►
From YouTube: App Runtime Deployments Working Group [May 25, 2023]
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
So
from
last
meeting
the
infrastructure
pipeline
has
been
moved
to
the
runtime
CI
project
so
that
we
can
create.
B
A
Requests
more
easily
Carson
has
done
this
well,
no
big
change.
A
The
image
with
the
G
Cloud
CLI
had
to
be
updated
so
that
these
manage
gcp
tasks
work
again.
This
is
also
done
so.
Basically,
they've
moved
this
particular
image
to
a
new
repository
and
gave
it
a
new
name
and
yeah.
It
contains
the
latest
gcloud
CLI
and
jobs
work
again.
A
So
this
is.
If
you
want
to
have
external
and
external
blob
store
for
your
bubble
environment,
then
you
have
to
adapt
the
terraform
files
a
little
bit.
We
have
already
used
the
updated
files
for
our
environment
yeah,
so
this
is,
but
now
it's
also
in
in
Porsche
bootloader
yeah.
Then
maybe
this
point
Carson
brought
this
up
last
time.
Bosch
bootloader
uses
the
new
cost,
saving
VM
types,
but
the
Bosch
deployments,
don't
I
think
I
have
not
fully
understood
the
point,
or
maybe
there
is
none
there.
It
is
welcome
Carson.
A
A
Hey
welcome
just
in
time
we're
discussing
the
point
that
you
brought
up
last
time
that
Bosch
bootloader
uses
the
cost,
saving
VM
types,
but
the
Bosch
deployments.
Don't
I
may
not
have
got
this
correctly.
So
Bosch
bootloader
creates
a
cloud
config
and
I've
checked
the
history
for
gcp
and
AWS,
and
the
cloud
config
configs
have
already
been
updated
to
Modern
and
smaller
VM
sizes.
A
But
I
have
not
checked
the
exact
coast
of
the
ones
that
are
listed
here.
C
I
think
those
are
fine.
The
if
I
remember
correctly.
The
issue
is
that
the
these
changes
were
only
made
in
Bosch,
bootloader
and
Bosch
deployment
maintains
its
own
list
of
VM
types
that
are
still
all
like
the
old
old
kind.
We
for
the
most
part
it
doesn't
affect
us
I.
Think
the
only
one
that
we've
had
to
override
still
is
the
Bosch
light.
Bmw.
A
Yeah
that
one
okay
I
have
not
checked
that
one
yeah,
no
idea
where
the
cloud
config.
B
C
C
B
C
Sorry,
it
was
in
that
folder,
but
that
one
we
actually
do
use
and
have
to
override
manually.
B
A
A
Snitch
config,
Posh
library
and
type
okay,
but
now
I
see
yeah
okay.
So
indeed,
although
we
are
not
using
those
directly,
it
would
be
nice
if
someone
updated
to
more
recent
VM
types.
Okay,
but.
A
Other
issues-
okay,
but
I-
can
take
a
note
and
and
asking
the
selection.
If
someone
could
take
care
of
that.
Okay.
A
Good
yeah
experimental
setup
still
fails
to
automatically
deploy
the
Route
53
Resources
with
bubble,
so
in
in
total.
The
state
is
now
that
all
our
bubble
ends
are
up
and
running.
The
pipeline
is
running,
so
this
is
all
file
infrastructure.
The
automatic
update,
which
runs
every
Tuesday
is
also
fine,
except
for
our
single
AWS
environment.
Here
the
automatic
update
is
paused,
but
everything
else
is
running
again
and
here
we
are
struggling
Stephen.
A
No,
it's
not
in
the
call
we
are
struggling
with
here
is
coming
this
particular
template
and
or
for
the
experimental
environment.
I
worked
around
manually
by
deleting
things
that
cannot
be
deployed
from
the
template
and
fixing
bus
fixing
the
stuff
in
the
AWS
console
directly.
So
that's
fine
for
the
moment,
so
hi
Dave.
A
Let
me
show
what
this
is
about.
A
B
A
Need
and
what
is
really
confusing,
but
slowly
we
are
getting
some
clarity
on
it
is
that
we
have
here
two
scenarios
with
and
without
parent
Zone.
A
So
let
me
briefly
explain
I
think
in
the
other,
fully
requests,
I
commented
and
I
also
asked
our
routing
guys
for
help.
