►
Description
Decision on how to parallelize the work
A
A
Except
when
you
start
thinking
about
security-
and
he
says,
and
then
it
starts
to
feel
like
it's
gonna
get
complicated
like
I'm
thinking
about
how
this
version
mapper
will
work
the
security-
maybe
maybe
it's
not
as
complicated
as
I,
think
it'll
be.
But
that's
the
main
concern
I
have
right
now
is
just
you
know,
we're
going
to
come
up
with
a
scheme
here
and
then
we're
gonna
have
to
kind
of
figure
out
how
it's
going
to
work
with
security
releases
as
well.
So.
A
B
B
So
if
we
go
down
the
route
of
what
we
discussed
here
with
a
version
mapper,
if
we
have
someone
merged
something
and
that
gets
tagged
and
then
we
have
a
security,
merge
come
in
and
some
tags
the
security
version,
which
version
do
they
tag
and
do
they
stop
the
next
version
from
bumping
on
public?
So,
for
example,
if
someone
just
merged
something
in
how
do
we
ensure
that
version
1,
one
that
was
previous.
B
B
B
The
problem
that
every
component
suffer
every
month
when
there's
a
security
release,
they
have
the
tag
freezing
on
public
because
they
start
tag
because
they
talk
in
advance
because
they
are
not
integrated
with
the
with
the
tooling.
So
they
have
to
pre
tag
the
security
fixes,
and
so
they
are
in
this
frozen
yeah.
This
still
brings
us
to
the
idea
that
we
should
drop
cember,
because
we
don't
need
it.
B
It's
it's
a
problem
and
yeah
here
it
but
I
think
I
was
not
really
scared
by
this,
with
a
security
release,
but
to
another
details
which
is
we
release
security
fixes,
which
is
the
latest
known
target
version,
plus
only
security
fixes.
This
doesn't
play
well
in
the
outer
loop
employee
word,
because
it's
it's
a
kind
of
stop
the
word
and
just
do
security
fixes.
B
We
can
use
the
new
images
and
things
like
that,
and
everything
is
more
or
less
like
the
previous
one,
but
my
idea
was
that
we
were
still
peeking
the
latest
development
plus
the
fixes,
but
this
will
end
up
that
when,
when
today
the
release
camps,
we
never
run
that
commits
at
that
that
version
of
github
in
production.
We
will
never
run
that
that
is
true
and
I,
don't
know
both
John's.
B
Confirming
unless
you're
the
that's
the
thing
where
I
think
this
is
not
gonna,
be
that
big
of
a
deal
for
this
process
specifically,
is
that
yes
you're
right?
It's
never
gonna
be
that
commit
running
in
production,
but
that
is
the
case
right
now,
even
because
even
with
we
stopped
the
world
thing
auto
deploy
is
not
stopping
so
that
continues.
So
we
are
back
into
this,
the
same
state
so
with
the
work
that
is
being
done
with
the
security
development
security
releases.
B
The
point
here
right,
like
security
releases,
are
always
going
to
go
into
the
auto,
deploy,
regardless
of
the
self
managed
release
for
calm
and
then
we'll
have
a
way
to
back
porch
things
into
a
stable
branch
and
then
with
a
click
or
command
tag,
so
that
part
I
would
basically
remove
from
this
discussion,
because
at
that
point
we
can
tag
whatever
version.
We
decide
whatever
version
scream
scheme,
we
decided
that
moment.
B
C
B
Need-
and
you
have
the
main
measure
and
minor
if
you
want
to
give
them
because
I'm
not
really
inclined
to
keep
all
of
them,
but
let's
say
we
keep
them.
Then
you
have
this
rough
idea
of
where
you
are
in
the
history,
because
it's
still
12
5
but
then
there's
a
shot.
I
mean
it's
some
somewhere
in
12
5
development,
some.
So
if
we
take
a
look
at
the
schema
that
was
written
in
the
previous
call
and
we
look
bullet
point
for
a
I
3,
so
get
Li
I'm
highlighting
it
right
now.
B
This
would
not
be
1
to
12,
or
rather
this
would
be
the
final
text
version,
but
we
would
be
tagging
it
on
top
of
1
to
12
and
then
shop,
for
example,
it's
not
proper
somewhere.
