Multicast-Unicast a...
 
Notifications
Clear all

Multicast-Unicast and xLights dummy setup

12 Posts
5 Users
0 Reactions
14.7 K Views
Wolfie
Posts: 2225
Topic starter
(@wolfie)
Noble Member
Joined: 11 years ago
[#41965]

A question sprang to mind after the expo.

I previously ran unicast only and not likely to change. Most of the presentations looked like they would be most appropriate to multicast.

In nearly every setup there were just a block of, say, 50,000 channels added as a dummy starter. I presume this is aimed at a multicast environment.

As stated, I run unicast and it may well be that universe 1, channel 1 will likely exist on 3 or more IPs.

Whats the procedure to set up the dummy channels in xLights to begin working on sequences such that its easiest to actually map to real IP and real controller and real lights so that xLights can activate them during testing (output to lights). I would like to be able to see blinkies while starting my sequencing until I get my brain swapped around what things will actually look like in lights compared to the preview on screen.

My intention is for at least one RasPi running FPP to take care of the show this year (another first for me). There will be 2 F16s and likely a J1sys D2 (e131 to two universe DMX bridge). Not sure but possibly one or more J1sys P12's but those may be retired. My intention is to run the FPP Pi to a switch then to each controller. I am hoping the FPP can use the WiFi dongle to connect to my home network for uploading and management but output on the wired eth0 port for sending packets.


11 Replies
Barnabybear
Posts: 1583
(@barnabybear)
Noble Member
Joined: 15 years ago
My intention is to run the FPP Pi to a switch then to each controller. I am hoping the FPP can use the WiFi dongle to connect to my home network for uploading and management but output on the wired eth0 port for sending packets.

Hi, yes that works I'm useing that set up. Worth noting though, if you use FPP as a master and remote you will need to remove the WiFi dongle when you have finished your setup and reboot or the remote will not work.
Tec bit: If there is a network connection on wlan0 by default (at the moment) the sync packet is sent over wlan0 not eth0. If you remove the WiFi dongle and reboot it will send the sync packets over eth0. The username and password and saved so if you need to change any settings you can just plug the dongle back in. I believe in the next update there may be an option to set which network connection is sent over.


Reply
Setarcos
Posts: 169
(@setarcos)
Estimable Member
Joined: 12 years ago
Barnabybear wrote:Hi, yes that works I'm useing that set up. Worth noting though, if you use FPP as a master and remote you will need to remove the WiFi dongle when you have finished your setup and reboot or the remote will not work.
Tec bit: If there is a network connection on wlan0 by default (at the moment) the sync packet is sent over wlan0 not eth0. If you remove the WiFi dongle and reboot it will send the sync packets over eth0. The username and password and saved so if you need to change any settings you can just plug the dongle back in. I believe in the next update there may be an option to set which network connection is sent over.

Under normal circumstances using the wireless interface (wlan0) for sync should be fine. A few folks have reported issues with noticeable delays on the slave device, but these are most likely due to wireless network congestion and/or poor WiFi signal reception. If you really must remove/disable the wireless dongle on the slave, you can add a route through the master's WiFi interface so you can still manage the slave device from your home network w/o plugging the WiFi dongle back in.

I know there was a discussion about making the interface used for sync selectable, but I haven't seen any code commits related to this yet.


Reply
Wolfie
Posts: 2225
Topic starter
(@wolfie)
Noble Member
Joined: 11 years ago

I am hoping not to get into the master-slave situation this year. This will be my first year with xLights and FPP so I am still holding onto my water wings for the time being.

I am surprised its not an option already to be able to choose the interface to which to send the sync packets. I am not overly sure the wifi will be sufficiently strong out in the front yard. I have aluminum siding so that attenuates the wifi pretty well but some gets out as the router is upstairs and can get through the roofing.


Reply
Setarcos
Posts: 169
(@setarcos)
Estimable Member
Joined: 12 years ago

The sync packets themselves are tiny, but anything that would induce large delays or continuously cause dropped packets can mess things up. One thing many people don't realize is that the total available wireless bandwidth on a typical home WiFi router is shared across the devices connected to that router (SU-MIMO). So when you fire up Netflix in the living room while your kids are watching Youtube, you may not see any significant issues during video playback because the protocols used to deliver those streams are far more tolerant to latency and have significant buffering to smooth out any network hiccups that often occur. On the other hand, if you are running anything latency-sensitive while this is going on even between two clients that are generating local network traffic, major problems can occur. A simple (albeit not entirely accurate) test you can perform to gauge jitter and packet loss is by pinging between the two devices you are concerned about. The results can sometimes be surprising. The latest 802.11ac routers support MU-MIMO which effectively reduces the number of WiFi clients sharing the wireless bandwidth between each other and the WiFi router and can help minimise the effects of this sort of thing, but the cost and the limited number of 802.11ac w/MU-MIMO WiFi dongles supported by the Linux kernels on the Pi and BBB/BBG used with FPP make this somewhat impractical right now.


