►
Description
The fourth meeting of the group dedicated to solving challenges around building and deploying open source CMSs on Kubernetes is our first time welcoming our Australian friends into the fold! In this meet-up we focus on classic topics such as Drupal’s files system management, taking a deeper look at object storage.
We also take a look at two K8s presentations from the Drupal space:
"How to manage Multiple Drupal sites with Lagoon" by Michael Schmid: https://bit.ly/38GUUDl
“Drupal CI/CD from Dev to Prod With GitLab, Kubernetes and Helm” by Yevgen Nikitin: https://bit.ly/3iaepaB
Catch up with the group on GitHub: http://bit.ly/338dXC5
A
Yeah
well,
I
guess
this
is
a
next
meeting
of
the
kubernetes
cig.
Drupal
we've
been
talking
quite
a
bit
about
kubernetes
and
its
impact
with
drupal
drupal.
I
think
over
the
long
term,
by
general
statement
of
I
guess
purpose
is
to
talk
about
the
things
that
we're
doing
with
kubernetes,
maybe
identify
some
of
the
common
problems.
I
think
we
finally
lab
write
on
the
problems
and
then
eventually
bring
those
problems
back
to
the
different
communities
in
this
case
Drupal.
A
This
is
the
first
time
we've
had
I'm
going
to
assume
mostly
Australian
audience,
but
I
know
that
it
has
been
previously
talked
about
earlier
to
start
alternating
the
schedules
and
the
first
time
they
are
actively
doing
that
so
I
think
as
a
first
order.
Well
just
heard
that
the
next
meeting,
but
the
meeting
after,
if
this
is
a
good
time
slot
or
this
particular
group,
will
go
ahead,
and
this.
A
And
typically,
what
we
like
to
do
is
just
start
with
quick
introductions.
Maybe
a
quick,
hello
and
who's
here
just
be
considerate
of
everybody
else.
Try
and
keep
it
about
30
seconds
to
under
a
minute.
My
name
is
Kevin
bridges,
I'm
the
CTO
and
founder
at
a
company
called
dread
and
the
product
and
have
a
lot
of
interest
in
kubernetes
I.
Think
it's
a
pretty
cool
thing
to
change
the
world
with
and
let's
just
go
around
the
room.
I
guess.
A
B
D
A
C
G
Yes,
when
it
says
microphone
off
means
microphone
on
I'm,
the
tech
lead
for
the
CMS
project
finance
in
Australia.
Whilst
we
got
amazing
doing
most
of
our
Canaries
OpenShift
Dockery
stuff,
we're
obviously
really
invested
in
the
future
of
docker,
but
most
importantly,
kubernetes
so
I'm
sort
of
here
listening
or
watching
trying
to
keep
notes.
So
this
side,
where
it's
still
doing.
A
F
Gonna
try
some
work
for
a
mozi
I/o
and
my
primary
job
is
really
working
with
Toby
and
managing
cluster
or
called
awesomest
the
program,
which
is
the
Australian
Government
common
web
platform.
So
we
run
a
about
your
site's
on
that
and
quite
a
large,
giving
each
cluster
so
on
yep,
big
scale,
things
always
always
interested
in
working
out
optimizations
and
anything
that
can
make
improvements
to
how
Drupal
runs
on
community.
H
Hello,
I'm
Ellie
I
work
for
open
strategy
partners
supporting
Kevin
and
D
dev
in
there,
whatever
they're
up
to
so
I'm
learning
a
lot
about
kubernetes
and
that's
part
of
why
I'm
here,
I'm
also
semi
in
charge
of
the
minutes,
and
if
anybody
wants
to
hop
in
down
there
and
just
drop
in
your
links
and
notes
what
you're
saying
as
it's
happening.
That's
super
helpful.
Thank
you.
A
E
Sure
Simon
Lindsay
longtime
dribbler
being
working
with
urban
shift,
in
particular
as
opposed
to
humans
for
2b
plus
years
now,
and
before
that
we
did
some
docker
orchestration
stuff
ourselves
and
started
that
was
sort
of
horrible
working
at
the
University
of
Adelaide
and
working
with
Nick
at
the
previous/next
and
the
other
places
as
well.
On
some
of
the
things
we're
working
on
I'm
alone.
Excellent.
