
this
connection
we
are
using
in
we're
caching
actually
in
the
distributor
code
somewhere
here,
pops
up
connections,
it's
just
a
string
of
topics.
It's
a
map
of
topic
to
actually
find
this
client
for
the
respective
con
for
the
respective
topic
used
is
not
usable
anymore.
I
I
Max
reconnects
-1
will
indefinitely
try
to
reconnect
until
it
has
a
connection
again.
So
fix
was
easy
and
the
bigger
part
is
here
actually
the
test
where
we
kind
of
I
indirectly
test
the
behind
the
behavior
of
the
library.
If
it's
behaving
correctly,
what
we
assumed
to
be
the
bug
or
not,
for
example,
here
we
adjust
for
a
specific
topic,
setting
the
max
reconnect
to
just
one
with
a
very
small
reconnect
weight
time
and
shutting
down
knots
for
that
time,
so
that
it's
exceeding
the
this
time
here.
I
So
we
are
dealing
now
really
with
a
with
a
client
that
lost
connection,
and
this
is
in
its
closed
state
does
not
use
level
usable
anymore,
and
we
we
then
later
on
just
verify
that
we
indeed
didn't
receive
an
event
using
that
specific
client,
but
we
received
an
event
that
was
not
altered.
That
way,
which
has
in
the
code
of
course,
the
minus
one
parameter,
which
means
reconnect
in
the
in
depth
indefinite
time
amount
of
time,
oh
nice,
that
we
could
test
it
using
a
unit
test,
not
an
integration
test.
I
C
A
K
Let
me
share
the
screen,
okay,
so
for
me
mostly
bug
fixes
for
the
bridge.
So,
for
example,
the
quick
filter
was
not
showing
all
elements
if
you
expand
it.
So
if
you
go
here
and
click
the
view
more
for
the
services,
for
example,
there
was
a
bug
that
there
were
two
scroll
bars
and
if
the
window
says
or
the
this
was
too
small,
the
second
scoreboard
disappeared
and
then
it
looked
like
yeah.
The
other
elements
are
not
here
and
yeah.
K
K
That's
for
a
quick
fighter
part.
The
next
thing
was
a
bug
that
we
had
some
duplicated
sequences.
K
So
if
a
sequence
was
selected
and
the
page
or
if
yeah,
if
it's
accessed
by
a
sequence,
the
page,
then
there
was
the
case
that
the
sequence
selected
here
was
also
at
the
bottom
here,
so
it
was
yeah
two
times
and
this
also
broke
the
show
older
sequences
functionality
because
it
took
the
latest
sequence,
and
this
was
the
duplicated
one
of
this
one.
So
this
is
also
fixed
by
now.
K
K
K
We
just
adjusted
this
message
here
that
it
doesn't
go
to
the
documentation
side.
Instead,
it
goes
to
the
service
creation
site
where
you
can
create
a
service
and
yeah.
If
you
create
a
service,
it'll
create
it
and
redirect
back
where
you
come
from,
so
the
environment
screen
and
yeah,
that's
basically
it
so
yeah,
that's
from
my
set.
So
then
they
will
hand
over
to
elizabeth.
L
Yeah,
I
will
keep
it
short
and
simple:
the
trigger
sequence
for
bridge
is
finally
merged.
We
put
in
a
lot
of
effort
in
pr
review
to
make
it
more
stable
and
removed
the
remaining
bugs
that
were
in
it.
So,
as
you
can
see,
there
were
lot
commits
added
and
comments
ended,
which
is
now
everything
is
fixed
and
yeah.
It's
ready
to
release,
so
it
will
be
in
the
next
release.
E
Thanks
yeah,
I
will
also
present
just
one
bug
fix
from
my
site
for
captain's
bridge
so
previously
in.
C
E
Evaluation
details
we
displaced
the
score
without
decimals,
and
that
happened
in
some
corner
cases
to
show,
for
example,
33
is
less
than
33,
but
actually
the
result
was
with
decimals
and
therefore
yeah.
We
extended
that
and
what
I
also
did
so
we
show
now
two
decimals
up
to
do
that.