So
one
scenario
is
that
the
parent
zone
is
specified
in
our
case
or
or
the
performance
tests.
We
also
have
a
bubble
environment
that
would
be
bundle,
subcloud
IO,
and
if
you
configure
this
parent
Zone,
when
you
do
a
bubble
up,
then
it
does
not
create
a
new
Zone.
A
Instead,
it
adds
these
four
records
to
the
existing
Zone,
so
you
get
C
names
for
the
load
balancers
and
an
a
record
for
the
Porsche
instance.
A
So
one
can
do
this,
then
your
Zone
might
contain
a
lot
of
Records.
If
you
have
a
lot
of
bubble
environments
and
they
are
all
in
one
in
the
same
Zone,
but
this
is
working,
there
are
actually
no
disadvantages,
except
that
it
might
get
larger
over
time.
B
A
A
I.
Think
I
can
show
you
if
I
have
not
messed
up
everything.
The
tricky
thing
is,
it
would
get
a
separate
name
servers
which
you
have
to
copy
over
to
the
parents.
Zone.
D
Why
are
the
child
Zone
into
the
parent
Zone?
Yes,.
A
D
A
Yeah,
so
so
this
is
our
performance
test
set
up,
so
this
is
now
with
an
Own
Zone
and
you
get
a
set
of
name
servers
and
you
have
to
wire
those
to
the
parent
Zone.
That
is
copying
this
over.
A
It
looks
then
like
this,
so
here
is
a
new
record.
It
has
the
same
name.
So
it's
this
one,
and
here
we
have
the
four
name
servers
from
the
other
Zone,
and
the
point
here
is
Dominic
explained
that
from
the
routing
team.
If,
if
the
zones
are
managed
by
different
teams
using
different
credentials
and
users,
then
this
might
make
sense
right.
A
Maybe
if
yes,
we
have
that
too,
we
have
sometimes
a
mix
of
gcp
and
AWS.
Then.
A
Good,
this
automatic
wiring
is
not
yet
implemented
in
bubble
for
that
we
have
in
our
performance
tests
a
small
help
profile
like
this,
so
you
have
to
specify
the
parent's
own
ID.
B
A
Then
you
can
create
an
S
record.
The
record
I
just
showed
with
the
system
domain
name,
and
you
can
like
this
copy.
The
name
servers
from
the
yeah,
so
to
say,
the
output
of
of
the
bubble.
Terraform
stuff
works
for
us
is
not
yet
perfect,
but
now
before
fixing
everything,
we
should
first
try
to
fix
what
is
really
broken
at
the
moment.
Where
is
the
whole
I
want
to
show.
A
It's
this
one
so
where
I
really
got
yeah
severe
notices,
I
wanted
to
show
the
the
original
file
give
me
a
second
spoot
loader.
A
Aws
templates
yeah,
so
this
is
the
current
file.
What's
what's
really
causing
a
lot
of
headache
is
is
is,
for
example,
just
this
piece
of
code
here,
and
so
you
have
a
list
with
an
empty
string.
You
convert
it
to
a
list,
then
you
use
a
list
literal
again
and
then
you
convert,
convert
it
again
to
a
list
just
to
flatten
everything.
Later
again,.
A
And
there
are
a
few
problems
here.
One
is
I
discovered
that
so
this
is
a
ternary
operator
and
even
if
this
is
not
executed,
terraform
still
tries
to
access
this
piece
of
of
this
data
structure,
which
is
not
present,
and
so
it's
it's
quite.
D
A
Because
you
didn't,
maybe
it's
happened
here.
It
could
have
happened
here.
I,
don't
remember
exactly.
Okay.
D
A
D
Original
issue
we
talked
about
was
specifically
the
name
servers.
So
that's
why
I
was
a
little
confused
because
I'm
like
well.
If
it's
working
for
the
Zone
ID,
it
should
be
fine
for
the
name
servers,
but
if
it's
not
working
for
both,
then
that
would
make
more
sense.
Yes,.
A
A
So
what
somehow
worked
for
us
is
so
I
tried
to
to
to
throw
it
away
and
start
from
the
beginning.
A
So
what
seems
to
work
quite
nicely
if
you
just
copy
this
Tannery
operator
logic
again
here
to
these
local
variables?
