ProSoundWeb Community
Sound Reinforcement - Forums for Live Sound Professionals - Your Displayed Name Must Be Your Real Full Name To Post In The Live Sound Forums => LAB: The Classic Live Audio Board => Topic started by: boburtz on June 13, 2019, 01:15:08 AM
-
I just experienced an issue with dante that I never never before seen.
2x CL5 consoles (one has redundant netgear gs110tpV2 switches), 2x RIO3224d with redundant switches (also netgear gs110tpv2), 4x Lab.Gruppen PLM20000q with redundant switches per pair (GS108tV2), one windows PC running dvs.
I am having intermittent issues with the clock slipping on the PLM amplifiers randomly (none of the amps are immune), when that happens, the amps that is slipping stops passing audio (!).
All of the Yamaha equipment is functioning flawlessly. I removed DVS from the network and it seems like the problem has gone away. That doesn't make sense to me, so I restarted dvs and unpatched everything from the PC. That didn't fix the issue. Anyone ever have issues running dvs along with PLM amplifiers on the dante network?
Before you spend time asking about EEE or QoS, all of the settings are set properly. It doesn't seem to be a cabling issue because wen I initially removed dvs from the network, I couldn't make the problem return, no matter how much I wiggled the cables or monkeyed with stuff.
I have used this same exact configuration in the past with only one console and with zero problems. Could it be too much network traffic for the 100mbps ports built into the PLM amps?
-
Assuming each amp is only getting between 2 and 4 channels of audio, the 100mb connection shouldn't be an issue, since each Dante channel only uses 6mbps of bandwidth at 48k.
My money is on DVS. I've said it before and I'll a it again, I have had very random issues, which sounded like clocking/sample rate issues when using DVS and have lost faith in DVS. I have been unable to reproduce the issues I had. But once DVS was removed from the equation I had no issues, and have had no issues with actual Dante hardware.
-
Are you 2 using the DVS for recording only or are you returning audio to the system thru DVS? I know next to nothing about Dante but just did a show with a Yamaha CL5 and was thinking of next time using DVS to record the show.
-
Assuming each amp is only getting between 2 and 4 channels of audio, the 100mb connection shouldn't be an issue, since each Dante channel only uses 6mbps of bandwidth at 48k.
My money is on DVS. I've said it before and I'll a it again, I have had very random issues, which sounded like clocking/sample rate issues when using DVS and have lost faith in DVS. I have been unable to reproduce the issues I had. But once DVS was removed from the equation I had no issues, and have had no issues with actual Dante hardware.
I'm inclined to agree, I've seen several threads pop up here and elsewhere with odd DVS issues. Some people solved it with restarting the software, others just removed it entirely as you did.
-
Are you 2 using the DVS for recording only or are you returning audio to the system thru DVS? I know next to nothing about Dante but just did a show with a Yamaha CL5 and was thinking of next time using DVS to record the show.
I've had issues with DVS with both recording to Protools and playback from QLab (not running at the same time). Once O switched to hardware Dante interfaces (Focusrite Red 4Pre and the Protools HDX card) everything worked fine.
In one case with playback from QLab, it was running fine for about 6 hours for an all day dance recital, then it started outputting hard digital noise right in the middle of the show after performing flawlessly for the previous 6 or so hours. I had to reboot the whole system (computer, console, RIOs and switches) to solve the problem. And then it worked fine for the last 2 1/2 hours.
-
Screenshots of Dante Controller with DVS connected would help, device info tab, clocking and latency monitor for each device. What's your IP scheme? IGMP? are you multicasting?
-
Screenshots of Dante Controller with DVS connected would help, device info tab, clocking and latency monitor for each device. What's your IP scheme? IGMP? are you multicasting?
I can get screen shots next week when everything is back in the shop. The only thing that shows up as an alarm is the PLMs lose clock sync. All of the latencies are fine. I do have one multicast flow set up for four of the dante output channels from the FOH console, as they are routed to the rio and to the plms. I tried multicasting everything from the rios to send all inputs to both consoles and dvs, that made things even worse, which is not what I expected. It's my understanding that snooping is not necessary with such a small network. The only traffic on this network is dante, lake controller, and dante controller.
The dante ip scheme is auto, no static ips, no snooping. There are no ip address conflicts.
Again, I have performed many (MANY) multitrack recordings with DVS and the QL5 using both Reaper and Nuendo Live with FLAWLESS results, and I believe that will still be the case with the CL/Rio system, but it will not work if I am using dante to get signal to the amps (my optimism is showing). At this point, I will be using analog to get signal to the amps and will not be using dvs for this weekend. Everything else works fine. Are the PLMs more sensitive to clock deviation than other dante gear?
-
Are the PLMs more sensitive to clock deviation than other dante gear?
100mb interfaces can become saturated in a Dante network if there is multicast traffic and no IGMP snooping. This is probably why you saw the problem worsen when you tried multicasting everything.
In your situation my guess is either the PLMs are being saturated or the multicast clock data is not getting through to them. Best to wait for the controller screenshots / log.
In the meantime are you running anything AES67 or do you have internet connection sharing on the machine running DVS?
-
In the meantime are you running anything AES67 or do you have internet connection sharing on the machine running DVS?
No AES67, no internet anywhere on this network. I do have the console control network piped into the primary switch, as well as a wifi access point with hidden SSID, which is what I use to connect both stagemix and lake controller. This has never given me a problem in the past. I removed those items and it did not rectify the problem.
-
PTP (clock) is multicast.
The fact your system falls over when multicasting suggests you should enable IGMP snooping.
This is Audinate’s standard recommendation (and general multicast networking rec). It’s the QoS that’s not necessary on a small/closed network. When you don’t do IGMP, you flood all devices with traffic.
-
PTP (clock) is multicast.
The fact your system falls over when multicasting suggests you should enable IGMP snooping.
This is Audinate’s standard recommendation (and general multicast networking rec). When you don’t do IGMP, you flood all devices with traffic.
I did not know this, and am completely unfamiliar with IGMP snooping as it relates to dante and switch settings. Is it a simple process of ticking a check box to "allow" or is the setup more involved? We have a rental operation and the surfaces sometimes go out without rio boxes (some clients have their own rios), so if the settings are device specific, I'll have to figure out how to most efficiently set it up so that it works with any and all devices that may be on the network.
-
IGMP Snooping in Dante (https://www.youtube.com/watch?v=xQh-iJJP07c)
Shure's take on IGMP Snooping (https://www.shure.com/en-GB/support/find-an-answer/dante-networks-and-igmp-snooping)
-
100mb interfaces can become saturated in a Dante network if there is multicast traffic and no IGMP snooping. This is probably why you saw the problem worsen when you tried multicasting everything.
In your situation my guess is either the PLMs are being saturated or the multicast clock data is not getting through to them. Best to wait for the controller screenshots / log.
In the meantime are you running anything AES67 or do you have internet connection sharing on the machine running DVS?
I did not realize that there was a newer version of DVS than the one I was using. I updated that and eliminated all of the unnecessary multicasting and it is working fine now. FWIW, the frequency offset for the PLMs was bouncing around all over the place, from ~+175 to ~-150 or so. Everything else on the network was pretty steady, occasionally fluctuating a point or two. Once I removed the multicasting and updated DVS, the PLMs are holding at a steady ~26ppm offset.
Thanks for the tip on snooping, it's something I hadn't looked into. In the future if I need to set up multicasting I will employ IGMP snooping.