Smells
here
in
the.
E
A
So
we
have
20
minutes
left,
let's
start
from
10
books
and
creates
10
minutes
for
each
project
and
then,
if
needed,
we
can
press
it
further.
Okay
and
yeah
thanks
to
everyone
for
the
demos,
so
you're
welcome
to
stay,
especially
if
you're
interested
in
backstage
integration,
because
I
think
it's
an
interesting
topic.
A
Okay,
so
maybe
you
would
like
to
briefly
summarize.
A
Your
current
research
and
what
other
obstacles
you
experienced.
G
Yeah,
so
I
did
like
investigation
on
backstage
like
what
it
does.
What
is
the
objective
of
that
application,
so
that
is
basically
a
like
cataloging
tool
for
all
the
micro
services
so
and
what
it
does
it
tries
to
solve
the
problem
of
a
developer.
Like
you
know,
we
in
a
microsoft
environment,
we
have
a
lot
of
teams
that
that
has
accountability
for
a
lot
of
different
components,
services.
G
So
what
backstage
does
it
helps
them
to
for
for
the
service
discovery
and
cataloging
them
on
a
a
single
website?
So
that
is
what
the
backstage
does
and
I'm
not
sure
like
like
what
we
want
to
like
want
to
achieve
with
the
character
integration
like
there
can
be
multiple
scenarios.
G
Like
one
thing
is
we
want
to
test
the
like
quality
gate
of
those
services,
the
other
other
thing
can
be
like
if
you
want
to
check
the
healthy
status
or
something
using
the
you
know,
the
multi
multi-stage
deployment
strategy
of
the
captain,
so
yeah.
A
Yeah,
so
actually
it's
a
good
question.
Maybe
it
makes
sense
to
clarify
how
google
summer
of
code
api
reads,
because
in
google
summer
of
code
the
approach
that
organizations
potential
mentors
provide
a
list
of
ideas
that
could
be
worked
on.
So,
for
example,
here
do
you
see
my
screen.
F
A
Can
see
so
captain
plugin
for
backstage?
So
basically,
this
is
initial
idea.
We
have
it's
quite
abstract
and
it's
quite
abstract
deliberately,
because
what
would
be
actually
done
is
to
be
defined
so
how
it
could
work
you
just
start
discussing
with
the
community.
You
do
your
own
research
and
come
up
with
the
proposal.
What
actually
you
would
like
to
address
in
your
project
here?
The
challenge
that,
to
some
extent
it's
already
happening
idea.
So
we
have
captains
and
box.
A
Sorry,
I
should
finally
disable
the
home
screen
in
internet
explorer,
so
here
there
is
a
backstage
plugin.
It
was
created
by
mauro
salah
about
one
month
ago,
but
after
that
the
project
wasn't
very
active,
but
it's
already
a
poc.
You
could
try
out
so
I
you
can
see
that
it
creates
some
integrations
some
dashboards,
but
obviously
there
is
a
lot
more
that
could
be
done
with
the
backstage
plugin.
A
So,
for
example,
me,
jorgen
and
andres
had
a
meeting
with
dmitry
maitin,
who
is
one
of
potential
mentors
in
this
project,
and
we
had
a
lot
of
discussions,
how
we
actually
provide
a
dashboard
for
blue
green
deployment
so
that
you
can
monitor
blue
green
deployment
status
on
multiple
environments
inside
backstage
and
they
rely
on
captain
as
an
engine.
So
I
asked
dmitry
to
share
this
example.
But
the
point
is
that,
even
if
you
have
some
dashboards,
there
is
a
lot
of
other
functionality
that
could
be
added
to
backstage.
A
So
my
suggestion
here
how
to
start
would
be
to
just
try
it
out
to
see
what's
there
to
explore
captain
use
cases
and
then
see
what
use
cases
you're
interested
in
and
basically
come
up
with
the
proposal.
How
you
could
extend
these
use
cases
better?
A
A
I
will
poke
people
about
that
again,
but
ultimately,
it's
up
to
you
to
come
up
what
other
areas
you
would
like
to
address
and
well.
This
is
how
gsoc
works,
and
I
hope
this
is
an
interesting
approach,
so
we
can
discuss
in
the
community,
so
you
mentioned
quality
keys.
It
could
be
one
of
the
options
managing
auto
remediation
is
also
an
option
just
visualizing
sli
slos,
maybe
with
some
analytics
graphing
integration
it
could
be
also
an
option.
A
So,
just
whatever
you're
interested
in
okay
and
yeah,
for
example,
one
of
items
which
could
be
considered
so
currently
there
is
a
pending
pull
request
for
manual
triggering
of
sequences,
and
I
guess
it
could
be
also
be
implemented
in
on
the
backstage
part.
In
theory,
who
knows.
G
A
So
easily,
how
would
happen
you
come
up
with
the
proposal,
then.
If
project
gets
accepted,
they
will
be
still
coming
in
the
bonding
phase
in
may
and
as
a
part
of
community
bonding
mentors
and
mentees,
and
the
community
actually
review
their
proposal
and
come
up
with
actual
plan.
A
A
A
A
A
Okay,
so
then
the
next
topic
rajiv,
would
you
like
to
take
over
asking
another
call.
A
Well,
the
piano
wanted
to
discuss
captain
documentation
sci
at
a
revamp
too,
so
we
can
still
briefly
discuss
this
topic.
A
A
I
will
just
briefly
go
through
this
project
id
and
discuss
way
via
okay,
so
we
have
new
documentation,
site
engine
project
idea,
and
I
think
it
basically
consists
of
two
parts.
So
one
part
is
the
engine
and
this
what
we
propose
for
google
for
google
summer
of
code
mike
actually
already
has
a
lot
of
feedback
about
the
current
implementation
so
mike,
if
you
want
to
summarize
your
experiences,
just
go
ahead.
C
Jensen
is
a
problem
and
I'm
amazing
it's
I.
I
don't
know
how
people
find
anything
in
there
frankly,
but
yeah
really
and
boy,
I'm
finding
captain
in
a
box
comes
up
at
the
top
of
every
search
at
google
and
youtube,
which
is
a
problem.
One
of
the
big
changes
that
I
would
like
to
that
I'm
proposing
is.
I
would
like
to
have
good
reference
pages
for
all
the
files
that
people
use
and
have
those
together
in
the
reference
section.
C
I
think
the
what's
there
for
at
least
structurally
for
commands
and
api
is
decent,
but
the
files
are
in
me's
and
matt
blobs,
which
I
I'm
I
spent
the
day
trying
to
figure
out
where
those
blob
files
are
etc.
But,
oh
that's
a
big
thing.
C
I
would
like
to
split
I
the
sort
of
the
getting
started,
the
installation
sort
of
I'd
like
to
have
one
document,
perhaps
with
an
accompanying
video
for
each
kubernetes
option
that
there
is-
and
this
is
how
you
install
it
and
then
have
the
tutorials
be
on
specific
topics:
quality
gates
to
to
measure
response
time
or
something
like
that
just
to
make
it
smaller.
I
looked
at
those
they
are
all
I
got
intimidated
when
I
started
trying
to
read
through
them
and
they
they
were
just.
C
C
There's
a
lot
of
places
where
just
an
additional
sentence
would
be
very,
very
helpful.
That
would
say
you
know
we're
going
to
populate
this
file,
which
does
such
and
such
at
the
beginning
of
the
tutorial
say.
What
are
we
trying
to
do
with
this
part
of
it?
C
So
I'm
looking
at
a
lot
of
that
there's
a
lot
of
really
good
information
stocks,
I'm
also
not
sure,
what's
happening
with
releases.
I
have
a
strong
hunch
that
we
are
just
copying,
doing
a
mass
copy
and
shoving
stuff
over
and
and
some
question,
I'm
I'm
not
sure
I
understand
in
the
current
docs
we
have
the
concepts
section.
C
We
have
the
quick
start
and
a
couple
of
other
things
that
are
not
really
specific,
and
then
we
have
all
the
stuff.
That's
really
specific.