A
Great-
and
you
know,
let's
just
go
ahead
and
jump
in
the
agenda-
didn't
unfortunately
have
enough
time
to
prepare
for
this.
So
I
definitely
appreciate
the
feedback
that
we've
gotten
so
far
right
off
on
the
agenda.
We've
got
a
couple
of
presentations.
We
last
week
or
last
meeting,
we
focused
a
little
bit
on
some
of
the
work
that
tests
has
been
doing
in
the
community.
She
did
a
presentation
that
is
focused
on
getting
doctor
and
kubernetes
to
play
well
with
Drupal
I.
A
Think
Jeff
Skilling
also
had
a
presentation
that
we
highlighted
a
little
bit
I'm,
not
sure
which
camp
specifically
they
were
speaking
at.
But
if
you
want
to
take
a
few
minutes
and
look
at
the
scroll
down
a
little
bit
in
the
meeting
notes,
documents
you'll
actually
see
links
for
those
but
yeah
I.
Guess
who
would
be
the
great
person
to
talk
a
little
bit
about
that?
First
link
that
we
have
there
how
to
manage
multiple
Drupal
sites
with
lagoon
yeah.
F
Okay,
I
think
I
posted
that
in
their
Channel
last
night,
I
thought
it
was
just
a
you
know:
I
actually
just
watched
it
the
other
day
because
I've
been
on
leave
but
I'm.
Yes,
it's
interesting
approach
to
how
we
run
things
and
with
massive
sort
of
multiple
Drupal
sites
which
are
kind
of
like
multi-site,
but
not,
and
just
a
really
nice
play
they
sort
of
manage.
F
That's
a
scale
means
that
we
can
actually
dispatch
instead
of
a
one
place
approach
and
a
base
image
and
heavily
focus
on
testing
that
before
deploying
out
to
you,
know
hundreds
of
other
sites
and
all
through
their
own
unique
CI
pipelines.
So
we
can
actually
test
and
roll
back
each
individual
site,
but
yeah
there's
just
you
video.
You
know
you
know
one
has
questions
just
asking
the
channel,
but
we're
hopefully
going
to
submit
a
decision
and
for
DrupalCon
incident
yeah
along
those
lines
as
well.
I.
A
Know,
there's
been
a
lot
of
conversation
about
how
to
actually
segment
out
core
Drupal
from
possibly
you
know
different
file,
directories
or
composer
based
directories
so
that
it
can
be
as
efficient
as
possible,
utilizing
docker
but
still
dynamic
enough
to
be
able
to
be
useful
and
it
seems,
like
generally,
people
are
focusing
on
and
correct
me
if
I'm
wrong
here,
but
just
a
strategy
of
keeping
core
as
static
as
possible
to
be
able
to
do
the
types
of
deployments
that
you're
describing.
Does
that
sound,
accurate
yeah.
F
A
And
have
you
yeah
I
guess
we're
going
to
talk
about
security
a
little
bit
later
on,
so
you
say
so.
Maybe
I'll
touch
on
that
there,
but
it
seems
like
you
know.
This
strategy
is
generally
becoming
predominant
as
far
as
being
able
to
be
a
component
of
managing
a
larger
security
stance
and
I'm
curious
if
you've
seen
benefits
in
that
area
as
well
from
this
approach,
yeah.
F
I
think,
if
you
put
that
note
in
about
community
security,
so
there's
definitely
benefits
because
you've
got
that
single
point.
We
actually
can
monitor
and
track,
and
you
have
image
scanning
and
things
across
that
layer
and
then
only
really
need
to
worry
about
the
you
know
the
customizations
at
them
at
the
very
end,
when
a
site
actually
becomes
a
site
which
is
often
adding
sort
of
some
sort
of
theme
customizations
of
things,
so
it
just
reduces
that
surface
area.
We
have
to
actually
test.
A
A
So,
unfortunately,
we
don't
have
anybody
here
to
speak
about
it.
How
long
has
been
a
previous
topic
of
conversation
for
those
that
are
interested?
There's
a
essentially
a
love-hate
relationship
with
helm,
I
think
the
general
consensus
right
now
is
that
helm
3
makes
grand
promises
and
sounds
amazing,
and
we
hope
that
it
fixes
everything
in
the
world
that
it
claims
to
fix.