It
still
gives
you
an
idea
of
what
we
are
in
like
working
towards
and
the
ABCD
is
the
shot
that
we
are
tagging
on
top,
which
means
we
kind
of
fix
the
clash
problem
as
well,
not
really
because
they
can
still
push
the
tag
and
expose
previous
versions.
I.
Think.
A
A
B
Today,
we
merged
the
fix
that,
in
the
admin
panel
will
show
you
an
up-to-date
literally
even
if
it's
not
tagged
just
if
the
sham
matches
it
would
show
us
up
to
date
so
that
we
untangle
the
rings
of
tagging
but
yeah
they.
They
do
something
like
this
today.
It's
not
that,
but
it's
kind
of
number
of
commits
you
are
away
from
the
from
the
real
tag.
Then
there's
a
G
which
they
forget,
and
then
you
have
the
chef.
This
is
what
I
do.
What
is
not
a
good
one,
13
to
10
minutes?
B
You
are
tilting
commits
away
from
that
tag.
We
we
that
in
omnibus
as
well,
if
you
want
to
use
that
we
have
this
exam
same
thing.
If
you,
if
we
need
it
Yallah,
does
it
get
described,
yeah
yeah,
so
this
is
what
they
are
using.
I
I
think
that
we
should
do
something
like
this.
You
are
not
allowed
to
tag
the
minor,
its
release,
dual
business.
B
That's
actually
a
interesting
idea,
so
you're
always
going
towards
zero
yeah
you're,
always
going
toward
zero
of
the
next
one,
and
then
we
can
selectively
pick
commits
that
are
newer
and
tag
them
independently
from
it's
actually
interesting
idea.
Yeah,
we
kind
of
keep
the
account
we
know
which,
which
is
the
last
things
that
we
tagged
haven't
say
now
we're
going
to
build
something,
but
this
will
break
that
idea
that
we
had,
which
is
team
tax
themselves,
what
they
want,
and
then
we
pick
up
the
SAG.
B
B
Let's
rope
this
back
into
the
discussion
about
how
are
we
migrating
components
to
kubernetes
and
see
how
this
fits
so
previous
time,
we
talked
about
a
possible
opportunity
for
us
to
not
do
any
work
and
have
the
platform
Auto
upgrade
itself.
So
every
time
developers
merge
something
into
master,
it
would
be
picked
by
the
platform
and
images
would
be
replaced
and,
if
necessary,
charts
would
be
replaced
independently
with
the
moving
part
of
this
version
mapper
as
well.
B
B
A
B
A
Right
I
mean
I,
see
that
big
impact
will
be
to
Omnibus
and
potentially
charts,
but
to
give
up
components.
I
don't
know
like
if
we
don't
care
whether
we're
taking
a
tag
or
a
sha,
then
it's
not
much
different
than
what
we
have
now,
where
we're
updating
versions
and
they
give
that
code
base
where
we're
updating
version
files
right.
It's
just
moving
update
to
this
central
yeah,
no
yeah.
B
B
A
B
I
mean
we
might
need
to
think
about
a
different
system,
because
there
is
another
thing
that
I
didn't
mention:
to
try
to
prevent
complexity,
but
we've
trained
the
whole
community
that,
like
we
still
have
people
who
are
installing
from
source.
So
what
they
need
to
do
is
just
pull
that
one
repository
and
they'll
get
all
the
versions.
If
we
do
that,
we
need
another
system
that
will
get
them
those
versions.
And
then
we
are
back
to
the
story.
You
had
on
Friday
about
services,
about
a
Verizon
service,
which
is
something
we
like.
A
B
Don't
we
keep
the
gitlab
as
a
source
of
truth,
because
it's
a
it's
already
there
right.
So
all
the
version
are
already
there.
Why
don't
we
use
that
one,
because
what
would
we
need
to
do?
We
would
need
to
like
get
release
tools
to
clone
this
humongous
project
that
keeps
moving
and
we'll
have
to
like
I
just
there
are
so
many
problems
that
that
are
like
flooding
to
my
head
right
now
that
I
would
like
avoid
using
it
lobbies
so,
like
github
rails,
there's
a
source
of
truth
for
anything
like.
B
Dependency
there
that
where
we
say
that
it
lab
rails
is
a
source
of
truth
for
everything,
but
it's
one
of
the
components
that
makes
up
a
bigger
ecosystem.
So
if
on
the
bus
were
came
first
or
came
in
time,
that
should
have
been
our
social
truth
to
begin
with
and
not
rails
yeah.
But
now
we
have
charts
that
are
aside
from
from
omnibus.
So
the
thing
could
be
it's
the
same.
It's
the
same
they're
encompassing
the
whole
ecosystem
that
we
have
built
so,
whether
you
call
it
charts
or
omnibus
it
doesn't
matter.
B
They
both
need
to
draw
from
a
source
of
truth,
that
is
on
a
higher
level
than
github
rails.
In
my
opinion,
yeah
yeah,
that's
what
you
mean:
this-this-this
willing
it's
fine,
but
this
would
means
that
then
also
to
get
Rob
rates
needs
to
check
that
source
because
it
need
those
information
forces
installation
for
running
tests
for
whatever.
So
it's
a
dependency
that
we
can
move
to
the
upper
level
and
say,
release
suits
or
whatever
and
having
this
single
source
of
truth.
But.
A
A
B
A
A
B
Idea
that
we
had
when
we
are
discussing
component
of
the
normative
component
was
that
we
just
ride
master
in
all
of
the
version
file
on
master.
It
was
the
biggest
okay,
so
it's
always
from
master.
So
you
can
write
whatever
in
advance,
because
some
problem
is
that
only
usually
has
real
content
test.
So
it
would
be
the
only
one
with
this
kind
of
security
mechanism
and
then,
when
you
release,
you've
read
their
version
that
you
were
using
for
that
release
because
it's
D,
but
then
you
lose.
B
A
B
B
B
But
I
do
agree,
but
we
didn't
tradition
of
introduction
of
the
charts.
It's
the
perfect
example
which
we
don't
want
to
have
in
charge,
the
same
logic
and
the
same
version
and
community
that
we
have
an
omnibus.
This
is
yet
we
are
building
on
here.
Okay,
and
now
we
want,
we
need
something
on
top
of
that.
B
It
has
the
throat
which,
which
is
fine,
I
mean
it's
okay,
but
it's
it's
more
something
that
we
need
for
our
own
case
than
what
we
actually
need
for
releasing
the
product
I
mean
I,
don't
understand
like
we
need
for
our
own
case.
We
need
for
our
own
case,
because
we
we
can't
duplicate
this
code.
If
we
add
another,
yet
another
packaging
and
release
yeah.
C
C
B
Sorry,
no
no
continue.
My
idea
was
that
in
the
time
frame,
the
time
frame
about
regular
release,
so
let's
say
one
month:
we
do
a
monthly
release
of
that
release
and
securely.
So
it
really
is
a
month,
let's
say
four,
so
it's
one
each
week,
okay,
so
it's
easier!
We
don't
do
only
for
our
deploy
releases.
So
our
own
case
and
I
mean
running
github.com
means
that
we
will
end
up
deploying
multiple
times
a
day.
Even
packages
were
only
registry
changes.
Only
decently
changes
only
pages
changes.
B
Yeah,
that's
the
point
so
how
to
deploy
no
having
this
master
source
of
truth
that
collects
every
versions
and
everything
it's
useful
for
also
deploy.
When
you
want
to
know
right
now,
I
have
to
build
something
which
version
should
I
pick
today
in
this
moment,
but
it's
less
needed
for
a
monthly
release
or
a
fetch
release
when
things
are
more
stable,
because
it's
the
thing
that
you
do
that
week
and
it's
gonna
be
packaged.
We
will
have
a
tag
in
the
end
which
tells
you
exactly
what
was
in
that
package.
B
B
A
A
A
It
was
probably
some
work
we
can
do
in
parallel.
I
mean
there's
a
lot
of
work
to
be
done
for
sidekicks,
specifically
about
how
we're
going
to
you
know
architect
the
cluster
and
how
it's
going
to
Brown.
Exactly
and
there's
probably
some
charts
work
that
we're
going
to
have
to
do,
but
as
far
as
like,
as
far
as
ensuring
that
what
we
have
running
on
telecom
maps,
to
exactly
what
we're
releasing
for
homnibus,
including
security
releases,
and
how
it
fits
into
that
process.
I
think
you
know
we
obviously
need
to
work
that
out
still.
