►
From YouTube: How to Load Balance Jitsi Meet
Description
Step-by-step tutorial on how to setup Jitsi Meet in a load balanced configuration to add redundancy and provide additional capacity. Jitsi Meet is a secure open source video conferencing platform you can setup yourself in minutes.
This guide assumes you have already installed Jitsi Meet: https://youtu.be/8KR0AhDZF2A
See more at https://jitsi.org
Try the free Jitsi video calling service at https://meet.jit.si
Get the code at https://github.com/jitsi
A
A
Now,
let's
see
what
the
basic
internal
structure
of
a
GT
meat
platform
is,
it
consists
of
several
major
parts
that
you
see
in
each
installation
and
also
several
optional
additional
components
that
provide
non
core
functionalities
like
zip
telephony
integration,
shared
real-time
text,
editor
conferences,
recordings
and
so
on.
In
this
tutorial,
we'll
focus
on
load,
balancing
the
platform
between
two
or
more
as
a
few
media
servers
or
selectively
forwarding
units.
These
servers
in
GT
meet
are
called
GT
video
bridge
and
are
one
of
the
most
important
parts
of
the
platform
the
video
bridge
hosts
each
conference.
A
Instead
of
mixing
all
incoming
media
into
one
single
stream
and
sending
it
to
all
peers,
like
the
traditional
MCU
servers,
do
the
jetty
video
bridge
only
relays
the
media
streams?
There
is
no
mixing.
The
advantages
are
that
this
way
it
is
faster,
lower
on
CPU
and
RAM
resources,
and
also
the
output
can
be
customized
for
each
participant.
For
example,
people
may
receive
videos
from
only
the
last
four
or
five
active
speakers,
and
the
quality
of
the
streams
has
their
size
may
be
lower.
A
If
the
receiver
has
lower
network
speed
in
the
default
installation,
each
GT
mid
front
end
has
only
one
video
bridge
running
behind
it,
but
that's
not
mandatory.
The
service
that
binds
all
the
other
parts
together
called
jacoco
can
manage
more
than
one
video
bridge
when
there
is
a
request
for
a
new
conference.
Jacobo
has
to
decide
which
video
bridge
to
call
for
hosting
it
in
the
standard
setup
the
one
that
comes
by
default
with
the
Debian
packages.
A
A
You
can
also
use
video
bridge
load
balancing
to
provide
failover,
but
remember.
Each
conference
is
hosted
on
only
one
video
bridge
and
if
this
bridge
goes
down
so
does
the
conference
when
the
participants
reload
their
browsers,
they
may
have
a
conference
that
looks
somewhat
like
the
failed
one,
but
underneath
technically
it
will
be
a
new
conference.
A
So
such
failover
will
work
above
the
conference's
level.
It
will
be
a
failover
for
the
whole
service.
As
far
as
the
very
conference's
are
concerned,
there
is
no
pain
over
regarding
the
load
balance.
It's
also
on
the
service
level.
Separate
conferences
can
be
balanced
or
moved
between
video
bridges.
Each
conference
is
born
lives
and
dies
on
the
video
bridge
that
chica
4
decided
to
direct
it
to
once
again,
the
video
bridge
load,
balancing
and
failover
are
on
the
service
level.
They
provide
availability,
non
interruption
of
the
whole
service.
A
A
A
Each
video
bridge
periodically
publishes
information
about
its
current
load
in
a
web
sub
note
on
the
XMPP
server.
Jacoco
is
subscribed
to
these
nodes
and
listens
to
the
changes.
All
you
need
to
do
in
order
to
have
a
load.
Balanced.
Video
bridges
is
to
configure
each
video
bridge
to
publish
its
statistics
on
the
pub/sub
node
in
the
internal
XMPP
server
prosody
in
our
case
and
by
default,
and
then
configure
check
offer
to
subscribe
to
this
node.
A
Let's
go
on
and
configure
our
first
GT
MIT
instance
for
a
load
balance
setup
by
adding
an
additional
video
grade
machine.
Now
we
already
have
two
virtual
servers
with
Ubuntu
1604
0
0
self.
Yes,
on
the
first
one
we
have
installed
GT
meet
as
it
comes
by
default
with
the
Debian
installation
packages
distributed
by
the
GT
upstream.
This
server
has
static
public
IP
and
a
DNS
name
pointed
on
the
second
server.
We
also
have
one
two
1604
with
just
a
GT
repository
added
and
nothing
additional
in
style.
A
A
A
A
A
The
next
part
of
the
procedure
config,
is
to
make
the
server
listen
on
the
public
interface
so
that
the
video
bridge
on
the
second
machine
can
connect
to
XMPP
and
register
as
a
component
by
default.
Prosody
listens
to
localhost,
so
we
add
the
following
lines
somewhere
in
the
main
config
file
setting.
The
component
interface
is
not
possible
from
a
separate
virtual
host.
It's
a
global
option,
so
we
open
the
main
config
file.
A
Remember
that
setting
the
components
to
connect
to
a
public
network
interface
means
that
you
will
have
to
configure
all
components
gqo
for
all
the
video
bridges
to
connect
to
the
public.
Address
of
the
server,
even
those
that
are
installed
on
the
same
machine
in
our
case
jacoco
in
the
first
video
bridge,
prosody,
can
take
only
one
IP
address
for
this
setting.
So
when
we
change
it
to
the
public
address,
the
local
installed
components
won't
connect
to
localhost
as
it
was
in
the
default,
is
now.
A
A
A
A
A
A
A
Once
all
this
is
configured
and
the
services
are
restarted,
you
should
see
in
the
G
couple
locks
in
bar,
lock,
GGG
computed,
lock,
details
about
the
video
pretty
choosing
for
each
new
conference
number
of
conferences
or
bridge
and
health
check
things
than
Chicago
sense
to
determine
which
video
bridge
is
available.
Here
are
some
typical
log
messages
that
you
may
wish
to
keep
an
iPhone
on
bringing
online
on
your
bridge.
It
registers
itself
as
a
component
in
the
X
EVP
server.
G
Kaval
locks
this
event
and
schedules
a
health
check
for
the
new
bridge.
A
Alternatively,
when
the
connection
with
the
bridge
is
lost
for
some
reason,
like
the
bridge
component
is
removed
from
XMPP.
Oh
check
off
fails
on
health
checks
on
both
you
see
in
the
log,
a
message
for
removing
the
bridge
and
stopping
its
health
checks.
When
a
new
conference
is
created
on
a
healthy
platform
with
at
least
one
working,
video
bridge
do
see
a
log
message
with
the
details
of
the
conference
creation,
then
you
can
see
on
which
bridge
it
will
reside
periodically.
There
will
be
messages
with
the
current
number
of
video
streams
on
each
bridge.
A
You
can
use
all
these
information
for
gathering
some
roasted
tea
sticks
or
which
is
more
important
as
an
alarm
for
problems
with
the
video
bridges.
Of
course,
there
are
more
powerful
and
complete
solutions
for
monitoring
the
usage
of
legitimate
platform
like
call
stats,
for
example,
but
you
can
always
use
the
log
file
for
system
troubleshooting.
A
Now
you
have
a
GT
mid
platform
with
two
balanced
video
bridges
and
you
can
use
the
same
procedure
to
add
more
to
add
a
new
video
bridge
just
do
the
following
on
the
main
server
add
a
prosody
component
for
the
new
video
bridge
with
a
subdomain
and
a
secret
string
and
then
we'll
load
the
prosody
to
make
the
config
changes
acting
on
the
new
server.
Add
the
GT
repository
and
install
GT
video
bridge
in
the
SIP
communicator
on
properties
file.