Troubleshooting inconsistent performance?

[etfz]

Hi,

I've been running a number of different NetBox (using Docker) versions over a few years, and for as long as I can remember, I've had issues with pages loading slowly, up to some ten seconds. On good days, pages load instantly, so the host clearly has the capacity. For example, just now I had the home page take ages to load, and then a couple of minutes later, it loads instantly. Sometimes it's somewhere in between. I think all pages might be affected at the same time, and it includes asynchronous requests such as for field autocomplete lookups.

I run a number of other Docker containers on the same host, and have never witnessed similar issues on any of them. Where can I start troubleshooting this?

[mattg66]

I have the exact same issue. Most requests are snappy, but seemingly randomly requests will take a while, and the main python docker image will use high amounts of CPU until the request is eventually completed

[etfz]

Glad to hear I'm not the only one, at least! Had not looked into the matter too closely before, but I can confirm that I am seeing the same behaviour with the main netbox container showing high CPU utilization for the duration of the requests.

[kkthxbye-code]

It's likely a netbox-docker related issue. The default configuration is like so:

https://github.com/netbox-community/netbox-docker/blob/477decd80af739e1eaa03c20a9ccd3ddd699818a/docker/nginx-unit.json#L69-L73

      "processes": {
        "max": 4,
        "spare": 1,
        "idle_timeout": 120
      }

This means that only one process will stay active and up to 4 can be spawned in total. After 120 seconds it will go down to 1 process. The spawn time is not instant at all, so if unlucky you might have to wait.

If memory is not an issue, try to replace the above with something like:

"processes": 4 # Or 8 or whatever.

This will spawn a static amount of processes on launch. You will have to mount the file into the container though. Something like:

my-nginx-unit.json:/etc/unit/nginx-unit.json:ro

[mattg66]

Thank you for your reply, will give this a try

[mattg66]

This can actually be achieved by modifying the file at docker/nginx-unit.json in the official netbox-docker repository

[kkthxbye-code]

Only if you build your own image. The file is copied during image build.

[mattg66]

Ah yes true, I am doing that so that is working for me, maybe this could be an environment variable

[mattg66]

Can confirm this has fixed the issue, I increased to a max of 16 and spare of 8. Thank you for the pointers @kkthxbye-code