C
That's
another
thing
with
the
reference
pages
definitely
want
those
two
specific,
because
each
reference
page
should
have
a
section
on
it
says:
differences
between
versions
also
for
commands
reference
pages-
I
I
you
know,
cut
my
eye
teeth
on
man
pages
in
unix,
but
they
should
be
copiously
complete
and
that's.
We
should
have
a
record
of
this.
C
C
C
We
want
it
all
in
our
documentation
set,
and
it
may
be
that
some
of
this
material
that
I
would
like
to
see
in
the
documentation
set,
because
it
would
just
look
nice
really
does
belong
in
readme
files,
etc
in
the
repos
close
to
where
it's
used
so
I'll
be
looking
for
a
lot
of
feedback
on
that,
and
please
feel
free
to
tell
me,
because
I
just
because
I
am
an
obnoxious
writer,
it
doesn't
mean
that
I
don't
know
about
the
problem
with
not
just
writers
who
try
to
do
too
much.
So.
A
Yeah,
so
we
had
the
kind
of
ad
hoc
knowledge
transfer
last
week
about
tutorials,
but
we
also
deep
dive
and
the
site
structure,
so
one
of
our
intents,
is
to
actually
introduce
the
real
documentations
code
when
all
documentation
is
basically
like
located
where
your
code
is
located
and
when
you
submit
the
pull
request,
you
can
also,
basically
you
edit
documentation
and
code
and
test
automation
in
the
same
repository
when
possible,
and
then
there
is
a
magic
tooling
which
actually
aggregates
this
documentation
and
for
this
magic
tooling,
we
are
currently
looking
at
entera.
A
So
it's
one
of
the
projects
focused
on
a
ski
doctor
that
actually
helps
a
lot
with
aggregating
documentation
from
multiple
repositories.
So
it's
actually
how
it
was
created.
So
you
can
reference
multiple
repositories,
versions,
tags
branches
from
there
and
then
as
a
tool
link.
It
can
just
aggregate
everything
it
can
also
do
preview
features.
It
can
also
do
staging
repositories
and
it
can
also
well
assuming
that
we
had
some
kind
of
identification
for
documentation.
We
can
also
hide
content,
have
feature
flags
for
the
content.
A
If
you
wish
but
yeah,
it's
definitely
not
something.
We
would
need
for.
Captain,
but
we
could,
for
example,
stage
unreleased
versions
and
have
a
flag
that
you
can
open
the
commentation
for
version
that
is
yet
to
be
released
and
we
can
create
this
documentation
incrementally
when
we
develop
new
features.
So
the
engine
is
quite
good.
A
It
adds
some
complexity
for
documentation
management,
but
I
think
for
the
content
it
would
be
much
better
and
for
gsoc
actually
implementing
this
engine
would
be
something
I
would
suggest
so,
basically
having
called
the
automation.
So
there
are
github
actions
for
enter.
We
will
also
likely
need
to
integrate
pandora
so
that
we
can
move
over
a
markdown
con
content.
A
So
it's
one
of
the
hollywoods
in
the
documentation
community.
What
you
use
either
markdown
or
ski
doctor
markdown-
is
really
simple.
Ascii
doctor
is
much
more
sophisticated
out
of
the
box,
so
it
supports
macros,
it
supports
includes
well,
it
supports
even
scripting
languages.
If
you
really
really
need
that
which
is
not
a
case.
D
C
Let's
not
scare
people,
because
I've
had
two
or
three
people
tell
me,
but
you
know
I
know
how
to
do
mark
down
in
that
ascii
doctor
for
doing
basic
coding
is
very
similar
to
markdown.
You
know
there's
a
few
things
that
are.
You
know
you
use,
although
I
think
I
think
you
said
antora
now
regulars,
but
you
use
two
equal
signs
for
a
second
level
heading
rather
than
two
hashtags,
but
but
for
most
of
the
basic
stuff
you
know
for
people
to
write
in
it.
It's
not
bad.
A
Right
and
we
will
still
need
a
kind
of
conversion
logic,
so,
for
example,
when
you're
moving
jenkins
wiki
to
ask
a
doctor
or
markdown,
we
created
the
converter
service
here.
