►
From YouTube: GitLab 15.8 Kickoff - Systems:Gitaly
Description
Gitaly kickoff for GitLab 15.8
Gitaly planning issue - https://gitlab.com/gitlab-org/gitaly-planning/release-planning/-/issues/6
A
Hello,
everyone
and
welcome
to
the
Italy
15.8
planning
kickoff
for
gitlab
version
15.8
here
coming
out
in
January
of
2023..
My
name
is
Mark
Wood
I'm,
the
product
manager
for
the
giddly
section,
I'll
be
going
over
our
plans
for
the
next
month.
How
we
plan
on
making
improvements
to
our
product
and
some
of
the
things
we're
research
in
the
meantime,
I
do
want
to
call
out
off
the
bat
that
there
is
a
pretty
steep
capacity
reduction.
That's
released
simply
due
to
the
holiday
season.
A
We
want
to
ensure
that
all
of
our
team
members
feel
like
they
can
take
the
time
off.
They
need
to
recharge
and
spend
time
with
their
family,
so
we're
trying
to
schedule
this
release
a
little
bit
later
in
hopes
that
we
don't
overload
our
team
members.
I
also
want
to
point
out
that
our
bug
fixes
here
have
been
worked
through
our
quad
planning
process.
A
So
some
of
our
objectives
and
themes,
we
are
in
the
process
of
planning
for
our
first
quarter
of
2023
calendar
year
and
therefore
we're
trying
to
set
up
our
okrs
and
understand
sort
of
what
our
goals
are
as
a
team.
Because
of
that,
we
are
working
on
a
handful
of
objectives
and
themes
here
that
we're
sort
of
planning
for
we're
spending
a
lot
of
time,
ensuring
we've
planned
them
out.
We
understand
the
implications
of
those
and
how
they're
going
to
work
within
our
product.
There's
a
couple
specific
ones.
A
We
really
want
to
figure
out
a
better
means
of
allowing
Italy
to
push
back
on
traffic
surges.
We
have
this
issue.
Actually
it's
an
epic
out
here.
We've
done
some
planning
as
we've
gone
through
it
right
now.
When
Gilly
receives
spiky
traffic,
it
will
throw
errors
or
performance
will
degrade
at
sort
of
the
same
time,
but
there's
no
easy
way
for
us
to
push
back
on
the
calling
application,
and
thus
we
just
continue
to
get
these
traffic
spikes.
A
What
we're
looking
to
do
is
find
a
way
where
we
can
provide
feedback
to
the
application,
provide
error
messages
back
to
reduce
the
spiky
traffic
and
overall,
improve
the
the
user
experience
and
reduce
downtime.
So
that's
something
we're
kind
of
working
through
we're
trying
to
figure
out
the
best
way
forward
on
that
and
we
hope
to
plan
some
issues
to
actually
Implement
that
in
the
future,
another
one
that's
really
important
to
us
is
how
to
make
object,
pools
work
in
a
clustered,
World,
we're
beginning
this
effort.
A
Now
we
are
actually
prioritizing
a
couple
things
right
off
the
bat.
Actually,
it's
right
here
slightly
above,
where
we're
trying
to
convert
object,
pools
to
use,
create
Fork.
This
is
really
important,
as
it
will
dramatically
speed
up
the
forking
workflow,
which
is
an
important
workflow
for
some
of
our
users.
A
Along
the
same
lines,
we
want
to
make
sure
object,
pools,
work
as
well
as
possible
in
with
cluster
with
Italy
cluster
and
with
our
upcoming
distributed
architecture,
we're
planning
on
rolling
out
here,
moving
forward
and
again
we're
ways
out
on
that.
But
again
we
want
to
lay
the
groundwork
for
that
and
ensure
that
we
understand
what
an
object's
pose
will
look
like
and
how
we
can
utilize
them
for
maximum
performance
and
efficiency
and
productivity
for
our
customers
and
our
team
members.
So
that's
pretty
exciting
work.
A
As
I
mentioned
there,
we
are
working
through
15
8
on
some
of
the
groundwork,
with
that
to
convert
object,
pools
to
use,
create
Fork
I,
don't
know
if
this
book
will
complete
in
a
single
Milestone,
but
we
definitely
want
to
prioritize
this,
as
it
does
impact
a
lot
of
our
4K
customers.
So
that's
really
important.
A
We
also
have
a
bug
with
moving
repositories
and
it
causes
them
to
be
left
read
only
this
was
found
very
recently,
and
this
is
something
that's
important
to
our
workflow.
So
we
want
to
also
prioritize
that
358,
as
well
as
the
work
down
here
that
we
have
worked
with
our
quad
planning
counterparts.
I
realized.
That
is
a
bit
light,
but
if
there
are
any
questions
or
anybody
has
any
comments,
please
let
me
know
there
is
some
carryover
work
for
15
7
that
we'll
be
finishing
up
as
well.