[jitsi-users] Scaling the jitsi infrastructure on AWS


#1

Hi,

We are following the ideas explained in this presentation to scale our
jitsi infrastructure on AWS:

https://app.box.com/shared/dq7eqqbr4xmm0l5zwkvotp3twpovzrqs/1/7747265085/67176057741/1

(got it from here: https://wiki.geant.org/pages/viewpage.action?pageId=53118178)

During the process of configuring our environment the following doubts arouse:

- whenever a new videobridge is added to a shard, there is the need to
reconfigure prosody (new admin and new component with correct
password) and reload it. How are you doing it? Are you using any sort
of auto-discovery solution?

- how to scale down a specif shard without impacting users? Is it just
a matter of shutting down a videobridge instance and then jicofo takes
care of moving the conferences to another bridges? Or is there some
other (smarter) approach?

Cheers,
-- Alex


#2

Hi,

- how to scale down a specif shard without impacting users? Is it just
a matter of shutting down a videobridge instance and then jicofo takes
care of moving the conferences to another bridges? Or is there some
other (smarter) approach?

There is a graceful_shutdown.sh script in the bridge. This marks the
bridge that its in graceful shutdown and reports it to jicofo, so no
new conferences will be created on this bridge. And it will wait for
conferences to end and then it will stop the bridge.
About adding new bridges, you need to add some logic to your
deployment environment to be notified to add new component to prosody,
or you can pre-configure some number of components in prosody and
re-use them with new bridges.

Regards
damencho

···

On Tue, Sep 6, 2016 at 4:21 PM, Alex Zanetti de Lima <trustlix@gmail.com> wrote:


#3

Hi,

Is there some one else leveraging that same infrastructure who could
help me with the two points below?

Thanks!
-- Alex

···

On Tue, Sep 6, 2016 at 6:21 PM, Alex Zanetti de Lima <trustlix@gmail.com> wrote:

Hi,

We are following the ideas explained in this presentation to scale our
jitsi infrastructure on AWS:

https://app.box.com/shared/dq7eqqbr4xmm0l5zwkvotp3twpovzrqs/1/7747265085/67176057741/1

(got it from here: https://wiki.geant.org/pages/viewpage.action?pageId=53118178)

During the process of configuring our environment the following doubts arouse:

- whenever a new videobridge is added to a shard, there is the need to
reconfigure prosody (new admin and new component with correct
password) and reload it. How are you doing it? Are you using any sort
of auto-discovery solution?

- how to scale down a specif shard without impacting users? Is it just
a matter of shutting down a videobridge instance and then jicofo takes
care of moving the conferences to another bridges? Or is there some
other (smarter) approach?

Cheers,
-- Alex


#4

Hi,

- how to scale down a specif shard without impacting users? Is it just
a matter of shutting down a videobridge instance and then jicofo takes
care of moving the conferences to another bridges? Or is there some
other (smarter) approach?

There is a graceful_shutdown.sh script in the bridge. This marks the
bridge that its in graceful shutdown and reports it to jicofo, so no
new conferences will be created on this bridge. And it will wait for
conferences to end and then it will stop the bridge.

Awesome. Didn't know about this script. Will take a look on it.

About adding new bridges, you need to add some logic to your
deployment environment to be notified to add new component to prosody,
or you can pre-configure some number of components in prosody and
re-use them with new bridges.

Thanks a lot, Damian!

···

On Mon, Sep 12, 2016 at 12:55 PM, Damian Minkov <damencho@jitsi.org> wrote:

On Tue, Sep 6, 2016 at 4:21 PM, Alex Zanetti de Lima <trustlix@gmail.com> wrote: