MarcusAlzona wrote:We are rapidly approaching the two-year mark for this (IMHO) critical ATEM bug.
In a thread started in July 2020, Henry Kim/@bigfundj wrote up a detailed analysis of the problem in October 2020:
https://forum.blackmagicdesign.com/viewtopic.php?f=4&t=117185#p675388I added information in October 2020 as well, on how the ATEM Mini Pro ISO had a cascade failure during a wedding:
https://forum.blackmagicdesign.com/viewtopic.php?f=4&t=117185#p678105In the *year and a half* since then, I’ve avoided using the ATEM in any streaming situations where I did not have a dedicated direct Ethernet connection at the venue, greatly limiting its usefulness as an event streaming platform. My multiple ATEM mobile rigs have become almost worthless.
I was very excited about the direct ATEM-Extreme-ISO-to-4G/5G-phone-connection feature….but the next time I tried it in May of 2021, it blew up in my face yet again….after streaming successfully for 20-30 minutes at some point it detected temporary network congestion, immediately filled the cache, locking up and causing failure of the stream AND the ISO recordings being done at the time.
I pulled out my iPhone and gimbal and got a stream and recording of the last 1/4 of the event, but all recordings of the middle half were lost forever, due to the complete failure of the ATEM.
I haven’t heard any updates addressing this problem from Blackmagic.
Raphaël Jacquot wrote:I have the same problem when the MTU of the internet connection is smaller than 1500.
I suspect the (misguided / broken) IP stack in the atem to not work properly in that case.
This seems like a really good theory on a potential key factor of the underlying problem. Doing a cursory search, I saw some people quoting 4G/LTE connections often having maximum MTUs in the 1400 and 1300 range
eheiger wrote:me too...unplug the network cable from the atem for 2-3 seconds when cache starts to fill then plug it back in and it recovers every time. Not acceptable but it's the only way to make the streaming engine useable. have ticket open with support, they sort of hint engineers are aware so hopefully they are tuly aware. I provide feedback to them via the ticket to help them narrow down the problem. I'm convinced it's a bug, my internet has no issues when this happens. Have fios 300/300 and barely using it. Constant ping to my provider ny.castr.io showed no issues or changes when cache fills. When you take the network cable out it seems it resets the ip stack because cache will empty very quickly and the stream rolls along flawlessy afterwards.
I did another event in the fall of 2021 using the ATEM Extreme ISO, and now that I think back at one point I did start getting the cache full syndrome, even though it was wired (but to a slightly congested Venue network. Thankfully, unplugging and re-plugging the Ethernet cable did eventually get the streaming cache un-frozen before total lock (and ISO recording loss) started to occur. But it really shouldn’t be this way!
Two years is a very very long time for this known critically fatal problem with a key functionality(streaming and recording) to go without being addressed. It is concerning that the issue can be documented and debugged, by Blackmagic customers and ATEM owners wishing the device to be successful, on the company’s user forums….but with no meaningful feedback or mitigation plans from the company.
It really feels like it’s being ignored or de-prioritized for some reason, and it shouldn’t be that the only solution for action is to document these issues in reviews on external vendor sites such as Amazon, B&H, etc. I hope that Blackmagic will try to address this issue ASAP.