►
From YouTube: Repeatable DB creation Demo (2021-08-18)
A
B
Yeah
sure
it'll
just
take
like
10
minutes
to
revision,
all
things.
So
if
you
want
to
go,
if
you
want
me
to
get
started
sure.
B
Okay,
so
we're
gonna
provision
a
a
dedicated
bg,
balancer
and
petroni
plus
bg
bouncer
on
a
separate
notes.
We're
gonna
do
it
through
the
ci.
So
just.
B
And
yeah
we're
gonna
do
it.
On
the
registry
chart.
Everything
has
already
been
provisioned
with
installed.
So
what
we're
gonna
do
right
now
is
just
run
and
so
on.
B
And
we
just
wait,
takes
around
nine
minutes,
both
of
them
to
get
provisioned.
A
Is
the
is
there
an
ilb
already
for
fiji
bouncer
created
as
part
of
the
terraform?
A
B
A
A
But
that's
that's
a
requirement
for
the
central
pg
bouncer
right
so
right
now
it's
just
creating
one
pg
bouncer
which
we
could
use,
but
eventually
we'll
have
like
several
pg
bouncers
for
four
rights.
B
Yeah
correct
we're
gonna
have
three
of
them
behind
a
an
internal
loop,
balancer.
A
Okay,
I
guess
we
should
have
an
issue
open
for
that
sure
you
know
already
yeah
I'll
open
one.
B
So
it
takes
some
time,
but
that's
because
the
zebra
constant
goes
to
installing
all
the
common
tasks
of
the
task
included
by
the
common
drone
from
the
get
and
installing
omnibus
and
configuring
takes
some
time.
But
otherwise
the
rest
of
the
steps
are
just
provisioning
files
and.
B
Okay,
yeah,
that's
unfortunate.
B
That's,
I
guess,
because
of
the
latest
mr,
which
was
merged
from
the
exporter,
queries.
A
Oh,
so
it's
like
complaining
that
give
that
prometheus
doesn't
exist,
but.
A
B
I
guess
because
it's
it
is
a
it's
running
a
handler,
so
the
handle
doesn't
seem
to
have
been
run
yet.
Okay,.
A
Absolutely
maybe
we
could,
I
mean
you
could
hop
on
the
box
and
maybe
yeah.
So
you
think,
like
maybe
running
reconfigure.
A
A
C
A
C
So
I
see
paulo's
just
opened
an
issue
for
us
to
actually
just
skip
ssh
and
just
use
the
aws
commands.
I
think
we
were
talking
about
it
the
other
day
as
well.
So
I
don't
know
how
that's
going
to
change
things,
but
presumably
that
still
needs
a
private
key.
B
A
Is
it?
Is
it
reconfigure
run
in
the
common
ansible
role
like
when
the
package
is
installed
or
not.
B
I
guess
we.
Yes,
oh.
A
D
It
doesn't
know,
and
the
reason
for
that
the
reason
for
that
is
each
each
each
role
is,
is
going
in
and
actually
set
up
the
specific
conflict
file
which
component
and
then
get
figures.
A
So
you,
you
drop
the
skip,
auto
configure
file
whatever
yeah,
that's
right,
okay,
that
makes
that
makes
more
sense.
So
it
sounds
like
then.
We
need
to
make
sure
that
this
directory
happens
after
the
reconfigure.
I
can
accept.
B
Okay,
so
bg
bouncer
provisioning
has
completed
still
waiting
for
this
one,
but
I
guess
we
can
demo
the
connection.
That's
all
the
pieces
are
already
there.
So
if
I
just
go
to.
B
A
Copy
and
paste
you're
still
sharing
your
browser
screen
by
the
way,
not
the
terminal.
That's
what
you
want
us
to
see.
A
B
Okay,
thanks
rosie
heads
up:
let's
do
two
chances.
B
Can
you
see
my
turn
now?
Yes,
okay,
so
just
so,
I
didn't
fake
anything
here.
We
are
in
peach
bouncer
one,
I'm
gonna
connect
locally,
so
I
wanna
connect
through
page
bouncer.
The
port
is
correct
and
we
have
a
database
that
doesn't
exist
in
postgres
itself.
That's
a
database
provided
by
pgbouncing.
B
B
B
B
B
A
B
B
From
boom
and
psychic
were
connected
to
a
single
cluster,
so
we
need
to
have
different
database
names
on
a
single
for
a
single
page,
bouncer
instance.
It.
Maybe
it's
not
relevant
anymore.
