Website unreachable in one region

Region failure is never random

A site that loads in Manchester but fails in Milan has a separate cause from a site that fails everywhere. When a failure is regional, the interesting changes are geographic: the DNS answers for that region, the CDN edge that served it, or a geo rule that blocks it. No server crash produces a country-only outage.

What changes by location:

• DNS answers. Anycast and geo DNS return different IPs in different regions. A bad record for one region produces a localized failure while the rest of the world resolves cleanly.
• The edge. A CDN brings a local cache and proxy. If that edge is down, that region gets timeout or error pages, while other edges are unaffected.
• Network policy. Blocks applied by the host, by the ISP, or by sanctions filters drop whole countries.
• Origin routing. If the site bypasses the CDN and connects only to a single origin, the round trip to that origin makes some regions slow or dropped.

Test with the right eyes

Do not reproduce from your own browser; your computer sits in the working region. Use tools that probe from others.

Read the CDN map

Turn the failing IP into a location. The CDN's control panel shows which edge serves a given region, and Cloudflare shows the data center used for the request. If the edge for that region is down or degraded, the site reflects it exactly.

Then probe the DNS the regional clients receive: dig @8.8.8.8 example.com, @1.1.1.1, and the regional resolver. If the answers differ, a cleanup geo record, or a record that was meant to be changed but still points at an old origin, is likely the cause.

Geo rules and firewalls

If the DNS and the edge are both healthy, the failure is policy. Ways a site gets geo-blocked by accident:

• A .htaccess or firewall rule allowing only some countries, written months ago and never removed.
• The host has a regional block, for example the origin is on a provider with a block on a specific region.
• The origin requires the visitor IP to match a whitelist, and a CDN edge forwards everything from within a region.
• A payment or a login plugin enforces a country check that does not fail gracefully.

Order of the check:

1. Ask a person inside the region to visit and report the error. The exact code (522, 403, timeout) halves the work.
2. Test the raw origin from that region, bypassing the CDN. If it loads, the block is in the edge or DNS; if it fails too, the block is at the origin or its firewall.
3. Read the security rules in the CDN: Cloudflare's WAF rules, and the host's .htaccess allowlists.
4. Confirm the domain's health records and the origin's IP are not geo-denied.

Loads but slowly only in that region

A site that "is down" in a region is often only slow there. The difference: where there is no local CDN edge, HTML and assets travel the whole route, and a page that is heavy on transfers or round trips fails and times out at the far end even when the same page renders instantly at home. The fixes are immediate and lasting:

• Use a CDN with an edge in the failing region, or test with a paid anycast DNS.
• Reduce the weight: compress (gzip, brotli), recompress every large image, serve next-gen images, and remove render-blocking assets so the far round trip matters less.
• For a slower origin, add more cache and keepalive and static edge caching so visitors in far regions fetch prebuilt files.

Prevention

• Keep the DNS records mirrored and short (TTL 300) so a change reaches a whole region quickly and a bad record dies fast.
• Add a monitoring probe located in every region you sell to, not just yours.
• Write the geo rules at the CDN and record each one, so a block written in a panic is removed on purpose.

When a single country fails

If one specific country is affected and it is not a regional ISP issue, check the domain record and the shared host's country block: both are common causes on smaller plans, and both are fixed at the DNS, rule, or hosting layer, not in the application code.