But
there's
a
little
bit
of
skepticism
around
that
as
well.
A
I
know
I
a
lot
of
times
it
generally
through
the
conversation,
we've
kind
of
picked
up
on
the
fact
that
people
tend
to
overuse
oh
and
make
it
do
a
bit
much
and
as
a
result,
it
tends
to
be
a
little
difficult
to
debug
at
times,
specifically,
if
you're
talking
about
home
I
can
lead
to
a
little
bit
of
frustration.
Trying
to
figure
out
what's
going
on
in
a
more
complex
system
does
not
fail.
A
B
A
B
Definitely
I
yeah,
we
played
with
with
home
III
write
a
couple
of
pull
requests
to
do
like
post-deploy
steps
and
things
like
that,
like
kind
of
because
I
know
they
do
have
that
hook
in
bits
like
ads,
screaming
and
all
that
kind
of
stuff
to
it.
But
yeah.
This
is
really.
We
act
between
2:00
and
3:00.
We're
like
I,
submitted
a
couple
of
pull
requests.
They
said,
but
I'm
really
gonna,
add
new
features
like
home.
B
3
is
coming
and
I
kind
of
said,
whiz
like
what
can
I
contribute
to
healthy
and
they
said
no,
it's
being
worked
on
actively
behind
the
scenes
say:
oh,
it
was
a
really
it's
been
in,
like
really.
We've
stayed
for
a
long
time
and
like
coming
back
now
and
they're.
Saying
hey
like
you,
didn't
keep
going
now.
If
you
want
rebase
so
on,
yeah
I've
had
an
opportunity
to
play
with
home
tree
yeah
but
yeah.
B
It's
they
had
me
bright
up
until
they
said
Lua
and
then
I
wasn't
sure,
say:
I,
don't
know
her
like
I
get
it,
but
then
like
I
yeah,
it's
it's
interesting,
so
yeah
I
I
would
I
want
to
give
it
a
go
like
I
love
like
ditching,
that
to
the
piece.
But
then
it
sounds
like
it's
going
to
compete
a
lot
with
like
customized
and
customized
customized
and
those
kinds
of
tools.
Now
a
mattering
and
bit
yeah
I
think
we'll
see
how
it
shakes
out.
It's
gonna
heat
up
again:
I
guess
we.
A
A
B
That's
that's
the
same.
He
and
Ruby
know
I'll
post
a
link
and
we
have
bunch
of
operators
that
we've
been
working
on
over
the
past
six
months,
like
kind
of
starting
with
like
a
drupal
operator,
and
then
it
kind
of
splits
out
and
so
I
guess
in
TV
database,
all
that
kind
of
stuff.
So
I'd
love
this
a
few
people's
feedback
on
that.
E
A
A
Be
able
to
get
the
word
out
about
those
things,
one
of
the
things
that
I
would
like
to
see.
People
talking
about
an
Amsterdam
is
how
to
collaborate
on
those
types
of
operators,
because
they
are
very
common
and
they,
you
know
we're
very
ambitious
open-source
engineers.
So
it's
very
likely
that
we
end
up
with
a
dozen
or
two
dozen
implementations
of
the
exact
same
thing,
and
you
know
we'd
stand
to
benefit
a
lot
if
we
work
together
on
those
things
so
yeah
take
your
time
and
share
absolutely.
B
Guess
that's
the
other
thing
with
home.
It
was
like
it
was
coming
with
an
operator
as
well.
It's
kind
of
back
in
two,
so
I'm
interested
to
see
how
that
plays
out
on
the
spectrum
of
operators
to
and
like
it's
yeah.
It's
definitely
got
some
grand
but
yeah.
To
reiterate
you
and
me
to
step
back
to
let
others
baby
definitely
sounds
like
there's
a
it's
trying
to
take
on
a
lot
of
pieces
and
yeah
I'm,
not
sure
which
one's
gonna
stick
yet
Intel
tree
Tommy
we're
gonna.
F
There
is
some
sort
of
convergence
at
some
stage
a
tree.
This
is
the
operator
SDK
approach
as
to
helm,
but
I
feel,
like
you
know
how
long
sort
of
taking
that
the
deployment
paying
away,
but
not
not
that
full
life
cycle
of
management
of
an
application.
So
you
know
if
helm
is
going
to
go
anywhere
and
sort
of
increase.
Its
feature
said
it's
going
to
go.
Move
into
that
space,
where
it's
sort
of
touching
over
into
the
operators
space.
F
A
Know
they
some
of
the
there's
a
office
around
the
corner
where
Diaz
was
originally
and
I
know
that
they
are
very
very
adamant
in
fixing
the
problems
surrounding
helm,
they're,
very
optimistic
about
the
future
and,
like
I
said
it
does
sound
like
an
amazing
amazing
project,
but
I
think
you
know,
the
pain
is
pretty
common
that
we
can't
really
sit
around
and
wait
at
this
point.
We
don't
have
that
luxury
yeah.
F
Well,
I
think
that
I
saw
the
Microsoft
blog
about
how
three
and
one
of
the
subclasses
said
film:
three
it's
built
for
production
workloads
with
one
thing
and
which
is
pretty
scary,
so
if
you're
using
alum
right
now
on
production.
Now,
what
do
you
mean
to
do-
and
this
is
right
from
the
you
know-
from
the
developers
so
yeah.
A
A
B
A
Anybody
in
the
group
have
any
words
of
advice
for
people
that
might
not
know
what
home
is
or
might
be
getting
into
kubernetes
and
might
be
looking
at
using
home.
Maybe
some
words
of
advice,
things
to
avoid
that
type
of
stuff.
I
guess
my
would
be
to
not
overuse
it.
It
does
have
a
purpose,
it's
very
good
at
general
things,
but
if
you
find
yourself
running
into
the
wall
and
trying
to
get
it
to
do
things
that
seem
exceedingly
difficult,
it's
likely
that
you're
you're
you
have
a
square
peg
in
a
round
hole.
A
F
Yeah
definitely
try
to
avoid
really
complicated,
charts.
I
think
that's
that's
the
most
common
way
that
I've
been
bandwidth,
helm,
just
charts
that
really
zoomed
in
it's
probably
just
doing
too
much.
That's
that's
back
to
Kevin's
point
is
possibly
it
could
have
been
done
better,
as
elsewhere
I
mean
at
the
end
of
the
day,
the
the
output
of
helm
is
just
a
set
of
templated
objects.
B
B
You
know
that
miles
and
miles
and
miles
of
you
know
comments
like
202
a
value
and
then
space
comment,
comment,
comment,
comment,
value
value
and
it
keeps
going
down
yeah.
So
I
guess
that's
a
very
good
sign
of,
like
as
your
charts,
getting
more
complex
that
files
getting
bigger
because
you're
trying
to
add
add
more
education,
so
maybe
start
with
way
more
defaults.
Just
in
the
chart
itself.
A
Next,
up
on
the
agenda,
we've
got
Drupal
con,
Amsterdam
I.
Think
that's
going
to
be
a
pretty
recurring
topic
of
these
talks
until
Amsterdam
does
occur,
lots
of
excitement
around
there.
We
were
able
to
talk
with
Michael
Schmidt
previously
in
the
last
night
up
about
some
of
the
efforts
that
were
trying
to
get
into
the
mindset
of
some
of
the
organizers
with
this
pupal
con
and
specifically
that
there
is
a
lot
of
interest
around
getting
kubernetes
and
drupal
to
work
together.
Well,
I
know
that
there
were
a
couple
of
sessions
submitted
by
amazing
we've.
A
Yeah
deed
abside
submitted
quite
a
few
sessions
as
well
that
are
directly
related
to
kubernetes.
It
seems
like
there's
a
good
opportunity
to
get
together,
even
if
it's
a
bit
of
an
informal
track
to
to
have
a
bit
of
a
talk
track
around.
You
know
the
efforts
that
we're
going
through
and
the
things
that
we're
learning
as
a
result
of
some
of
the
work
that
we're
doing
the
goal
being
to
remove
pain.
A
If
anybody
has
any
thoughts
about
answer
damn
or
any
insight
into
some
of
the
planning,
that's
been
happening
there
or
we've
got
an
open
slot
from
the
last
conversation
about
organizing
at
kubernetes
social,
which
might
just
be
a
bit
of
a
smaller
group,
I
think
the
challenge
around
discussing
kubernetes
in
the
way
that
we
might
want
to
add
a
place
like
Drupal
con
Amsterdam
is
specifically
that
we're
not
really
looking
to
have
hi.