So
if
we
do
not
have
a
parent
Zone
and
then
we
take
the
Zone
ID
from
the
new
Zone,
just
plain
and
simple
like
this,
and
if
there
is
a
parent
Zone,
we
take
it
from
the
existing
one
accessing
the
data
structure.
So
it
worked
for
both
cases.
A
But
I
can't
tell
for
sure
if
it's
now
100
correct,
if
you
start
from
fresh
again,
but
it's
at
least
it
looks
a
lot
easier
than
than
what
we
have
now,
also
trying
to
avoid
any
any
unnecessary
type
conversion
by
by
splitting
a
comma
separated
string
and
then
doing
a
list
from
that
again.
A
A
A
A
And
this
does
not
seem
to
be
necessary
because
because,
if
you
create
a
Zone,
you
automatically
get
an
NS
record.
To
my
knowledge.
A
A
So,
okay,
so
that's
what
I
would
like
to
ask
you
kindly
review
this?
Where
is
my
chat.
B
A
B
B
A
So
I
I
also
tried
yours,
but
with
this
one
but
didn't
accept,
then
I
think
one
of
those
was
NT
or
null
or
something.
But.
A
D
A
A
It
worked
for
both
cases
with
and
without
parent
Zone,
but
of
course,
as
I
messed,
a
lot
around
with
the
terraform
State
I
can't
tell
for
sure
would
have
to
clean
bubble,
destroy
again
and
then
Bubble
Up
times
to
see
if
this
is
really
working
this
year
with
the
one
in
the
iso
sequence,
which
seems
to
work,
fine,
uh-huh
and
and
this
I've
deleted,
because
it
was
already
there.
So
it's.
D
B
D
D
B
D
Just
says:
when
you
create
a
Zone,
you
must
you
need
to
create
an
NS
record
as
well
as
as
the
zone.
D
B
D
That
is
that
is
now
unnecessary,
which
is
it
makes
life
a
lot
easier.
Yeah
yeah,
yeah
I
mean
it's
entirely
possible
that
terraform
handling
these
kind
of
sort
of
conditional
resources
has
come
a
long
way
since
0.11
when
all
of
this
was
written
and
that
it
now
better
handles
this
sort
of
setup,
where
it's
not
necessary
to
mess
around
with
creating
lists
and
flattening
them
and
concatenate
them
and
picking
the
first
element
out
of
them
again
and
all
of
that
kind
of
nonsense.
B
D
D
So
that
that's
fine
I
think
that
would
be
a
much
cleaner
solution
if
it.
A
Came
that
way,
but
then
then
we
are
trying
to
verify
it
with
a
clean
bubble,
destroy
in
bubble
up
again
and
and
would
propose
this
as
final
nice
yeah.
B
A
D
Put
a
note
on
on
my
peer
about
you,
you
mentioned
having
a
an
example
plant
patch
for
wiring
in
the
the
new
Zone
into
the
parent,
Zone,
yes
and
I.
I
think
we
we
could
do
that.
D
It
would
be
nice
if
it
worked,
regardless
of
which,
as
you
were
using
and
the
other
challenge
there
is
that,
for
some
reason
the
presumably
different
people
wrote
them.
The
different
plan
patches
for
different
is
use
different
outputs
for
the
name.
Servers
that
get
that
you
need
to
wire
in
so
we'd
have
to
try
and
standardize
that
so
that
you
could
have
one
plan
patch
that
didn't
care
whether
it
was
gcp
or
AWS
or
or
Azure.
D
It
could
just
take
the
output
and
create
the
create
the
parent
zone
or
create
the
parent
record
for
you,
because
right
now,
there's
it
would
be
a
different
one
for
each
so
I'm,
not
sure
whether
that's
worth
undertaking
or
whether,
like
you
say
it's
just
here's
an
example.
If
you're
using
AWS
on
both
sides,
this
plan
patch
will
work
if
you're
using
some
other
combination,
you're
going
to
have
to
craft
your
own
yeah.
B
A
If
it
works
between
other
combinations,
yes,
yes,
okay,.
B
A
A
Yeah,
so
you
see,
this
is
the
template,
and
here
what
is
red
is
not
is
undefined,
but.
