#general
Hi Everyone,
How can I contribute to repository? I have gone through contributor's video and could see that PR to add me as an external contributor is also merged. But I am still not able to open PR.
👋 - We're evaluating Archestra for our enterprise and I've hit a wall with a pdf upload > 10 mb. I see that I can't submit an issue on the GH repo unless I've contributed before but this seems like a simple issue to address and it would really help out our assessment.
Problem: The Next.js middleware has a hardcoded 10MB limit that blocks file uploads, even though the backend supports 50MB by default.
Evidence:
• Uploading >10MB PDFs through the UI fails with "Request body exceeded 10MB" (413 error)
• Direct curl to the backend API successfully accepts 10.7MB files - The limit is purely in the frontend middleware (
middlewareClientMaxBodySize)Fix: Add to
next.config.js: middlewareClientMaxBodySize: process.env.ARCHESTRA_MIDDLEWARE_BODY_LIMIT || '50mb'
}```
Can a contributor please open an issue for this please?Read-only live mirror of Archestra.AI Slack
👋Join the discussion withAI enthusiasts!👋 - We're evaluating Archestra for our enterprise and I've hit a wall with a pdf upload > 10 mb. I see that I can't submit an issue on the GH repo unless I've contributed before but this seems like a simple issue to address and it would really help out our assessment.
Problem: The Next.js middleware has a hardcoded 10MB limit that blocks file uploads, even though the backend supports 50MB by default.
Evidence:
• Uploading >10MB PDFs through the UI fails with "Request body exceeded 10MB" (413 error)
• Direct curl to the backend API successfully accepts 10.7MB files - The limit is purely in the frontend middleware (
middlewareClientMaxBodySize)Fix: Add to
next.config.js: middlewareClientMaxBodySize: process.env.ARCHESTRA_MIDDLEWARE_BODY_LIMIT || '50mb'
}```
Can a contributor please open an issue for this please?7 replies
Hi there, we have https://archestra.ai/contributor-onboard that makes the first contribution for you after you pass captcha
Here is the one for tracking purposes: https://github.com/archestra-ai/archestra/issues/4741
👋 @user Thanks for the detailed report and workaround.
I'm working on the upload limit now. The fix is a bit more involved than just raising the middleware cap because larger files also stress the stored chat message read path.
To size this properly around your real usage:
1. What file sizes do you typically upload?
2. What maximum file size do you need this flow to support?
3. Which LLM provider(s) are you using for these uploads, for example Anthropic, OpenAI, or Bedrock?
That will help me verify the fix against the path you'll actually use.
Hey @user - Typical uploads will be fairly large (75-100mb I'm told from my team). Average is around 10-25mb.
👍1🙌1
Good news @user — the
10MB truncation issue is fixed in v1.2.52, so chat uploads handle much larger files now. The current default is 50MB per attachment, which should comfortably cover the 10-25MB average you mentioned and is a sensible starting point for the first round.For the larger ones at the higher end of your range (75-100MB), a few things come into play beyond just the platform cap:
- The 50MB ceiling is tunable via
ARCHESTRA_API_BODY_LIMITon the backend — env var, no rebuild needed, just a restart. - Each model provider enforces its own per-attachment size limit, which can reject very large files independently of our config.
- Postgres and K8s memory budgets affect how smoothly large attachments flow through the system end-to-end.