►
From YouTube: Kyma Prow Migration WG meeting 20181026
Description
Meeting notes: https://docs.google.com/document/d/1ljEAoCBJXlxx_ATPyvKZ1KoyFOSIBzEAOkN-2H-HhUY/
A
A
A
A
A
C
B
C
Yes,
we
can
see
it.
Okay,.
B
So
I
would
like
to
present
you
the
status
of
the
progress
in
the
test
in
the
repository
and
during
this
week.
So
we
march
for
pull,
request
and
first
is
about
reading
city
secrets
from
duteous
installation
scripts
then
run
integration
tests
on
virtual
machine.
This
will
be
presented.
Balanced
women
next
issue
was
about
configuration,
error,
cancellations,
flag
and
the
last
one
generating
minimal
configure
a
config
file.
We
closed
six
issues,
so
two
of
them
were
closed
without
providing
full
request.
B
One
was
about
investigating
how
to
provide
source
code
for
jobs,
because
it
was
done
in
the
mija
who
did
ask
when
he
defined
I
plan
for
component
and
also
create
a
bit
back
docker
file.
It
was
also
done
in
the
anyhow
request:
let's
have
a
look
on
our
board,
and
so,
as
you
can
see
currently
in
the
review,
we
have
documentation
for
pro
secret
management.
Next,
one
is
immigration,
tokamak
components
to
pro.
So
it's
very
important
issue
and
we
have
some
discussions
there.
So
please
also
have
a
look
on
that.
B
B
B
Okay
and
the
next
one
in
review
is
automatic,
update
of
pro
after
merge
to
the
master
branch
and
in
progress.
We
have
issue
about
pro
trade
modeling,
which
will
be
discussed
today
by
Andres
provisioning,
a
cluster
for
integration
test
and
next
running
FEMA
integrations
activity,
because
currently,
what
was
done
by
Suleiman?
We
trigger
that
every
time,
and
here
we
want
to
achieve
the
same
configuration
and
as
we
have
in
the
internal
CIA
and
next
issue
is
about
defining
strategy
for
organizing
jobs.
In
the
confusion.
B
Currently,
we
are
using
only
one
M
file,
but
it
will
be
bigger
and
the
bigger.
So
we
decided
to
define
strategy
and
the
last
thing
is
about
providing
secrets
for
a
beautiful
machine
and
we
also
read
a
prioritize
a
little
bit
our
backlog.
So
we
have
two
tasks:
block
right
now:
prepare
post
submits
for
components,
it's
blocked
by
this
issue,
about
immigration
document
components
and
prepare
documentation
for
that
and
on
the
top
of
the
to-do
list.
B
We
have
static
analysis
of
fish
or
shell
scripts
and
defining
job
for
testing
for
repository,
because
currently
we
only
require
approval
for
from
code
owners
and
that's
all
all
no
more
checks.
So
here
we
want
to
define
some
of
them,
and
next
one
is
about
deploying
double
networks
which
allow
us
to
click
in
the
github
two
to
get
to
the
pro
and
dashboard,
and
next
one
is
about
integration
providing
logs
from
integration
jobs.
B
So
when
you
want
to
start
working
on
the
pro,
please
take
issues
from
the
top
of
that
list,
and
the
last
thing
I
would
like
to
say
is
that
recently
probably
noticed
that
we
start
a
new
meeting
called
contributor
syncs
meetings,
so
we
meet
every
Tuesday
and
Thursday
at
8
a.m.
and
it's
very
short
meeting.
It
looks
similar
to
scrum
stand-up
meeting
when
we
ask
questions
what
was
done,
what
we
plan
to
do
and
we
discuss
problems
and
obstacles.
So
if
you
participate
in
our
work,
please
attend
that
meeting.
Also,
that's
all
thank
you.
A
D
Does
it
draw
the
question?
Do
we
have
a
documentation
to
please
with
you
which
you
can
start
work
on
right
now
because,
for
example,
I
prepared
one
task
for
this:
how
to
create
pipeline
job
documentation,
which
is
already
blocked
by
my
paw
request?
But
do
we
have
something
what
you
can
start
work
right
now,
because
battery's
prepared
good
documentation?
For
example,
we
are
using
preset
or
the
decorator
from
Pro,
which
is
not
documented
and
other
people
may
don't
know
how
to
use
it.
D
E
Generally,
we
need
to
come
up
with
a
documentation
structure.
