Poor Website performance

Manager UI
1

Hello Everyone,
Im still getting comfortable with Open Remote but have noticed a reoccurring issue that I for the live of me can’t find a solution for.

Im currently running the dev setup of Open Remote and am accessing the Manager UI through the local IP of the device OR is running on.

In the beginning the website ran really smooth but after a few hours of adding assets and agents etc. The performance declined drastically.
I think the main issue is related to the Websocket which often takes up to 40sec to respond upon reloading the page or saving changes to an asset.

Network from reloading the page

image
image1373×84 16.5 KB

Its unprobable that the issue is related to hardware since the device OR is running on still has many unused resources

If anyone has experience with this kind of issue any help would be much appreciated.

2

I actually experienced something in reverse of this.

When we first boot our servers, the manager website seems slow. Almost unresponsive. But after some time it’s much more responsive.

I guess I our case some caching warmup or something.

But looking forward to hear what the problem. Is with your setup.

3

Hi all!

Thanks for sharing your experiences on performance issues.

It’s important to note that issues during local web development are often different than ones found in production environments. So if more details can be shared, we can narrow down this issue. Although I have seen this issue before, but couldn’t consistently reproduce it.
.

Network from reloading the page

image
image1373×84 16.5 KB

This screenshot you shared @mito, shows two WebSocket connections:

  • /ws, which is the RSPack web development server ran through your IDE / terminal.
  • /events is the WebSocket connection to the OpenRemote Manager instance. Is this ran through your IDE or through Docker?

The fact that they show similar timings does suggest it’s something outside of OpenRemote.
I also don’t know whether the network tab shows the correct data, as the 40 seconds could also be the lifespan of the websocket connection itself :joy:
But I might be wrong there.
.

In the beginning the website ran really smooth but after a few hours of adding assets and agents etc. The performance declined drastically.

This would be an amazing metric for us if you can share more details.
We’re constantly improving performance using load tests, but we might not cover your scenario.
Is this only an issue locally, or also in a remote production environment?
And do you experience the degrated performance through the UI, or also while using APIs?

.
Any additional information that can be given is worth a lot!
Thanks for sharing in advance :+1:

4

Hi @martin.peeters,

Thanks for replying.

I’m running the Manager through VSCode, and am accessing the site through the Local IP using my laptop which is in the same network as the machine running OR.

So far OR has only a few assets with some agents running and a rule that runs every minute. The current load on the system averages at 5% CPU and 8GB RAM unused.

I’ve tried using the API and couldn’t find any performance issues there.

I also tried running the whole custom deployment stack in Docker which for now doesn’t seem to have the same issues.

If there are any specific metrics that I can provide that would help you please say so and I will try my best to provide them.

5

Hi!

This sounds like the degraded performance is specific to the Browser / Web UI, right?
If the Manager runtime does not show any issues, it would be specific to local development.

You could check if it is specific to NodeJS / the local web development server,
whether the task manager of your operating system shows any weird behavior.
I’ve also seen cases where the Chrome Dev tools really spiked in memory usage,
causing the web browser to hang all the time.

Because, according to your comment, everything is OK once the whole stack runs in Docker.
So it would be interesting to get that web browser / NodeJS metric.