I often have the BBC R3 320kbs stream running while I’m doing something else. A few times over the years it’s just stopped, but I can manually restart it. No great problem. Over the last few weeks, it’s dropped more frequently - maybe three times over a couple of hours. Hmm. Maybe it is a problem…
A quick look at the logs suggest the stream is being interrupted on the ‘internet’ side, rather than anything local, but (to me) doesn’t tell where/why.
10/14 21:26:54 Warn: [easyhttp] [8933] GET https://as-hls-uk-live.akamaized.net/pool_904/live/uk/bbc_radio_three/bbc_radio_three.isml/bbc_radio_three-audio%3d320000.norewind.m3u8 web exception without response: socketmsg (SocketError): The SSL connection could not be established, see inner exception. The SSL connection could not be established, see inner exception.
Chain is Virgin fttp 250mbs to Virgin Hub 3 (modem mode) to MikroTik hAPax3 to NUC/W10pro/Roonserver (wired) to Naim ND5XS2 (wireless). No issues with continuous stream from Tidal. Cloudflare diagnostics suggest connection is all good.
My internet connection is plenty fast, but the VM Hub 3 is a DOCSIS modem, which is apparently not brilliant for latency. It’s possible it’s an SSL timeout, but the error message (same one every time) doesn’t tell whether it’s Roon being impatient, or the BBC servers giving up.
I have a set of logs from the frequent-drop period.
Describe your network setup
Virgin fttp 250mbs to Virgin Hub 3 (modem mode) to MikroTik hAPax3 to NUC/W10pro/Roonserver (wired) to Naim ND5XS2 (wireless).
I’d first test out simplyfing your local network setup. Set up a direct ethernet connection from your Roon Server directly to your primary router - do the same for any of your endpoints.
I’d also double-check your windows firewall settings, since you are using a Windows PC, I would verify that both Roon.exe and RAATServer.exe have been added as exceptions to your Windows firewall.
You can use these instructions to add the exceptions and the executables themselves would be located in your Database Location/Application folder path.
I would also add these exceptions to any Antivirus or other Firewall blocking applications you may have and ensure that you connected to your network via a Private network, not a Public one, see this guide for more information.
If you still run into issues after the above, please share a specific track name where you experienced the issue, and we’ll take a closer look!
There is no specific ‘track’ - we are talking about the BBC Radio 3 stream in ‘live radio’.
The Roon Server is directly conected to my primary router (the MikroTik hAPax3).
The radio stream is fine most of the time - as are streams from Tidal - so Firewall configuration is ok. The local network is private.
There’s an example of the error message in the post above. I can upload a complete set of logs if required.
Wired connections to the router for endpoints are impractical, but if the team think it will help I can connect a local DAC and leave a BBC R3 stream running.
Edit - I’ve managed to enable local output on the NUC - had to reboot because remote desktop insists on enabling remote sound and won’t release it! I’ll kick off an R3 stream tomorrow morning and see if it drops… can’t make the system much simpler.
Hi @AndyR,
Thanks for getting us those timestamps. We were able to find the dropout events in your logs. We have seen users have a better experience in the past if they change their Router’s DNS servers from the ISP provided ones to Cloudflare DNS, Quad9 or Google DNS. Can you please give this a try and let me know if it helps?
The NUC running Roon Server is already configured to use Cloudfare WARP - I’ve just checked that it is currently enabled; all appears to be ok, so should already be using Cloudflare DNS.
Ruling out address allocation issues is just a step we take to clarify conditions for due diligence.
Here’s what we can see in logs:
The BB3 station is returning steady 200 status codes on the URL each time RoonServer talks to it; however, metadata and discovery service requests all begin to time at the precise second that playback drops off. All upstream services other than the stream appear down.
The Naim first pauses after the stream abruptly ends, and then tears down the TCP connection due to inactivity.
At this same time, mDNS appears to fail across the Cast ecosystem, including to a Pixel Roon controller.
None of this appears causally related to the issue. However, these network issues all appear when the bandwidth and throughput might be undergoing stress, so we can’t fully rule out the modem/router.
Evidence points to this explanation more than anything, but we want to escalate this to Devs to determine whose side is causing the breakup. If this were endemic to BBC station URLs we’d see broader reports. It’s possible that BBC’s HLS implementation isn’t playing well with Roon. Did you notice any changes from release to release, or has this changed gradually over the last few weeks?
My memory says that there have always been very occasional drops. It’s the last month that has seen clusters where the stream will drop every few minutes, with no apparent correlation with any observable network issues.
However - you mention a pixel control device… my Pixel 4a died last month, and I’m now using a Pixel 7a - but I think the first cluster occurred a few days before that.
I did try to stress the bandwidth a few days ago - two 192/24 Tidal streams and two BBC radio streams concurrently to different endpoints for about an hour. No drops.
I’ve just checked. I think I first noticed increased frequency of drops on October 2nd - I made a copy of the log files then. That’s a day or so after an early-access release, but likely (it’s a Wednesday… ) the first time I’d had the R3 stream running since the release.
The release notes don’t mention any changes to live radio explicitly, but were there any updates that may affect connectivity?
… while attempting to look into how hls works I’ve rediscovered the forum threads relating to 30 minute delays on the Danish radio hls streams; I think I deduce that there has been work on the Roon implementation of hls streaming in roughly the same timeframe that we’ve seen frequent drops on the BBC hls streams.
I know it’s as likely to be a coincidence as not, but are the issues related?
As a side note - currently running BBC R3 using the ‘new’ Naim implementation. Likely the same stream. No drops so far…
Hi @benjamin - thanks for confirming that. Incorrect deduction from misreading posts on the ‘Danish Radio’ thread then…
I’ve been running the BBC R3 hls stream directly to the Naim streamer for ~ eleven and a half hours today with no drops. Yesterday evening (uk) via Roon I had half a dozen drops in as many minutes and gave up. Obviously inconclusive, but to me that points to the issue being the Roon hls implementation rather than anything intrinsic with the stream. I’ll turn the volume down and leave the direct stream running overnight…
It’s inconsistent. It’s run via Roon for 12 hours or more with no drops, but also six in six minutes the other day. The stream I’m currently running directly to the streamer has now been running nearly 21 hours though.
I’m still thinking my (and a few others I think) issues may be latency related - with a Virgin Media Hub3 DOCSIS modem giving average latency around 15ms, occasional marginally longer latency could push a slightly impatient Roon implementation into dropping the stream.
Do you know what type of modem your fibre connection is using and/or what sort of latency/jitter you achieve?
For as-hls-uk-live.akamaized.net [104.86.110.224], I’m seeing a 10-minute average latency of 7.6ms, with 0.3ms jitter. Rare spikes (once or twice per 10 mins) to a max of 11.6ms.
)