This
is
kubernetes.
This
is
what
docker
is
this.
A
F
F
Unfortunately,
but
the
idea
is
that
we're
gonna
have
three
sessions
in
one
stream,
all
in
a
row
which
will
be
like
a
mini
summit.
I
guess
we
can
pull
it
off
yeah,
but
it'd
be
great
to
have
you
know.
People
have
come
along
and
obviously
submit
ideas
and
I'm
sure
not
only
those
sessions
but
great
conversations
outside
of
that
as
well.
If
we
all
get
together
to
look
and
call
to
her.
A
F
B
B
A
And
we
have
several
events
that
we
can
practice
that,
to
you
know
bad
campus
coming
up,
there
might
be
a
little
bit
more
informal
flexibility
to
to
organize
and
Paul
tracks
and
maybe
record
them
as
well.
Did
you
know
the
remote
in
distributed
nature
of
working
on
kubernetes
is
just
something
that
were
open.
Source
in
general
is
something
that
I
think
that
we
can
focus
on.
A
F
A
A
different
venue
entirely
this
year,
so
yeah
so
I
think
that
the
camp
got
a
little
dispersed
because
of
the
logistics
of
working
with
a
campus
and
as
a
result,
I
don't
know
that
the
conversations
were
as
cohesive
as
they
had
been
in
previous
years.
So
there's
sense
that
I'm
getting
from
the
people
that
I'm
talking
to
or
that
it's
going
to
be
in
the
hotel
Shattuck,
which
is
just
down
the
street
from
where
they
typically
have
it
at
Berkeley.
A
But
it's
going
to
be
a
smaller,
more
intimate
venue,
there's
going
to
be
three
dedicated
rooms,
and
forgive
me
if
I'm
slaughtering
this
we're
not
getting
it
accurate
I'm,
not
an
organizer
for
bad
cam,
but
I
believe
that
they're
going
to
have
three
larger
rooms
and
then
a
kind
of
a
social
area
for
food
and
other
hallway
track
kind
of
conversations.
So
I
think
there's
a
good
opportunity
to
kind
of
you
know
really
get
intimate
with
that
conversation
and
have
it
be
a
little
bit
more
effective
this
year.
A
D
A
A
B
Crickets
nan,
Marie
I,
read,
read
only
file
system,
actually
gentle
humble,
but
I
was
the
one
that
over
the
line
and
got
emerged
into
kubernetes.
For
that
we
go
into
context
like
docker
exposed
it
and
then
I
just
did
the
API
pieces.
She
pretty
awesome
yeah
so
that
that
one's
kind
of
special
to
me,
but
every
donor
file
system
and
I
think
that's
a
really
big
deal
for
you
know
for
us.
Given
you
know
like
a
lot
of
these,
exploits
they
come
out
like
the
first
thing.
B
A
There's
a
lie.
You
know
everything
from
somebody
uploading,
a
Bitcoin
miner
written
in
PHP
to
just
god
only
knows
what
it's
all
over
the
place,
but
I
think
that
some
of
the
upcoming
lightning
talks
might
be
a
bit
of
a
good
forum
to
focus
on
security
specifically
and
I've
got
a
couple
of
people
in
mind
that
can
help
there
thanks.
F
Yeah
so
I
put
that
for
the
question,
and
now
it's
mine
you
more
about
the
tool
surrounding
security.
If
anyone
had
any
recommendations,
you
know
we're
doing
a
review
at
the
moment
on
twistlock,
which
which
looks
pretty
good,
but
I
was
the
other
image
scanning
tools
or
all
things
in
the
ecosystem.
That
could
could
be
useful
if
anyone
has
had
any
experience
or
just
any
ideas
on
things
to
research.
B
B
F
A
We've
looked
at
twist
lock
a
little
bit.
We've
looked
at
cops
a
little
bit
to
help.
I
can't
strongly
recommend
using
cops
for
several
reasons
and
then
there's
a
couple
of
services
that
we've
looked
at
as
well
we're
starting
to
find
genders
that
are
stepping
up
to
offer
to
install
you
know,
I
a
pod.
