Check a rotation sample
Use the bounded local sampler to collect observations. It opens a separate curl process for each request and writes a CSV. Run it on your machine with your own proxy account, then import the file here.
Read the change count correctly
A usable observation has HTTP status 200, a valid IP address and an empty error field. The tool compares two adjacent observations only when both are usable and their timestamps differ. It does not connect observations across a failed sample.
If A, A, B are recorded successfully at successive times, there is one observed change in two comparable pairs. If A, failure, B are recorded, there is no comparable pair across that failure. The addresses differ, but the missing observation is preserved.
Rows are sorted by timestamp. IPv6 spellings are normalized before comparison. The observation window runs from the first timestamp to the last; it does not establish continuous availability between them.
What an IP sample cannot establish
Two endpoints can share a public address, and one endpoint can use different addresses over time. Repeated IPs therefore cannot prove device reuse. Distinct observed IPs are the addresses in your sample, not the provider’s total inventory.
Keep new connections separate from requests over an existing connection. A browser can send many requests through one CONNECT tunnel. The rotation test guide explains how that affects the experiment.
A sample of successful echo responses also does not prove the target website worked. Record the task’s own pass condition and test duration alongside the IP observations.
Match the test to your routing choice
Use the updated username builder to keep the pool, country, session ID and mode deliberate. Timed modes need observations before and after the selected interval. Sticky and hard select the endpoint-reuse behavior; neither promises an unchanged public IP.
For the new-connection preset, omit the explicit session ID and open a fresh connection for each sample. Verify the configuration and current limits in your dashboard. Compare buying options on the rotating mobile proxies page, then check current country stock.
Common questions
Does this test a proxy server directly?
No. Use the downloadable sampler locally to produce observations. This page analyzes the CSV and never receives the proxy credentials.
Why are failed samples excluded from adjacent change counts?
There is no valid address to compare at that timestamp. They remain visible as failures and interrupt the comparison sequence.
Can this certify sticky routing or a pool size?
No. It reports the sample you provide. Public IP addresses alone do not identify the underlying endpoint or unobserved inventory.
Sources and review scope
Counting rules describe this tool’s implementation. The example uses documentation IP addresses and fictional observations. Gateway behavior was checked against the local source on 13 October 2026; no live paid rotation test is claimed.
- Local rotation method and sampler
- curl connection options
- Shared address space and carrier-grade NAT
- Pool Gateway documentation
Agent service index and Pool Gateway documentation provide the API entry points. The tool index as JSON lists these tools and their limits.