C
A
There's
nothing
else,
I
mean
we
can
use,
we
can
use
pre,
prod
and
possibly
even
you
know,
have
a
cluster
that
we
can
or
have
the
ability
to
shut
it
down
when
we
want
to
use
pre
prod
for
something
important.
It's
just
a
matter
of
ensuring
that
cycling
isn't
picking
up
jobs
in
the
cluster,
and
that
would
allow
us
to
test
in
parallel,
and
you
know,
without
affecting
staging
I.
C
B
A
Think,
where
we
end
up
is
going
to
be
using
a
tagged
version
of
charts
for
deployment,
how
we
get
that
tagged
version
and
the
process
for
tagging
it
I
think
is
what
we're
talking
about
here.
I
think
we
all
sort
of
agree
that
I
think
where
we
want
to
end
up
is
that
we're
just
gonna
bump
the
helm
version
in
case
work
loads.
We're
just
gonna,
have
the
appropriate
sub
component
versions.
B
Yeah
I
do
agree
with
job
here,
because
if
we
assume
that
we
already
have
solved
this,
we
are
not
ready
to
start
migrating.
So
we
have
so
many
things
that
we
still
have
to
figure
out
on
the
infrastructure
and
size
that
yeah,
taking
the
assumption
that
it's
just
a
matter
of
bumping
the
helm
charter.
She
I
think
it's
a
good
one,
because
it's
where
we
want
to
end
up
with
the
release
process.
B
Top
of
my
mind,
I
was
thinking
about
the
the
job
trace
problem
that
we
had
last
week
when
we
removed
the
NFS
and
I
think
this
is
clashing
with
our
plan
in
general,
because
that's
something
that
we
cannot
remove
from
production,
the
NFS.
So
do
we
know
if
there's
some
active
development.
Oh
no,
we
don't
know
we.
There.
A
A
A
B
Reelected
as
we
speak
the
same
to
completely
work
differently
so
pages
we
should
ignore
for
now
what
I'm
suggesting
and
what
I
suggested
at
the
beginning
of
the
sidekick
discussion
is
that
we
should
look
for
the
cues
that
have
the
least
impact
and
that
I
have
no
need
for
shared
storage.
There
are
cubes,
there
are
a
lot
of
them,
yeah
that
we
can
do
without
actually
having
to
depend
on
NFS.
B
What
we
should
do
during
this
investigation,
though,
is
expose
the
ones
that
do
have
a
dependency,
because
that
will
allow
us
to
paralyze
some
of
the
work
given
that,
as
far
as
I
know,
no
one
is
actually
working
on
this.
The
sooner
we
expose
those
properly
with
the
data
of
hey.
This
is
actually
not
working
on
whatever
environment.
We
have
not
the
not
all
the
epochs
that
we
have
going
on,
but
you
actual
work
done
where
we
can
provide
provide
some
evidence.
This
is
working
or
not
working.
B
If
we
have
that
we
can,
we
can
find
a
way
to
prioritize
that,
or
rather
I
can
find
a
way
to
prioritize
that
while
we
work
on
the
other
things,
that
was
the
idea
all
along
by
the
way.
Like
I
mentioned
that
a
couple
of
times,
we
want
to
start
with
a
queue
that
we
know
has
no
problems
with
shared
storage
and
then
expose
it
slowly.
B
A
A
B
A
B
B
Besides,
you
come
in
because
there
are
sub
points
there
right,
like
even
in
the
sidekick
migration.
There
are
sub
points.
One
point
is
thinking
about
like
how
are
we
gonna
partition,
the
psychic
use
like
which
ones
are
we
gonna
select
all
of
that
thing,
and
then
the
other
part
is
which
environments
and
how
are
we
going
to
test
this,
like?
Those
are
two
big
big
things
within
one
epic
to
be
done
so
who
wants
to
take
a?
What
or
do
we
want
to?
Have
someone
decide
that.
A
That
sounds
fine
to
meet
you
I
mean
I
definitely
want
to
work
on
it.
So
like
skarbek,
do
you
want
to
try
to
lead
it?
And
then
you
know
you
can
just
see
how
it
goes
sure
and
if,
if
you
feel
like
you're,
not
free
enough
to
work
to
delete
it
and
we
can
come
back
and
figure
out
something
else,
you
know
you
know.