I
mean
we
need
to
describe
our
setup
in
general
so
that
we
as
a
team,
understand
how
how
to
operate
it
and
so
on,
and
the
other
thing
is
a
usage
so
how
people
can
activate
now
a
new
pipelines
for
repositories
or
that
we
had
the
typical
flow
speed.
I
think
we
need
to
talk
him
in
and
yeah.
B
D
E
Page
M,
so
I'm
new
to
the
working
group
and
I
still
want
to
figure
out
a
place
to
think
so.
That's
not
currently
I'm
shared
with
the
public,
but
I
would
would
to
do
today,
probably
I
like
to
just
summarize
in
roughly
what
we
did
and
the
threat
modeling
for
now
and
what
outcomes
are
and
further
steps.
Armed
and
generally,
we
had
two
session,
so
this
yeah
and
Suleiman,
and
and
we
are
just
drawing
the
landscape
which
we
are
targeting
to
so
the
assumption
was
that
we
have
one
production
cluster
for
now.
E
I
think
it's
important.
If
you
plan
to
have
more
clusters
for
I,
don't
know
testing
or
whatever,
then
probably
the
same
outcomes
should
be
applied
yeah.
So
for
now
we
just
jump
draw
out
the
landscape,
meaning
of
what
what
systems
we
have.
What
are
the
agents
inside
the
systems
and
then
we
have
end
over
the
communication
channels
and
discussed
on
others
huge,
irrelevant
topics.
For
now
what
is
we
will
still
have
at
least
one
other
session
where
we
want
to
go
over
the
agents
itself
now
and
discuss
potential
threats
on
that
agent.
E
So
it's
not
finished
yet
and
as
result
just
so
that
picture
I
have.
Of
course
we
had
a
big
picture
on
the
wall
and
now
I
try
to
transfer
it
into
it
and
electronic
diagram,
and
there
that's
how
it
looks
like
so
that
we
have
that
our
production
cluster
on
on
the
Google
cloud
platform
provision
with
the
internal
components.
E
We
have
the
Google
cloud
platform
services
which
we
are
using
to
like
the
key
management
service
for
the
secret,
storing
then
the
cloud
storage
using
the
cloud
storage
for
archive
the
archiving,
the
locks
having
the
computer
in
kubernetes
engine
for
our
integration
test
to
spin
up
clusters
or
VMs
dynamically.
We
have
two
cloud
registry
to
release
our
docker
images
and
use
existing
images
from
them
and
on
the
other
side
here
we
have
some
github
integration.
E
So
the
event
notification
on
one
side,
status,
check
of
the
jobs,
publishing
it
to
github
and
other
integrations
via
the
plugins
to
changing
the
labels
of
a
query
request,
for
example,
and
other
tooling.
So
if
you,
if
you
can
think
of
other
dependencies,
which
we
have
just
now
feel
free,
then
to
contribute
to
that
diagram.
That
we
are
aware
of,
but
you
are
more
important-
is
from
the
outcomes
perspective.
So
for
now
we
identified
no
block
us
so
that
no
something
cannot
be
designed
in
a
way
that
we
cannot
go
live
with
the
solution.
E
E
In
general,
we
we
said
that
we
are
just
any
external
communication
around
our
cluster
should
be
really
yeah
that
must
be
secured,
but
it's
also
acceptable
for
now.
That's
internal
communication
is
not
secured,
but
V
it's
recommended
to
achieve
it
and
we
should
address
it
after
we
have
our
first
installation
in
place.
E
We
discussed
that
planned
secret
mass.
We
could
secret
marriage
meant
using
the
key
management
service.
That
was
fine.
We
agreed
on
that.
We
probably
should
have
more
technical
service
accounts
in
our
Google
account
and
really
a
royal
technical
role
which
we
have
so
that
we
have
one
service
accountability
for
publishing
the
release.
E
Imagist
was
a
cloud
registry
and
one,
for
example,
form
for
the
provisioning
phase
using
the
key
management
service,
and
what
is
not
really
finished
yet
yet
is
was
a
topic
of
how
secure
that
clusters
VMs,
which
we
are
spinning
up
dynamically
for
integration
tests,
must
be
so
having
in
mind
that
at
Kemah
cluster
itself
has
maybe
unsecured
endpoints,
because
we
all
want
to
test
the
feature
there.
