Timeout on /forest/sessions-google requests

We are getting a “request timeout” on a Forest internal request:

High Response Time: Heroku/Syslog drain

2021-01-12 14:34:14.267
370 <158>1 2021-01-12T14:34:13.687693+00:00 heroku router - - at=error code=H12 desc=“Request timeout” method=POST path="/forest/sessions-google" dyno=web.1 connect=0ms service=30001ms status=503 bytes=0 protocol=https

What could be the cause for this?

Thanks!

Hi @Eyal_Hakim !
Are you still getting this issue ?
Can you share a screen capture of the error with loom for example ?

Hi, yes we are still getting it. There is nothing really to record, we are seeing those issues in the console and Forest becomes very slow.

I think this happens when we upload large images to Forest, it consumes the server’s memory and hangs the node process. Can we upload images directly to our storage from the client-side without the need to first upload them to the forest backend and from there to our storage?

We are using Cloudinary.

Thanks

So your issue is not on connection to Forest ?
If so, are you using a smart action to upload those files ?

No the issue is not about connection to Forest.
We are using the built-in file picker widget, which then sends the file as Base64 to the forest-backend and from there we upload it to Cloudinary.

Is there a way to skip the backend and upload directly to Cloudinary?

BTW the images are just a guess, we are getting slow response times and timeouts from many different routes that do not have images.

Hello @Eyal_Hakim :wave: Welcome to our community :confetti_ball: :tada:

Sadly, uploading an image form the UI to a remote bucket is not possible. You have to hit the agent first, which will handle authentication data, and then upload to your bucket.

If I may, I would to ask some questions to have a better idea of what is causing the latencies:
1- Do you observe the high latency only on the route /forest/sessions-google or some other routes are impacted as well ? Do you have any other example ?
2- Have you tried to restart your server ?
3- Are you able to reproduce the latency by performing a specific series of actions ?
4- Are you able to monitor the memory used by your agent over time ?

Thanks in advance for your answers, they will help to investigate :raised_hands:

Steve.

  1. Many of our other routes have the exact same issue. They are private so I wouldn’t share them here.
  2. Yes
  3. Not sure, maybe uploading an image as this is my guess, can check
  4. Yes

Will it be possible to upload the image directly if we build a custom component?

Hello @Eyal_Hakim :wave:

Sadly this is not possible. You can create smart views where you can build whatever you want, but this is for collection list pages only. And I would not recommend to put you cloudinary credentials there.

If you get a 503 http error code, this usually mean that the server is online, but not able to respond. So my first guess would be a memory leak, which slows down your application.

Are you able to catch anything on your memory monitor ?

Steve.