Reply
Barnabybear
Posts: 1583
(@barnabybear)
Noble Member
Joined: 15 years ago

Hi, I went for that setup for ease. I aleady have an old router setup as an AP / switch with DHCP mounted on the front of my house and cabled back to my office. This is on it's own network, I also used to have a show PC on this but it died. When this happend I replaced it with a Pi running FPP and switched to sequencing on my own PC that is on the home network and wirelessly connected to the Pi through a WiFi dongle. This works well for me as I can sequence on my nice PC but transfer the files to FPP and I don't have any show trafic on my home network. It was only last week when I tried to run some remotes that it became apparent that the sync was going over wlan0, but it's an easy fix so not a problem as everything else works so well.


Reply
Wolfie
Posts: 2225
Topic starter
(@wolfie)
Noble Member
Joined: 11 years ago

I want no lighting traffic on the internal network short of testing and infrequent hits to the http config pages as necessary. I have two Kodi RasPi boxes and the o/l watches netflix on her tablet and the girls youtube on their tablets. So there is quite a bit flying around. One kodi is wired while the other is wifi as are the tablets. I also have 2 RasPi running as security cams via Wifi.

If and when I go to a master-slave situation, I will put in a dedicated router and place it out near the display area for the various pi and bbb to connect to. This way there will be zero issue with netflix, Kodi and youtube and keep the lighting traffic on its own subnet.

The initial question wasn't really directed at the wifi FPP setups. I was more asking about null channel configs in xLights relative to running in a unicast environment.


Reply
Posts: 1015
(@gilrock)
Noble Member
Joined: 14 years ago

You can't control lights from xLights using NULL channel outputs. They are really meant for channel placeholders to allow you to write data to a block of channels like for a P10 panel. I'm setup where I can easily switch between unicast and multicast. I got a small cheap router to use as a show network. I wanted my controllers to be configured as DHCP because I don't want problems if I connect them to a different subnet but I also wanted to be able to memorize a hard-coded IP address for each controller. So I solved that by setting them as DHCP but in the router I created a specific IP addresses to be assigned to the MAC addresses of my controllers. So they always get the same IP when plugged into my show network. Using that scheme I've easily switched between unicast and multicast. Never saw much difference for my channel count.


Reply
Wolfie
Posts: 2225
Topic starter
(@wolfie)
Noble Member
Joined: 11 years ago
Gilrock wrote:You can't control lights from xLights using NULL channel outputs. They are really meant for channel placeholders to allow you to write data to a block of channels like for a P10 panel. I'm setup where I can easily switch between unicast and multicast. I got a small cheap router to use as a show network. I wanted my controllers to be configured as DHCP because I don't want problems if I connect them to a different subnet but I also wanted to be able to memorize a hard-coded IP address for each controller. So I solved that by setting them as DHCP but in the router I created a specific IP addresses to be assigned to the MAC addresses of my controllers. So they always get the same IP when plugged into my show network. Using that scheme I've easily switched between unicast and multicast. Never saw much difference for my channel count.

Thank you, but I was fully aware you can't control them via null channels. They are nulls, I get that they don't go anywhere. I am asking whats the best arrangement to adding null channels NOW so that LATER (say a month or two) as I get props constructed/rebuilt, I can easily remap the null channels to real controllers and real pixels for testing and visualization.

I have zero experience with xLights short of what wonderful tidbits of details you managed to pour into my brain at the expo (thank you by the way). As such I have no foundational experience on which to base the initial null setup. One big block of 50k didn't seem too friendly to be able to tear apart into real universes and channels going to real IPs. Maybe it is. I just don't know. So I am asking how to best set up the null blocks to facilitate migration into the real world later. Maybe 50k as one block is fine. Would 5 blocks of 10k be better? Or another option? The only given right now is its almost 100% I will utilize two F16s, one for house elements, one for yard elements. And the ECG-D2 for the floods (yea, I know the F16 does DMX, I already have connectors done for the D2 but I will migrate them to the F16 at some point just not right away).


Reply
Posts: 1015
(@gilrock)
Noble Member
Joined: 14 years ago

It's so easy to recreate the setup just one big block will be fine. The main thing is don't use the "Output" number to map your start channels on the models because if you change outputs then you are moving the start channel. So use the absolute channel numbering or use model chaining or the new universe/channel assignment options for your start channels. You can blow away the setup and recreate it without it hurting your models.


Reply
Page 1 / 2
Share: