Bizarre high-user issues

Hi All,

We’re running JVB on a 10 core, 20 thread dedicated server with a gigabit pipe, 32GB ram and SSD storage. At around 30-40 users the site was unusable and couldn’t handle all of the video streams. I’m sure our resources are sufficient. Is there anything I am missing here?

Any help would be greatly appreciated.

If it helps, here’s the current system usage with just 10 users:
CPU [ 13.3%] CPU - 13.3% MEM - 11.8% SWAP - 0.0% LOAD 20-core
MEM [ 11.8%] user: 9.9% total: 31.4G total: 976M 1 min: 3.53
SWAP [ 0.0%] system: 3.2% used: 3.69G used: 0 5 min: 3.04
idle: 86.7% free: 27.7G free: 976M 15 min: 2.91

NETWORK Rx/s Tx/s TASKS 319 (767 thr), 1 run, 160 slp, 158 oth
eno1 22.8Mb 25.9Mb
eno2 152b 0b Systemd 7 Services loaded: 155 active: 155

There is a known issue where Firefox does not suppport simulcast and always sends HD resolution which is then forwarded by the videobridge to each client. Therefore, the client’s network may become overloaded if several clients are using Firefox. Not saying that this is definitely the issue you are facing but its the first thing that comes to mind.

You can try to set the resolution option in config.js to some low value or to add Firefox to the unsupported_browsers list in interface_config.js (you may also want to add safari there, I believe there are some problems as well).

1 Like

We’re all on chrome though.

So I have an update from my team. We had about 35 users in one room. The issue was that the connections were really poor, people cutting out or black screens. This is for everyone. Looking at the server stats we had plenty of bandwidth overhead. Any ideas?

What about the jvb sysctl settings, are they applied, about udp buffers? You can check whether there are packets dropped from the kernel.

Sorry I am new to this, where can I find those settings?

Its mentioned in the quick-install guide, see and

Thanks. Those are all set correctly. I can’t see any config issues based on what you showed me. We just ran a stream with 8 users and videos keep cutting out etc. Any more insight?

Thanks all for your help so far. Im happy to provide someone access to the server to have a look at everything if you have the time.

jashton: does that Gigabit pipe actually has a bandwidth more than 20 Mbps, or is it just a Gigabit pipe? Test your speed with speedtest-cli to be sure.

Yes we have gigabit symmetric upstream in the datacentre. Speedtest is giving me about 900Mbps both ways at the moment.

Hey there. Same here… With 5 users some get black screens or freezes. Server is running with like 20% load, has plenty of RAM left…

Yes same problem here! Very choppy video despite a high-end system and plenty of bandwidth. We love the platform but this is the last hurdle for us to get over! We are trying a conference of about 10 tonight so we will see how we go.

Check whether you have these settings:
It maybe the buffers are not set and the kernel is dropping the packets.