Is
that
acceptable
or
not?
So
that
was
in
discussion.
E
C
C
You
can
use
this
if
we
are
using
this
configuration.
You
know
it's
already
a
part
of
its.
You
can
see
it's
here.
Actually,
as
Adam
said,
it's
just
a
press-up
me
test
right
now,
so
it
will
run
for
each
PR,
but
later
on,
we
have
a
follow
up
task
to
discuss
how
we
are
going
to
do
it
in
the
production
environments
like
a
possible
job
or
how
we
are
going
to
chain
them
together
and
how
we
manage
to
do
that.
C
Well,
first
from
the
probe
job,
it
runs
this
script,
which
is
authenticated
authenticating
with
the
Google
cloud
first
and
then
it's
creating
an
instance
on
the
fly
right
from
here
and
then
it's
just
copying
over
one
installation:
script
to
name
the
created
VM
and
after
that
is
running
this
installation
script
and
finish
it's
just
removing
to
the
newly
created
VM
so
that
there
wouldn't
be
any
left
over
yams
and
inside
the
VM.
This
is
the
script
which
is
being
run.
It's
basically
installs
all
the
needed
dependencies.
C
C
Before
before
doing
that
before
configuring,
the
job
you
just
need
to
create
a
secret
created,
kubernetes
secret
out
of
Google
cloud
service
accounts
so
that
this
job
can
use
this
service
account
to
communicate
with
the
Google
cloud
API.
So
here
you
can
find
steps
to
do
that
here
under
integration,
job
folder-
and
there
is
one
more
topic
actually
am
a
follow
up
task
is
using
this
preemptable
VM
instances.
C
That's
also
topic
I
want
to
discuss
here.
These
are
just
like
regular
VM
instances,
but
Google
Cloud
can
delete
them
anytime.
It
they
needs
resources
so
well.
The
good
thing
is
they
are
costing
really
less
than
the
regular
VM
instances.
Actually
it's
like
one
fifth
of
the
cost
of
the
regular
cost,
but
the
downside
is
they
can
just
wait
at
any
time.
They
need
it.
While
the
documentation
says
the
ratio
is
about
5%
to
15%,
so
it's
like
every
one
job
in
10,
so
it's
not
too
bad
and
also
they
they
are.
C
A
G
B
B
C
Been
testing
it
from
yesterday
it
depends
on
the
zone.
Actually
in
the
first
one
I
was
trying
and
they
didn't
have
any
preemptable
instance.
That's
why
it
was
failing,
and
then
it
was
trying
to
create
a
regular
BIA,
but
I
changed
to
another
zone.
I
think
it
was
West,
3b
or
something
and
they're
just
deletes
it's
Iran's
one
job
in
every
five
jobs,
so
it
was
around
20%
for
me,
I
think
it's
still.
Ok,
all
the
integration
tests
is.
C
Configurable,
we
just
need
to
change
how
how
we
are
creating
the
instance.
We
need
to
enable
or
say
well
it's
from
there.
E
F
C
F
That
mean
that
if
you
trigger
a
complete,
suet
or
I
know
the
food
that
we
consider
at
some
point,
it's
going
to
fail
completely
in
order
to
what
you
want
to
make
it
well
to
retry
in
this
particular
case,
or
maybe
how
to
avoid
that
there
is
a
broken
integration
four-and-twenty,
whatever
tests.
Only
because
this
happens
can
we
retry
automatically
somehow
in
your
procedure,
I.
F
E
F
C
F
So,
at
least
from
my
point
of
view,
only
thing
you
might
want
to
check
before
you
go
with
these
approaches
that
you're
able
from
the
pro
tool
to
identify
this
situation
and
to
handle
it
without
requiring
human
action,
because,
if
you
say
one
out
of
five
or
whatever
one
out
of
ten
even
is
going
to
fail
and
I
trigger
a
suite
to
type
tunnel.
Ten
integration
tests-
I
don't
want
in
the
middle
of
these
two
to
have
to
take
many
actions
because
some
VM
got
kicked
out.
F
F
This
is
the
pram
civil
one.
Yes,
but
the
important
thing
is,
like
you,
said:
you're
going
to
spin
up
a
VM,
use
it
for
half
an
hour
and
then
throw
it
away,
anyways
right.
