►
From YouTube: GitLab 13.12 Kickoff - Enablement:Memory
Description
Kickoff for the 13.12 release for the Memory team.
Planning issue: https://gitlab.com/gitlab-org/memory-team/team-tasks/-/issues/89
A
Hello,
everyone,
my
name,
is
fabian
simmer
and
I'm
the
product
manager
for
the
memory
team
at
gitlab
and
I'm
quickly
going
to
walk
you
through
the
13.12
plans.
We
have
please
keep
in
mind
that
during
the
13.11
release
there
was
a
lot
of
work
as
part
of
a
rapid
action
to
improve
the
performance
of
our
database,
where
the
memory
team
made
a
lot
of
contributions.
A
So
some
of
these
items
are
carry
over
from
the
older
the
13.11
release.
So
the
first
thing
that
I
would
like
to
talk
about
is
an
overall
effort
to
reduce
the
memory,
consumption
for
puma
web
server
and
sidekiq,
which
is
processing,
background
jobs
and
so
we've
merged
functionality.
That
allows
us
to
actually
measure
the
amount
of
memory
that
specific
end
endpoints
allocate,
and
that
then
helps
us
to
understand
if
there
are
specific
endpoints
that
use
a
very
large
amount
of
resources
and
we're
trying
to
reduce
that.
A
So
we
are
in
essentially
the
process
of
having
identified
specific
candidates,
and
now
the
team
is
working
through
that
candidate
list
and
is
trying
to
understand
why
some
of
these
endpoints
are
using
large
amounts
of
memory,
and
we
are
going
to
document
why
that
is.
We
are
attempting
to
reduce
the
overall
memory
consumption,
but
in
the
sort
of
medium
term,
we
are
also
trying
to
provide
better
tooling
to
our
developers
to
either
fix
these
kinds
of
problems,
but
also
to
raise
the
visibility
of
the
sort
of
overall
memory
consumption.
A
A
The
other
item
is
that
unicorn
support
will
be
removed
and
you
will
have
to
use
puma
as
an
alternative
and
from
14.0.
So
there's
a
little
bit
of
work.
That
needs
to
be
done
to
ensure
that
everybody
is
actually
aware
and
that
everybody
is
using
puma
before
upgrading
to
14.0.
A
So
in
puma
itself
in
puma
version
5,
which
we
are
now
using.
There
are
a
few
changes
to
do
with
memory
management
and
we've
tried
to
roll
that
out
in
production
and
matthias
is
going
to
essentially
talk
about
some
of
the
learnings
that
we've
had
during
the
that
process,
and
I
hope
that's
going
to
be
useful.