No metrics on apps.sockets

Help
1

Hi, I have more than 30 servers running netada, all fully funtional with the same configs.

But today I found out a cluster of 3 with no metrics on apps.sockets.

I reinstalled netdata and nothing changed.
Netdata error logs doenst say anything about apps.sockets, or that something is broke.

How can I debug this issue?

Thanks for your atention.

2

Thank you for answers @Ricardo-Teixeira!

I will create a VM with CentOS 7 to try replicating the problem.

Now about the Kickstart, I forgot to ask something about it, did you use kickstart-static64.sh or kickstart.sh?

Best regards!

3

Thanks Thiago for chiming in!

4

Let’s continue the discussion here, unless we see that it’s a bug, so we create a bug issue on github.

1 Like
5

By the way, it is better to continue here or at github?

6

Yes I am.

  1. Yes
  2. KickStart
  3. 2Gb
  4. SELinux is disabled
  5. Yes I have. I will work on them and will give you feedback.

Thank you for your attention.

7

Hello @Ricardo-Teixeira ,

I am not sure if you are the same person that creates the issue on github, but I will assume that you are not, and I am copying the answer that I gave there:

I tested now on my machine and I can see without any problems the sockets, so I have few questions for you:

1 - Did you upgrade for the latest Netdata version(1.25.0)?
2 - Did you compile Netdata? Or did you use kickstart to upgrade?
3 - Which memory mode are you using with these Netdata agents?
4 - I already had problems with CentOS and SELinux that affects ebpf.plugin. I believe that apps.plugin also could have problems to read /proc depending of how it was configured. Can you see any reference to apps.plugin inside your audit.logs?
5 - Do you have any information related to apps.plugin inside your error.log?

I hope with your answers we will be able to fix this problem.

Best regards!

1 Like