Subject: Large asset downloads stall/timeout on custom domain (www.lumamap.ru) but work fine on *.netlify.app

Site: stellar-kangaroo-b25f8e.netlify.app
Custom domain: www.lumamap.ru (primary domain, set 2026-09-06)
Affected asset: https://www.lumamap.ru/screenshot-editor.jpg (228,515 bytes)

Problem:
Downloads of this file consistently stall partway through and time out when
requested via the custom domain www.lumamap.ru, while the exact same file on
the same deploy loads instantly via the netlify.app subdomain.

Reproducible test (curl, run 2026-09-07 00:27 UTC+3, two consecutive attempts):

  curl -s -o /dev/null -m 20 -w "TTFB: %{time_starttransfer}s | Total: %{time_total}s | HTTP: %{http_code} | Size: %{size_download}\n" https://www.lumamap.ru/screenshot-editor.jpg

  Attempt 1: TTFB 1.07s | stalls, times out at 20s | HTTP 200 | only 11203 of 228515 bytes received
  Attempt 2: TTFB 0.40s | stalls, times out at 20s | HTTP 200 | only 11203 of 228515 bytes received (identical byte count both times)

  curl -s -o /dev/null -m 20 -w "TTFB: %{time_starttransfer}s | Total: %{time_total}s | HTTP: %{http_code} | Size: %{size_download}\n" https://stellar-kangaroo-b25f8e.netlify.app/screenshot-editor.jpg

  Result: TTFB 0.55s | Total 0.84s | HTTP 200 | full 228515 bytes received

Notes:
- TTFB (headers) is always fast and normal on both hosts — the stall happens
  in the response body, always at the same byte offset (11203 bytes) across
  repeated attempts on the custom domain, which points to something
  deterministic rather than transient network noise.
- This has been reproduced consistently over 24+ hours (first observed
  2026-09-06, confirmed again 2026-09-07), so it is not a transient edge
  incident — nothing is listed on netlifystatus.com for this window.
- Originally the apex domain lumamap.ru was primary, resolving via A record
  to the documented load-balancer fallback IP (75.2.60.5). On 2026-09-06 I
  switched the primary domain to www.lumamap.ru (CNAME to
  stellar-kangaroo-b25f8e.netlify.app) per your own in-app recommendation,
  which cleared the "Pending DNS verification" warning — but the stall on
  large assets persists identically on www.lumamap.ru.
- Checked Web security center for this project: Firewall Traffic Rules,
  Rate Limiting, and WAF are all "Not set" / Disabled, so this isn't an
  account-side security rule blocking/throttling the request.
- Also reproduced from a real browser (not just curl), so it's affecting
  real visitors, not just automated requests.

Could you check whether there's an issue with the edge node(s)/CDN path
serving requests for the custom domain www.lumamap.ru specifically —
possibly something related to how the custom domain is routed vs. the
default netlify.app subdomain, or a proxy/buffering issue that only
triggers on larger files?

Happy to provide HAR files, additional curl -v output, or run more tests
on request.