Basically,
that
will
monitor
the
kernel
essentially
and
give
you
a
fair
amount
of
insight
and
then
for
me,
what's
actually
happening
and
there's
a
lot
of
variance
in
those
vendors.
A
Specifically,
you
know,
I've
had
speaking
in
u.s.
dollars,
quotes
everywhere
from
$30,000
a
year
to
250,000
dollars
plus
a
year
to
help
secure
a
specific
cluster.
So
I
think
that
at
some
point,
there's
likely
to
be
a
more
of
a
focus
group
on
security.
Much
like
Drupal
has
the
security
team
for
the
rest
of
the
project,
and
it
might
be
a
good
opportunity
to
start
getting
more
of
a
cohesive
answer
about
what
we
do
from
a
security
perspective
and
how
we
deal
with
it
as
a
community,
because
it
is
a
very
difficult
conversation.
F
B
At
the
same
token,
some
of
these
leap
levers
that
we
have
now
when
they're
in
the
traditional
hosting
you
know
like
like
when
were
you
able
to
like
just
turn
on
read
on
you?
You
know
for
your
app
we'll
talk
about
like
what
your
app
does
and
then
contain
it
down
into
facts.
You
know,
by
applying,
like
you
know,
all
your
like
secret
routes
and
things
like
that
to
it.
So
if
I
yeah,
it's
like
buy
one
like
it's
more
complicated
but
yeah
we're
all.
I
B
B
F
D
A
D
One
thing
the
like:
we
don't
use
at
this
point
because
we've
basically
I've
entirely
trusted
workloads
but
like
pod
security
policies,
so
having
those
set
up
and
you
can
enforce
stuff
like
you
know,
you
IDs,
that
the
containers
run
as
file
system.
You
know,
must
be
read-only
or
not.
So
those
are
just
a
really
handy
scene,
easier,
yeah,
you're,
running
containers
that
you
don't
have
embedded.
D
A
J
It's
gonna
say
that
from
FCMs
point
of
view,
we're
sort
of
security
for
us
is
paramount.
The
whole
concept
of
trusted
versus
untrusted
workloads.
We
have
various
levels
of
trusted
workloads
so
obviously,
as
Tom
said,
the
ones
that
we're
building
and
deploying
on
our
own
images
we
know
all
about,
but
as
we
start
to
build
offerings,
agencies
can
do
more.
How
much
are
we
comfortable
with
people
doing
inside
are
trusted
before
they
become
untrusted?
So
that's
why
tools
like
twistlock
or
AK
or
any
of
those
things
we've
got
to
start
looking.
J
F
A
And
then
you're
in
a
world
of
discomfort
when
you
hit
a
scenario
where
you
think
that
you're
working
with
trusted
code-
and
you
think
that
you
know
it
inside
now,
but
in
fact
it's
been
compromised
in
some
way
shape
or
form.
And
ultimately
you
end
up
help
propagating
some
of
the
security
issues,
because
you're
under
the
impression
that
there's
a
level
of
trust
there
that
might
not
be
warranted.
J
Around
evidence
as
well,
you
can
know
that
you
know
you
trust
people,
but
how
do
you
know
what's
the
evidence
and
for
us
particularly
having
to
go
through
sort
of
all
the
specific
government
security
accreditation?
It's
having
that
reassurance
that
what
you
say
really
is
you
can
adapt
that
all
the
time
yeah
well.
A
A
Cool,
so
it
sounds
like
two
more
items
remain
on
the
agenda.
The
first
is
a
prometheus
maker
for
drupal
and
what
that
might
look
like
and
then
the
second
is
just
kind
of
an
opportunity
for
anyone
to
share
any
challenges
or
celebrations
that
they've
recently
come
across.
We
may
not
have
an
immediate
answer
here,
but
getting
it
out
there
might
help
us
find
it.
So,
let's
go
ahead
and
dive
into
prometheus
I'm,
not
quite
sure
who
raised
that
one
I.
F
Did
yeah
actually
puts
up
a
couple
of
weeks
ago?
Wasn't
too
many
around,
but
I
was
just
interested
to
know
if
anyone
was
keen
to
to
join
forces
and
look
at
look
at
waste.
Basically
a
previous
scraper
for
the
Drupal
laser
module,
that's
just
pulling
out
matrix
within
Drupal,
so
you
know
what
kind
of
matrix
that
like
no
counts.