E
A
If
you
do
terraform
plan
and
applying
all
this
stuff.
B
A
A
D
A
Yeah,
it's
it's
I
I
borrowed
one
of
your
commits
and
then
put
one
sure
one
on
top
yeah.
A
Okay,
once
we
have
the
pull
requests,
they
will
have
to
build
a
new
bubble
release.
Then
we
can
cut
a
new
CF
deployment,
Concourse
tasks,
image
and
then
Stefan.
We
can
integrate
this
into
the
performance
tests
because
they
use
the
fixed
version
of
CF
deployments,
Concourse
tasks,
and
so
this
might
take
a
little
longer
until
we
have
a
final
release,
fully
released
solution.
A
A
On
DNS,
if
not,
then
we
move
to
the
next
point:
am
I
still
sharing
yeah
yeah,
okay,.
B
A
A
If
anyone
still
wants
to
use
the
deprecated
stack,
then
we
have
provided
this
large
ops
file,
which
adds
everything
you
need
back
again
and
yeah.
We
would
keep
this
Ops
file
as
long
as
there
are
updates
for
fs3,
which
is
right.
A
A
You're
good
was
a
bit
tricky
because
we
had
to
move
from.
A
Releases
to
Ops
releases:
this
is
now
where
the
fs3
update
is
happening,
and
then
you
don't
have
any
pre-compiled
releases,
so
we
had
to
move
it
out
from
this
use,
compiled
releases
Ops
file,
but
okay,
that's
not
much
of
a
concern
for
the
consumers
of
CF
deployment.
Pipelines
are
green
again,
so
everything
is
running
fine.
Now.
A
Good
yeah
any
questions
on
that.
No
that's
good!
Okay,
fs4
is
updated
very
frequently
and
yeah.
We
are
still
waiting
for
the
fs4
compatibility
release,
which
adds
Python
and
Ruby
again
yeah,
but
this
is
not
for
our
working
group.
A
Yes
and
I
I've
seen
one
or
two
people
also
complaining
about
this.
Certain
incompatible
change,
don't
know
from
where
they
are,
but
yes,
others
also
noticed
that
Python
and
Ruby
is
suddenly
missing.
So,
yes,
we
could
add
a
simple
Ops
file
and
that
replaces
the
default
fs4
with
the
compatibility
and
possibly
also
having
validating
that
one
in
yeah
one
of
our
scenarios
right.
You.
B
B
A
Good
but
right
now
there's
nothing
more
to
do
then.
The
last
point
from
my
side,
I
think
with
oh.
A
Cf
relent,
Concourse
I
think
it's
used
for
storing
the
compiled
releases
right
and
I
think
also
for
uploading,
Bosch
locks
of
failed
deployments
and
I
think
we
talked
about
this
already
one
time.
B
C
Think
that's
right,
I'm
not
sure
how
we
move
the
compiled
releases
over
successfully,
though.
A
It
was
something
or
Carson
you,
you
wanted
to
start
with
that,
and
then
it
wasn't.
It
proved
to
be
too
difficult
or
something.
C
I
guess
wasn't
sure
how
to
do
it,
because
we
okay,
we'd,
have
to
move
we'd,
have
to
shift
the
whole
pipeline
over
and
then
probably
at
the
same
time
shift
the
bucket
over
which
we
can
do.
We
just
have
to
be
really
quick
about
it,
because
we'll
lose
the
bucket
name
as
soon
as
we
unclaim
the
bucket.
C
So
it's
just
like
a
little
bit
of
like
a
risky
operation.
Also,
what.
C
Historical
releases,
we
should
be
able
to
copy
them
all
over
and
because
buckets
have
unique
names.
They
shouldn't
be
tied
to
oh.
B
B
D
B
D
D
C
D
D
C
But
either
way
the
The
Concourse
one
I
guess
we're
gonna
need
to
change
the
pipeline
and
the
folder
pretty
close
to
each
other.
So
maybe
pausing
the
pipelines
moving
the
folder
over
updating
the
pipeline
and
then
unpausing.
The
pipelines
makes
the
most
sense
or
having
the
update
to
the
pipeline
ready
to
go
then
pausing
the
pipeline
moving
it
all
over
and
then
pausing.