If
we
have
a
dedicated
cluster
for
each
type
of
connection
puma
inside
it,
but
that's
like
a
relic
from
the
past.
We
carried
on
along
the
way.
A
Yeah,
so
this
came
first
and
then
and
then,
after
that
it
came
where
we
had
multiple
pg
bouncers
per
node.
So
what
we
could
possibly
do
is
just
have
different.
We
could
have
multiple
or
we
could
have
two
clusters
or
multiple
pg
bouncers
running
on
the
same
node,
even
listening
on
different
ports
and
then
have
two
ilbs
in
front
one
that
we
use
for
rails
and
one
that
we
use
for
sidekick.
D
B
Yeah
for
for
the
kid
business
you're
right,
we
can
just
have
the
you
we
can
have
like
a
almost
identical
configuration.
We
don't
need
to
change
the
database
name.
We
just
need
to
change,
maybe
yeah
pull
sizes
once
the
database
name
can
stay
the
same,
given
that
we
change
the
same
thing
in
the
application
because
they
are
configured
to
connect
to
this
particular
database.
A
A
A
All
right
cool
great:
are
there
any
questions
before
we
end
the
call.
A
Grant
for
your
benefit,
I
in
the
beginning
I
talked
about
like
where
we
are
with
the
get
stuff
and
what
we're
using.
If
you
have
any
questions
about
that,
let
me
know
so
far.
It
looks
like
on
the
exporter
front,
I'm
going
to
use
the
omnibus
postgres
exporter,
I'm
going
to
use
the
omnibus
node
exporter
where
we
can
and
and
then
for
the
others,
we'll
have
to
create
new
roles
and
maybe
we'll
look
into
seeing
whether
it
makes
sense
to
move
those
roles
to
get
or
not.
A
This
is
for
like
ebpf,
intel
and
others.
D
Yeah,
it's
it's
actually
a
classic
problem
for
me,
if
you're
saying
with
explorers,
they're
small
files,
but
actually
to
set
them
up
and
actually
get
them
running
in
linux
and
as
a
service
and
stuff
it's
actually,
it's
actually
quite
involved.
It's
more
involved
than
you'd
like
it
to
be
anyway,
especially
across
mobile.
A
D
So
that's
why
I
was
kind
of
pushing
it's
like
if
on
the
bus
is
already
doing
it
and
that's
fine
and
that's
why
I'd
still
say
again.
If,
if
mtl
has
benefit,
then
I'd
say,
the
argument
is
clear
that
it
could
be
added
to
on
the
bus
as
well
as
an
optional
explorer,
to
be
added
to
be
configured
just
for
you.
A
A
D
A
Yeah
I
mean
I
could
potentially
see
us,
you
know
taking
advantage
of
process
exporter.
I
don't
know
like
I
think,
if
we're
gonna
be
starting
to
ship
recording
rules
and
slo
definitions
with
omnibus,
then
like
for
some
of
these
exporters,
you
know
we
rely
on
metrics
from
some
of
these
exporters,
so
we
should
probably
include
them.
D
C
A
D
D
Is
something
to
work?
Well,
I
mean
genuinely,
though,
if
if
mtl
does
have
a
benefit
to
to
you,
then
truly
it
could
have
a
benefit
either.
So
it
might
still
pass
the
threshold
of
adding
it's
on
the
bus.
C
I
I
would
question
so
I
think
process
explorer
definitely
intel.
I
would
question
whether
we
want
to
ship
that
with
a
product
like
yeah,
I'm
like
like
we've
kind
of
moved
away
from
intel,
thank
goodness
like
for
almost
everything,
and
there
are
a
few
things
that
we
use
it
for
still.
But
it's
very
rare-
and
I
don't
know
if
it's
worth
all
the
effort
of
setting
it
up
and
you're,
maintaining
it
engineering,
it's
an
omnibus
thing
rather
than
you
know,
I'm
not
talking
about
for
your
specific
use
case,
but
as
an
omnibus.
A
C
Yeah,
I
would
I
would
question
whether
any
of
those
are
needed
anymore,
because
a
lot
of
that
stuff
is
in
that
queries.yaml
file
in
the
postgres
exporter
now,
and
there
was
something
else:
oh
yeah
and
the
fluency
we're
doing
some
kind
of
interesting
stuff
as
well,
where
we
are
scraping
from
fluentd
right
with
the
pg
stat
statements
and
pg
stativity
and
the
vacuuming
stats.
And
all
of
that.