Bulk-Operation Dashboard cloudflare limitations

[ksatter]

Fleet version:

---

💥  Actual behavior

When adding a new software package that is under 200 MB, the request is rejected

Dashboard error:

Cloudflare error - Request Entity too large

These requests do not make it to the Fleet server.

When adding existing software to multiple additional teams, or modifying software that is applied to multiple teams, the request often times out.

If adding one team at a time, this succeeds. I suspect it's due to the async nature of the API ca
Dashboard error:

dogfood-msp-dashboard.fleetdm.com | 524: A timeout occurred

Fleet errors:

{"component":"http","err":"Couldn't upload. Please ensure your internet connection speed is sufficient and stable.","level":"info","path":"/api/v1/fleet/software/package","ts":"2024-12-13T15:21:00.987214255Z","uuid":"75b89e3e-8bb6-4c65-a227-50d08fd036f2"}

{"err":"feature not supported","level":"error","msg":"http middleware failed to override endpoint read timeout","ts":"2024-12-13T15:47:05.264968813Z"}

In the Load balancer, these requests are returning a 408 after 25 seconds, indicating that the client is ending the connection.

After one of these operations fails, the dashboard becomes slow to load. This may not be related, but may also be lingering operations that are still running after the request has been terminated on the front end.

Anectodally, I'm fairly certain that at least one operation that "failed" later appeared to have been completed, but I haven't been able to verify.

🧑‍💻  Steps to reproduce

1. TODO
2. TODO

🕯️ More info (optional)

N/A

[lukeheath]

@sgress454 This is the first bug for next week in the bulk operations dashboard. We're tracking on the Endpoint Ops board so @xpkoala can QA any dashboard changes.

cc @sharon-fdm