D
E
C
C
D
If
we
wanted
to
make
it
easier
to
switch,
we
could
do
that
with
a
different
variable
name
and
update
that
the
template,
with
a
different
variable
name-
I,
don't
know
whether
that's
easier
or
not,
but
yeah.
The
the
tricky
thing
is,
is
getting
the
bucket
moved
over
I'm,
assuming
that
Google
doesn't
have
any
nice
tools
for
transferring
buckets
to
different
projects.
D
C
They
didn't
last
time
they
didn't.
They
did
not
last
time.
I
did
this,
they
have
a
nice,
they
have
nice
tools
for
transferring
data
in
and
out
of
buckets,
but
if
you
want
to
actually
move
the
bucket
between
projects,
the
like
pathway
that
we've
been
following,
which
was
established
during
the
great
last
pass
issues,
was
rename
or
it
was
like
makeup-
make
a
backup
bucket
in
the
new
destination
transfer
all
the
data
to
the
backup
bucket,
delete
the
old
bucket
and
quickly
recreate
a
new
bucket
Within
within
the
new
folder
I.
C
Don't
think
you
can
even
rename
buckets
so
you
have
to
like
do
multiple
transfers,
which
kind
of
sucks
or
you
have
to
like
yeah
I,
think
I
think
you
have
to
do
multiple
transfers
from
what
I
remember.
C
We
can
have
a
go
at
it.
I
can
I
can
have
a
go
at
this
since
I've
done
it
a
couple.
A
Times
and
this
other
project
transferring
the
ownership
of
the
whole
project
is
that
is
that
an
option
or
is
there
something
else
inside.
C
C
A
The
same
name,
okay,
yeah,
but
but
if
we
yeah
I'm
good
I
mean
transferring
ownership
is
likely
challenging
yeah,
oh
okay,
good!
Fine,
then,
could
you
explore
the
possible
options
and
then
we
have
to
decide
which
one
to
take.
C
Sure,
oh
I
can
type
something
in
the
slack
about
the
possible
steps
we
could
take
and
we
can
always
find
time
to
pair
on
it.
If
you
want
as
well
yeah.
C
B
B
A
Oh
this
one:
okay,
no.
D
B
D
Yeah
I
was
just
reading
doc,
saying
we
should
probably
spin
up
a
gcp
VM
and
just
use
the
GS
util
tool
to
transfer
files.
So.
C
B
B
D
C
D
B
C
Oh,
it
looks
like
some
of
the
branch
steel,
Branch
jobs
have
been
failing
just
because
there's
some
protected
yeah.
C
I
I
might
be
able
to
delete
those
actually,
but
there's
the
update
releases,
stale
Branch
one
which
should
manually
delete
stuff
kind.
A
Of
delete
stuff
because
it's
protected
this
one.
A
A
Indeed,
yeah
I
noticed
this
quite
some
time
ago,
but
put
the
topic
back
for
now.
Yes,
it
should
clean
up
stale
branches,
but
it's
not
compatible
with
the
current
Branch
protection
rules.
Interesting,
okay,.
E
Maybe
it's
a
leftover
from
the
old
punch
protection
I
mean
we
had
this
back
in
the
branch
protection
with
it
found
every
branch
that
had
a
v
inside
but
that's
fixed.
Meanwhile,
that
may
mean
you
just
go:
Chapman
writes
once
over
those
branches
removes
protection
and
from
then
on,
it
should
work,
but
there
should
be
no
new
branches
anymore
get
protected
unless
they
are
the
main
branch
and
real
version
branches.
B
A
A
E
E
Yeah
yeah,
it
was
not
perfect.
E
E
Was
there's
a
regular
expression
to
find
version
branches
and
it's
V
and
then
a
number
of
arbitrary
links?
Number
even
zero
is
a
lot
okay.
So
far,
so
good
but
match
was
not
a
full
match
so
from
the
beginning
to
the
end
of
the
line,
but
just
find
it
in
the
string,
and
that
was
a
misinterpretation,
because
the
the
documentation
of
this
Branch
protector
tool
is
not
that.
A
A
A
Delete
yeah
rule,
okay,.
E
Okay,
yeah,
we
need
main
and
develop
and
release
candidates.
This
is
something
yeah
you
you
might
want
to
have
that's
good.