►
From YouTube: GitLab 13.6 Kickoff - Enablement:Database
Description
Kickoff for the GitLab 13.6 release
https://gitlab.com/gitlab-org/database-team/team-tasks/-/issues/94#note_430887473
A
So
that
is
the
goal
of
this,
which
is
to
ensure
that
we
can
roll
back
migrations
if
we
need
to.
As
early
as
possible,
the
testing
cycle
along
similar
lines,
we
also
are
working
with
postgres
ai,
with
their
database
labs
ee
product
to
arm
our
engineers,
with
better
tools
to
be
more
effective
in
working
with
database
queries
and
migrations
and
other
type
of
content.
A
Postgres,
ai
and
their
tool
allows
you
to
do
some
thing
cloning
and
also
allows
you
to
do
some
masking
anonymization
as
well
for
content.
So
you
can
more
easily
test
on
production
like
content
without
the
challenges
that
typically
go
along
with
that
in
terms
of
cost
time
to
create
a
copy
as
well
as,
of
course,
some
of
the
data
concerns
as
well.
A
So
we
want
to
go
ahead
and
test
this
out
so
that
we
can
give
engineers
we'll
start
with
our
subset
of
users
who
already
have
access
to
this
content
today,
but
give
them
a
nicer
interface
to
it,
with
a
faster
thin
clone,
where
they
can
more
easily
test.
Some
of
these
queries
and
changes
of
migrations
without
worrying
about
you
know,
what's
going
on
with
a
larger
database,
so
that
is
the
first
step
here.
We're
pretty
excited
about
it
and
again.
A
We
want
to
make
sure
we
can
just
make
our
engineers
more
effective
and
more
efficient
in
working
with
this
content
and
particularly
testing.
This
content
at
scale
essentially
calm
scale,
so
that
we
so
that
we're
more
confident
we'll
find
any
kind
of
performance
problem
or
scaling
problems
before
it
goes
into
production.
A
So
pretty
excited
about
that
one.
On
the
topic
of
performance,
we
are
continuing
to
work
on
our
efforts
to
automatically
re-index
some
of
the
con
indexes
within
the
notification.
Currently
we
rely
on
administrators
to
run
the
re-index
manually,
and
this
is
less
than
ideal,
as
you
could.
Probably
imagine
people
forget
about
it.
If
you
don't
your
documentation,
you
might
not
be
aware
you
have
to
go
do
this,
and
this
is
just
one
example:
on.com
of
some
of
the
index
flow
that
you
can
see
building
up
over
time.
A
So
the
next
step
that
we're
taking
here
in
13.6
is
we
want
to
develop
a
heuristic
so
that
the
application
can
determine
what
which
indexes
we
should
attempt
to
rebuild
at
any
given
time
effectively,
which
ones
that
we
have
the
most
opportunity
and
are
worth
the
time
and
effort
to
go
ahead
and
run
the
rebuild
process.
So
that
is
what
we're
working
on
in
13.6
for
this,
as
we
continue
to
really
make
this
something
you
don't
have
to
worry
about
anymore
with
gitlab,
which
would
be
very,
very
nice.
A
Finally,
we
are
working
to
hopefully
wrap
up
the
overall
partitioning
project
that
we've
been
going
on.
We
have
started
off
with
the
part
the
audit
events
table
and
we've
built
a
lot
of
tooling.
We've
been
working
on
documentation
with
the
goal
of
ideally
arming
the
broader
community
to
be
able
to
partition
tables
on
their
own
without,
depending
upon
the
database
team,
to
go
ahead
and
do
this.
The
audit
events
table
was
relatively
simple,
with
a
relatively
known
and
singular
access
pattern.
A
If
we
can
do
this
and,
of
course,
we'll
update
our
tables
and
tools
as
needed
along
the
way,
then
I
think
we
feel
pretty
confident
that
we
can
turn
partitioning
over
to
the
broader
gitlab
development
community
and
have
them
be
successful
in
partitioning
their
own
tables
as
they
see
fit
and
as
they
need
to
we'll,
of
course
be
available
to
help.
But
again
we
don't
want
to
be
blockers
here
for
partitioning
any
table
that
needs
to
get
partitioned,
because
over
time,
hopefully
many
of
them
will
to
improve
performance.
A
So
that
is
the
major
items
we're
working
on.
If
you'd
like
to
join
the
conversation,
please
do
so.
I
will
include
the
planning
issue.
You
see
here
the
link
to
it
in
the
video
on
youtube
and
so
jump
in
you
can
click
through
explore
the
issues.
Of
course
you
can
see
our
direction
page.
This
is
working
on
there
and
we
look
forward
to
talking
with
you
and,
if
not
we'll
talk
to
you
in
about
a
month
and
tell
you
what
we're
working
on
in
13.7.