New install of OV running at 100% CPU Useage. What's wrong?

Issues with deployment
1

Till now we have been using a single KMS, which we have tested to handle up to 60 watchers without problem. Our KMS Linux server has 2vcores and 7GB of Ram. Yesterday we did a presentation hosting 26 watchers using our single KMS and CPU useage was less than 30%. Screenshot attached.

We recently installed OpenVidu on a second separate Linux server and been doing tests over the last few days to assess its capabilities and hopefully integrate OV into our project. Our test OV media server has the same 2vCores and 7GB ram, with OV installed. In testing, we find that even with no load on it, the CPU is running at 100% useage! Screenshot #2 attached. I know that the OV Docs state: “ on the cluster, have at least a minimum of 2 CPUs and 8GB of RAM, and a generous network bandwidth.” …but one would think 7GB of RAM would be enough to keep it going at least with no load.

So I have to assume something is not working correctly but we are not sure: (i) whether we are not configured correctly, or (ii) we are configured correctly but don’t have enough server power for what OV needs, or (iii) it’s something else.

Anybody have any ideas on what we may be doing wrong, and how we fix things and get back on track?

2

You have to take care to close ports with a Firewall as recommended in deployment procedure. For example UFW. If you deploy on a cloud, security groups are enough.

We have been reported that if you install OpenVidu without closing ports from outside, someone can connect to your redis database and hack it.

Review what is the process with 100% usage.

3

Here is info from server monitoring:
Process Information: redis-server CPU 196 %
OV works in Docker container

4

kurento 196 % /tmp/kdevtmpfsi

5

You have been hacked

In next OpenVidu version we will have redis password protected to avoid this issues with open ports. But please, close all not used ports.

6

After firewall enable and allow/open all ports from OV documentation can’t start:
./openvidu start
Creating network “openvidu_default” with the default driver
ERROR: Failed to program FILTER chain: iptables failed: iptables --wait -I FORWARD -o br-44be968f5c32 -j DOCKER: iptables v1.6.1: Couldn’t load target `DOCKER’:No such file or directory

Help please :slight_smile:

7

It seems your firewall installation has broken docker in some strange way. Nothing related with docker.

We have installed OpenVidu in ubuntu with UFW without problems.

I recommend you to reinstall everything from scratch.

Regards

8

Micael, Svet says he only can do 3 replies but he sends this msg:
After system reboot (Ubuntu 18.04.4 LTS) all works :slight_smile: thank you for firewall info - now CPU is in normal range (2 - 3%)
Many thanks!!!

2 Likes