We
would
also
need
some
kind
of
automation
so
that
we
won't
force
developers
to
actually
migrate
all
the
content
to
ask
your
doctor
to
have
this
documentation
published
as
used
to
be
so.
A
Hopefully
it
will
be
minimum
impact
on
capital
contributors,
but
we
might
move
some
content
over
especially
to
integration
services,
so,
for
example,
primitives
tutorials
would
likely
go
to
parameter
service,
etc,
etc
and
yeah.
It
would
require
some
collaboration
with
maintainers,
but
once
it's
done,
I
hope
it
will
be
a
much
better
experience
for
everyone
and
we
definitely
won't
need
to
maintain
so
many
versions
anymore.
A
M
Have
two
questions?
Actually
one
is
so.
I
guess
antora
supports
some
some
form
of
version
drop
down.
Then
right,
yes,
okay,
awesome
and
then
also
do
you
have
some
example:
documentation
page
that
uses
antara
that
we.
A
A
C
A
C
M
In
production,
basically
yeah.
C
References
one
of
the
problems
that
we've
got
right
now
is
the
the
internal
file.
Well,
not
the
internal.
It's
on
the
url,
the
file
name
for
all
those
files
has
the
release
in
it.
So
one
of
the
first
things
I
discovered
was
all
these
broken
references
because
they
were
going
to
release
0.8
and
I
think
that
just
got
pulled
down
pulled
out
of
the
set
and
torah
will
cross-reference
a
file.
C
So
if
we
did,
you
know
if
we
didn't
make
any
changes
still,
the
old
version
would
still
cross
reference
files
in
that
version
of
the
docs
and
the
new
one
would
cross-reference
the
new
version,
which
would
be
a
tremendous
help,
although
I
think
people
have
done
some
scripts
in
the
last
week
or
so
to
pick
up
some
of
that.
So
it's
not
so
bad,
but.
A
Yeah,
so
I
shared
a
link
to
the
public
demo,
so
you
can
see
that
there
is
antara
playbook.
So
basically
it's
a
configuration
file
which
actually
shows
how
to
arrange
the
content.
So
basically
it
grabs
the
things
together.
So,
for
example,
here
it
takes
two
components:
one
from
a
head,
another
one
from
multiple
branches
and
puts
them
in
dogs.
So
for
each
branch
it
will
have
an
entry
point
and
also
there
is
a
lot
of
things
plus
some
ui
bundles,
which
configures
the
interface
obviously
for
production
grade.
A
It
would
be
quite
a
long
yaml
file,
but
yeah
it
should
be
still
feasible.
So
I
hope
that
I
will
finally
publish
a
poc
I
started
working
with
and
the
my
pc
will
be
published
just
to
github
pages
with
some
github
actions.
So
even
if
the
engine
is
complex,
we
will
use
the
common
static
site
to
link
so
in
principle,.
M
Nice,
okay,
thanks:
I
need
to
leave
for
the
next
meeting
so.
A
Yeah
sorry,
for
going
over
time,
yeah
thanks
to
mike
for
deep
dive
and
thanks
to
everyone.
So
thanks
bye-bye,
hopefully
we'll
have
a
pure
season.
A
C
A
G
I
think
I'll
just
have
a
look
at
the
the
the
plugin
that
is
already
there
for
the
backstage
and
from
there
we
can
like
find
out
like
what
we
need
to
do
and
like
first
of
all,
like
we
can
break
down
the
functional
part
of
that
and
later
we
can
decide
how
to
implement
that.
G
A
A
Meanwhile,
I
will
try
to
push
in
the
working
group
about
putting
their
use
cases
as
github
issues
and,
if
needed,
we
can
organize
a
special
project
meeting
for
this
working
group
so
that
we
can
discuss
few
the
what
we
want
to
do,
because
I
already
have
a
quite
a
good
demo
by
dmitry,
but
I'm
just
not
sure
that
I'm
allowed
to
publish
it.
Maybe
he
would
like
to
present
on
his
own.