►
From YouTube: Geo HA upgrade demo 12.0 to 12.1 - Part 1
Description
In this video we're attempting to perform an upgrade from GitLab 12.0 to 12.1 on a Geo HA setup. We're evaluating the new documentation, which is not merged yet.
In addition, an unusual thing in 12.1 is PostgreSQL is automatically upgraded to version 10, which requires downtime on a Geo installation, so in this demo we attempt to follow the instructions that tell you to skip the PG upgrade.
The recording is in two parts because I did not have a Pro account with Zoom, at the time.
A
B
A
A
A
A
A
A
A
A
A
A
A
A
A
A
C
A
Here
upgrade
the
package
on
the
primary
upgrade
the
package
on
the
secondary
reinitialize,
the
database,
okay,
so
I
I
think
I
think.
Is
it
the
case
that
these
are
the
instructions
for
if
we
do
want
Postgres
to
be
updated
and
the
this
is
the
instruction
for
if
we
just
want
to
skip
that
this
time?
Okay,
so
since
I
don't
know
since
this
doesn't
seem
to
be
working,
then
we're
going
to
have
to
not
do
that.
I.
B
Mean
that's
a
it's
an
interesting
question,
so
I
have
done
this
on
a
non
aged
a
system
following
the
host
instructions
and
I
think
it
has
worked
in
some
installations
that
has
failed
in
others,
but
I
think.
The
first
question
is
like:
is
it
okay
that
that
require
returns?
Nothing
right
and
I
think
I
think
almost
like
I'm
more
interested
in
understanding
how
we
can
actually
upgrade
the
database?
Okay
I
see,
but
it's
a
good,
its
I'm
making
notes
in
in
the
geo
discussions
to
say
that
this
returns.
C
Me
have
some
insights
here
so
that
the
the
replication
slots
will
only
be
available
before
we
upgrade
the
database
from
9.6
to
10.7.
10
got
something
if
we
installed
gitlab
the
way
that
it
is
on
our
documentation.
We
do
the
replication
using
replication
slot.
It
is
something
that
the
Postgres
keeps
some
metadata
about,
how
many
in
which
different
instance
is
replicating
from.
So
if
you
have
like
three
other
post
squares
servers
replicating
from
the
primary,
it
knows,
which
one
is
like
the
the
one
that
is
lagging
the
most.
C
A
C
A
That's
interesting
because
I
thought
I
thought
that
get
led,
P
sequel
would
use
the
correct
there
correct
the
connection
info,
but
okay,
so
I
suppose,
if
we
put
into
the
instructions
here
run
on
the
node
that
is
running
the
Postgres
database,
that
this
would
work.
And,
additionally,
if
there
is
an
issue
for
like,
as
Gabriel
said,
running
P
sequel
on
a
different
node,
then
that
would
be
helpful
as
well.
A
B
Think
the
first
thing
to
note
here
is
that
I
don't
think
these
instructions
were
written
with
AJ
in
mind
Ryan's.
Looking
at
this
yeah
I
mean
you
could
say
that
upgrading
gitlab
e
and
on
the
primary
and
secondary
is
still
correct,
but
that
entails
very
different
things,
depending
on
your
setup
right.
C
A
C
B
A
A
B
B
Which,
by
the
way,
just
prints
out
something
where
it
doesn't
really
modify,
any
state
I
I
would
have
said,
follow
the
instructions
for
a
multi
node
upgrade
and
then
at
the
end
you
know
have
formed
the
step
number
four,
which
is
reinitializing
the
database,
because
that
is
kind
of
how
I
read
these
instructions.
It's
like
find
your
slot
name
do
a
regular
grade.
A
A
A
A
A
A
A
A
A
A
Oh,
this
is
speaking
only
about
the
primary.
It
is
okay,
so.
A
A
A
A
A
A
A
A
B
A
A
A
B
B
A
A
B
C
So
worst
case
scenario
in
a
car.
The
best
case
scenario
would
be
that
considering
they
are
using
PG
bounce
bouncers
of
like
that,
whenever
the
primary
goes
down
for
you
to
be
upgraded,
the
master
posters,
you
have
a
downtime
there
and
when
it
goes
up
you
can
route
everything
to
that
don't
so
there
is
like
this
minimal
amount
of
downtime,
and
after
that
you
can
start
restoring
every
other
follow.