A botnet does not need to know which plugins your application runs before probing it. It can request familiar plugin directories, look for version disclosures, and test exposed endpoints across many sites. Distributed sources make individual IP addresses a weak signal: the useful pattern is often the same sequence of paths appearing across unrelated clients.
For a security owner investigating how to block scraping botnets probing for outdated plugins, one configuration mistake deserves an early check: the WAF forwards requests to an old application instance instead of the patched production backend.
This troubleshooting walkthrough describes a configuration scenario, not a reported incident. The concern is concrete: a team patches the intended production instance, but requests still reach a legacy instance with vulnerable code.
The misconfiguration: a stale upstream address
SafeLine receives application traffic as an HTTP/HTTPS reverse proxy and forwards it to an upstream application. Its application configuration includes Domain, listening Port, and Upstream, which the documentation defines as the real address of the original application. See the official application configuration guide.
Suppose a deployment replaces an older backend with a patched instance. The public application continues working, but the SafeLine application's Upstream value still names the older backend. A successful deployment check against the new instance does not establish which instance serves public requests.
An attacker probing plugin paths can therefore encounter code the defender believes has already been removed. Reconnaissance may progress to exploitation if an accessible endpoint has a vulnerability. A directory request alone, however, does not prove exploitation or compromise.
Establish which backend answered
Start with application access logs. Look for repeated plugin-related paths, request methods, timestamps, response codes, and any available request identifiers. Compare the same interval across the old and new backend logs.
A successful response to a plugin readme or asset request can justify checking for version exposure. A missing-page response suggests that particular request did not find the resource, although application routing can obscure the meaning of response codes. Neither result proves that the installed plugin is safe.
Next, open the relevant application configuration in SafeLine and inspect Upstream. Compare that address with the deployment inventory. Confirm the destination's identity and installed plugin versions directly; a familiar hostname is insufficient if its resolution or backend assignment changed.
The decisive evidence is agreement between configuration and traffic: the configured upstream identifies the legacy instance, and that instance's access log records the public requests.
Correct the destination and prove the result
Update the application's Upstream to the verified production destination using the application configuration workflow. SafeLine documents this field as the destination to which it forwards application traffic. Official configuration reference.
Then make a harmless request through the public application with a unique query marker, such as /?routing_check=pluginreview001. Check the backend access logs for that marker. The patched backend should record it; the retired backend should not record that test request. Also verify that the expected page loads and a representative legitimate workflow still succeeds.
This log comparison confirms the routing correction. A screenshot of the corrected Upstream field is useful change evidence, but it does not independently prove where a request arrived.
Verify filtering separately. The official guide provides sample attack requests and explains that a blocking page indicates successful blocking. It also directs administrators to the console's attack list and the "detail" view. Use an authorized, controlled test and inspect the corresponding entry. Official verification procedure.
An attack-list entry confirms handling of that test request. It does not establish that every plugin-discovery request will be blocked.
Decide what to block after routing is correct
Remove unused plugins and verify patches on the instance actually serving traffic. For remaining probes, distinguish discovery from exploit payloads and ordinary application use.
SafeLine documents web attack defenses, IP-based rate limiting, and CAPTCHA challenges. These provide possible controls, but the supplied documentation does not establish a plugin-inventory detector or a botnet-wide correlation mechanism. Official capability overview.
Use request volume and legitimate-client behavior to assess rate limiting. Distributed probing may stay below each source's threshold. Before applying a browser challenge, check whether affected routes serve API clients or integrations that cannot complete it. Keep observing backend logs after changes rather than treating a quieter dashboard as sufficient proof.
Does a plugin probe mean the application was compromised?
No. A probe establishes attempted discovery or access. Investigate suspicious execution, unexpected file changes, and application activity before concluding that compromise occurred.
Why can probes continue after plugins were patched?
Bots can keep requesting known paths without knowing your patch status. Continued requests are different from continued exposure. Verify the serving backend and its installed code.
Which log confirms the upstream fix?
The intended backend's access log should contain the uniquely marked public test request. Correlate the timestamp and marker, and check that the legacy backend did not receive it.
Is a SafeLine blocking page enough to close the issue?
It confirms blocking for the tested request. Closing the routing issue also requires backend evidence, while closing the vulnerability concern requires checking the actual plugin installation and reviewing any signs of exploitation.
SafeLine has surpassed 700,000 installations worldwide as of September 30, 2026. SafeLine Pro annual promotional pricing is $650-$880, compared with the regular $1,000 per year.