So
it's
what
you
experienced
from
my
point
of
view
would
make
a
lot
of
sense
if
you
say
have
a
cost
with
ten
thousand
nodes
and
I
don't
care
if,
like
ten
percent
of
them
are
gone
half
of
the
time
over
the
year,
but
I
have
saved
like
final
50
percent
or
like
save
on
more
than
50
percent
out
of
the
cost.
F
E
F
Think
that
is
very
good.
The
only
thing
is
we
have
to
be
careful
to
be
able
to
handle
this
situation
without
actually
crushing
the
pipoint
without
actual,
but
that's
that's
what
I
think
we
should
think
about.
If
we
can
do
that,
fine,
if
no
I
would
really
recommend.
We
don't
do
that,
because
then
you're
going
to
have
to
put
a
lot
of
manual
work
on
some
person
to
troubleshoot
that
and
that
will
probably
cost
more
than
this
I.
Don't
know
five
ten
cents
that
we
are
safe
for
execution.
So
that's
my
only
we.
B
A
D
Okay,
right
now
I'm
a
little
confused
who
should
act
about
our
tasks
and,
as
you
could
see,
who
has
a
lots
of
tasks?
Okay,
right
now
for
issues
in
to
accept
column,
because
it
is
confusing
when
we
start
working
on
the
some
tasks
we
are
taking
into
our
team
sprint
or
for
another
two
o'clock,
and
then
with
we
made
the
review
from
people
from
working
group.
And
then
we
don't
know
who
should
accept
at
that
task
and
also
with
with
who
we
should
discuss
the
details
of
tasks
and
that's
a
problem.
D
D
B
I
will
see
how
it's
in
our
team
so
I'm
moving
it
to
accept
and
I'm
waiting
for
my
product
owner
to
to
move
it
to
closed
and
currently
I'm
most
of
the
tasks
in
these
working
groups.
So
I'm
trying
to
be
up-to-date
with
all
the
tasks
which
are
kind
currently
in
progress
and
also,
if
you
have
any
questions
of
you,
you
can
ask
on
instruction
or
you
can
approach
me.
But
if
you
have
any
other
opinions,
we
can
discuss
it
so.
B
E
That
working
group
should
decide
on
what
is
really
needed
to
fulfill
the
task
and
then
also
to
accept
it.
So
maybe
we
can
say
that
this
group
of
working
group
leaders
and
the
backlog
refinement
meeting
or
whatever
it's
doing
that,
but
as
involving
people
from
outside,
we
are
not
in
the
topic.
I
think
it's
not
cannot
be
the
solution.
B
C
D
So,
as
Alan
mentioned,
we
already
set
up
the
let
free
production
Pro,
which
is
available
on
page
status,
beat
cannot
rely
on.
It
has
an
eye
that
has
certificate
saw
the
connection
to
the
github
is
secure
and
we
setup
it
for
testing
our
Pro,
it's
Stabler
and
it
will
no
work
properly
with
crema
and
viral
crema
project
organization.
D
B
E
As
you
discussed
last
time
about
this
orchestration,
so
we
will
end
up
with
probably
20
30
drops
and
I'm
just
afraid
that
this
will
confuse
for
now
as
long
as
that
is
really
not
introduced,
and
so
that
there's
a
people-
and
maybe
so,
we
really
should
work
on.
Oh,
maybe
as
part
of
the
chemo
incubator
organization.
For
now.
E
B
D
Right,
yes,
because
we
need
to
that,
may
merge
it
to
the
chemo
project,
because
I
mean
everyone
has
different
forked,
that's
okay!
That
right.
E
B
Maybe
I
would
say
why
I
like
to
run
our
current
row
against
original
kima
project,
because
we
can
check,
for
example,
how
take
performance
of
our
pro
cluster,
for
example,
provisioning
these
virtual
machines.
Maybe
we
can,
we
will
have
some
problems
later
when
we
provision
I,
don't
know
ten
at
the
same
time
so
and
I
think
it's
also
good
good
good
place
to
notice
such
kind
of
problems.
We
can
also
remove
these
notifications,
as
you
can
see
here,
to
do
not
confuse
people,
but
I
would
like
to
run
it
against
original.
B
E
D
C
It
was
just
a
question
actually,
as
you
know,
we
have
All
Saints
Day
next
Thursday.
That's
why
I'm
most
of
the
people
won't
be
in
the
office
on
Friday,
so
I
didn't
think.
Moving
to
next
week's
meetings
to
the
other
Monday.