►
From YouTube: GitLab 15.7 Kickoff - Enablement:Geo
Description
Planning issue: https://gitlab.com/gitlab-org/geo-team/discussions/-/issues/5049
Geo group past kickoff Videos: https://www.youtube.com/playlist?list=PL05JrBw4t0KofRx080zwe0VxUEv5yG_nO
Presentation by: Sampath Ranasinghe, Sr. Product Manager, Geo
A
So,
let's
get
started,
we've
been
on
a
long
journey
to
moving
existing
data
types
to
the
self-service
framework.
The
self-service
framework
provides
an
easy
way
to
add
new
data
types
to
the
Geo,
replication
and
verification
process.
It
also
empowers
other
developers
and
teams
within
gitlab
to
to
add
new
data
types
to
to
this
process,
as
in
when
they
develop
new
features.
A
So
in
order
to
reduce
our
maintenance
burden
and
also
manage
code,
one
piece
of
code
when
it
comes
to
this
replication
and
verification
process,
we
want
to
consolidate
under
this
self-service
framework
and
and
part
of
that
process,
is
moving
existing
data
types
that
use
a
different
type
of
logic
and
bringing
them
into
into
this
framework.
A
So
in
this
Milestone
we
continue
the
work
that
we've
been
doing
in
moving
projects
and
wikis
into
the
self-service
framework,
and
this
has
been
somewhat
of
a
journey
for
us
over
the
last
couple
of
Milestones,
but
we
are
making
very
good
progress
on
on
getting
this
into
the
self-service
framework
and
and
we'll
continue
that
effort
in
this
milestone
next
up
is
fixing
the
timeout
issue
related
to
pushing
via
SSH
from
a
secondary
site.
A
Currently
we
have
a
60
second
Timeout
on
proxied
requests,
push
request
from
the
primary
from
the
secondary
side
to
the
primary
site.
As
you
can
imagine,
this
creates
problems
for
customers
who
have
large
push
requests.
They
eventually
result
the
the
connection
between
the
primary
and
the
secondary
times
out,
resulting
in
a
failed
push
request.
A
The
solution
for
this
will
likely
require
an
architectural
change
to
to
Geo
and
how
we
handle
a
SSH
proxy
requests
and
we'll
entail
a
significant
level
of
effort
in
order
to
address
the
ultimate
goal
is
to
remove
the
restrictions
on
size
of
SSH
push
requests
that
are
sent
via
the
secondary
sites.
A
Next,
we
continue
cleaning
up
the
the
terminology.
We
made
a
shift
to
to
adopt
to
deprecate
node
in
favor
of
sites
or
somewhat
change
change,
the
meaning
of
node
and
site
in
our
documentation
and
in
our
product.
So
in
the
new
terminology,
a
node
denotes
a
single
server
and
a
site
is
a
collection
of
one
or
more
nodes
are
running
a
single
instant,
single
gitlab
application.
A
So
we
will
be
tackling
updates
to
the
gitlab
RB
file
in
this
in
this
Milestone,
alongside
some
updates
to
the
API
to
to
make
the
alignment
and
then
finally,
we
will
be
looking
to
make
some
up
improvements
to
ntp.
Handling.
A
Jwt
authentication
requires
clocks
between
all
involved
machines
across
geosites
to
be
in
sync.
So
therefore
we
we
have
pre-def.
We
have
a
predefined
ntp
server
at
pool.ntp.org,
which
cannot
be
easily
overwritten,
as
you
can
imagine.
This
poses
a
challenge
for
air-gapped
customers
whose
systems
are
not
able
to
reach
this
address,
but
when
ntp
checks
fail,
the
error
message
does
not
indicate
which
address
the
system
was
attempting
to
contact
to
retrieve
the
time
information.
So
again,
this
this
helps
with
troubleshooting
and
and
are
fairly
easy
fixes.
A
That
will
improve
the
experience
for
our
customers
and
and
we'll
be
looking
to
tackle
this
in
this
milestone.
We
will
also
start
to
look
into
improving
some
of
the
observability
experience
of
for
forgeo
when
it
comes
to
the
replication
and
and
the
side
statuses.
This
work
is
is
beginning
with
some
improvements
to
the
terminal,
outputs,
making
things
a
little
clearer
and
increasing
The
Clarity
on
on
when
scientists
from
site
promotion
is
completed,
and
then
we
have
a
number
of
bugs.