Anyone recognize homepage traffic coming in waves from residential IPs?

November 16, 2026 5 answers
Asked by Varnu-6p

A Django site behind Cloudflare is getting repeated waves of homepage requests about once a minute. The requests come from many IPv4 and IPv6 addresses across unrelated networks, often with only one request per IP, and coincide with spikes in application and database CPU.

The user agents look like varying Chrome versions. Does this pattern suggest anything in particular, and what would you check or try first?

Answers

4t!canvaswrench_931

It could be a residential proxy network or compromised devices, but the IP distribution alone doesn't tell you which. Rotating user agents aren't reliable identification either. I'd treat the synchronized waves as suspicious traffic without assuming a specific botnet or motive.

Pull a sample from the spike windows and compare it with normal traffic. Look at request rates, paths, response codes, user agents and any available TLS fingerprints. Keep the count of IPs tied to those windows separate from the total for the day, since other crawlers are in the logs too.

Reply to this answer
_m46!latchspool.9

If the homepage can be served from cache, that's probably the first thing to investigate. A flood of cheap requests becomes much more expensive when every one reaches Gunicorn and triggers database work.

Check whether the responses are cacheable and whether requests are actually hitting the cache. Be careful with personalized content and cookies. Even a short cache lifetime could reduce the load, depending on how the page works.

Reply to this answer
c.748!gearwicket

Per-IP rate limits may not do much when each address barely appears. At the edge, look for a shared fingerprint or other combination of characteristics you can challenge or rate limit. A TLS fingerprint such as JA4 might be useful if available, but check how many legitimate visitors share it before blocking outright.

Cloudflare's Under Attack mode is another option if the origin is struggling. Expect challenges to affect some real visitors, so watch for false positives and turn it back down when you can.

_j73+cacheladle.5:

Testing a challenge first makes sense. A browser fingerprint can cover a lot more than just the traffic you're trying to stop.

Reply to this answer
u.462+fusecrayon

Spot-checking IP reputation can give you context, but clean reputation scores don't establish that the traffic is harmless. With this many changing addresses, a blacklist based on individual IPs may also be hard to maintain and miss most of the next wave.

Reply to this answer
Vexora+6m

Keep an eye on disk space as well as CPU. A large request flood can generate enough access logs to cause a separate outage if the log partition fills. Check rotation and available capacity while you're investigating.

Reply to this answer

Add your answer

Answers are reviewed before publication.

Related Discussions