You
know
that
inter
needs
counts
and
things
I
guess
just
having
that
plumbing
in
place
means
that
we
can
just
add
different
metrics
and
module
versions
would
be
a
good
good
point
and
things
like
that.
B
Cool
yeah,
like
the
big
push
for
something
like
this,
was
for
like
Hughes
Hughes
is
like
yeah.
It's
a
really
really
big
deal,
especially
when
you're
you
know
moving
through
a
community
space.
You
know
hosting
architecture
and
then
you're
pushing
stuff
SMTP
off
you
know.
Traditionally,
you
know
a
lot
of
people
might
just
use
the
host
use.
The
VM
used
essentially
host
switch
really
super
awesome,
but
when
you're
going
into
a
container
based
hosting
infrastructure,
then
you
need
to
stop
looking
at
those
services.
You
probably
need
to
be
a
bit
more
resilient.
B
F
A
Cool
and
then
I
guess,
the
next
item
that
we
have
up
is
just
kind
of
you
know
were
stories
happy
stories.
However,
you
want
to
say
it
I,
just
you
know
anything
that
comes
to
mind.
That
might
have
been
a
success
for
you
anything
that
you're
having
a
particular
challenge
in
now
is
the
opportunity
to
raise
that
and
share
it
with
the
group
and
I
know
ours
right
now,
we're
looking
specifically
at
a
lot
of
firebase
usage
to
handle
authentication
and
authorization
in
different
capacities
across
multiple
products.
A
So
for
us
you
know
paying
attention
to
how
to
get
firebase
to
work
well
with
a
cube
native
API
has
been
quite
the
challenge:
I'm,
not
entirely
sure
that
those
two
products
were
meant
to
work
well
together,
but
you
know
we
have
gone
through
the
process
of
understanding
the
bootstrap
cycle
of
kubernetes
and
how
to
get
those
bearer
tokens
working
properly
has
been
very
interesting,
so
hopefully,
we'll
have
some
good
talks
coming
out
of
that.
Pretty
you.
D
We
had
a
good
win
over
the
weekend
in
the
last
six
months,
so
we've
just
launched
a
rebuild
of
a
government
public
service
application
serves
220
million
requests
a
month.
Now
the
interesting
thing
about
this
was
we
had
it:
a
Drupal
7
site
running
on
our
kubernetes
cluster
and
during
the
Drupal
8
rebuild
the
team
decided
to
do
incremental
migrations,
so
take
one
feature
from
the
Drupal
7
site,
build
it
in
Drupal,
8
and
then
start
routing
traffic
to
that
feature
on
the
new
version.
D
So
you
know
we
achieve
this
by
using
like
ingress
rules,
so
our
H,
a
proxy
ingress
controller
we
configured
to
basically
ya,
be
able
to,
rather
than
just
host,
like
straight-up
domain
names,
to
a
specific
deployment
to
be
added
to
it
at
a
path
based
level
to
so
where
all
the
default
traffic
would
go
to
the
Drupal
7
site.
We
could
say
if
you
go
to
slash
service
centers,
which
is
the
path
where
people
could
look
up
a
specific
location
that
would
then
be
routed
to
the
Drupal
8
site,
transparently
and
so
yeah.
D
We
did
that
a
couple
times
during
the
project
and
then
the
final
cut
over
was
basically
just
removing
the
the
ingress
from
the
Drupal
7
foot
and
magically
was
going
to
the
Drupal
8
site,
so
that
was
yeah.
Really
good.
Win
allowed
us
to
have
the
site
in
production
for
a
very
long
time,
rather
than
having
to
do
like
a
big
kind
of
one-off
Big
Bang
at
the
end,
and
also
meant
that
we
didn't
have
to
rely
on
the
client
performing
any
DNS
changes
at
the
critical
moment.
D
C
A
Alrighty
then
again
to
be
respectful
of
everybody's
time.
I
will
go
ahead
and
keep
it
on
schedule.
So
thank
you
all
very
much
for
your
participation
and
I
definitely
look
forward
to
speaking
with
everybody
in
the
near
future
and
continuing
this
I
love
the
fact
that
we
can
alternate
the
times
and
get
even
more
people
involved
so
nice
to
meet
you
and
looking
forward
to
doing
more
with
you
thanks.