Error 1202 concerns the Snippet subrequest allowance. Count the requests made during one failing invocation, including redirects, retries and fallback branches.
The number of visitors and the number of fetch calls written in code do not describe that complete sequence. Use the actual failure path and the current Snippet allowance.
Guide reviewed: October 8, 2026. Interface labels may vary by version, account and device.
Count the complete subrequest path
| Part of the path | Record | Why it matters |
|---|---|---|
| Initial dependency call | Endpoint and response | Start of the actual sequence |
| Redirect chain | Each hop | One code call may produce several requests |
| Retry | Maximum attempts and trigger | Failure cases expand work |
| Fallback | Alternate endpoint and condition | Success path may hide the worst case |
| Rule scope | Requests selected | Unrelated pages may invoke extra work |
Follow one request from start to finish
- Identify the matching Snippet and failing input.
- List dependency calls in order.
- Add redirect hops to each call.
- Include the maximum supported retries and fallback branch.
- Compare this sequence with the current Snippet allowance.
Record the affected URL and identify the Snippets that apply to it. Then review what outside resources or destinations the feature contacts. The developer needs the full path through the code, including conditional calls that do not run in the simplest test.
For example, a feature might fetch information from a service, then fetch an alternative when the first response is unsuitable. A test in which the first service succeeds may use fewer requests than a real visit that follows the fallback branch.
This example illustrates why the failing input matters. It does not establish that a fallback is responsible for every 1202 result. The actual request sequence should support the explanation.
Include redirects in the count
Cloudflare’s Snippet documentation states that redirects within a subrequest chain count toward the allowance. A call to one address may therefore involve more than one request before it reaches its final destination.
A common pattern worth examining is a destination that redirects to another hostname and then to its HTTPS or canonical address. If the intended final endpoint is documented and stable, using it directly may avoid unnecessary intermediate steps.
That change should be verified with the service provider’s requirements. A redirect can carry intended routing or authentication behavior, so bypassing it by guesswork is not always appropriate.
Check fallback and retry logic
Keep a path table for normal success, first dependency failure and full fallback. Concurrency changes waiting time, not the number of calls. Count the complete sequence before deciding which request is unnecessary.
Retries can expand the request count when a dependency is unavailable. A developer should review how many attempts are possible and what causes another attempt. A limit with a clear purpose is easier to reason about than a loop that continues until a response looks acceptable.
A fallback should also have a defined endpoint and result. If several branches can each contact another service, document the worst case rather than estimating from a successful first response.
Making calls concurrently may reduce elapsed waiting time, but it does not eliminate the calls from the count. The correction needs to reduce unnecessary requests or use a design whose allowance fits the required work.
Keep the matching scope aligned with the feature
A Snippet that needs extra data only for a particular set of pages should not automatically run on unrelated resources. Compare the matching rule with the intended feature and test requests outside that scope.
Likewise, repeated requests for information already available in the current request may be unnecessary. A developer can review which data is genuinely missing before contacting another service. Any reuse or caching design needs to account for freshness and user-specific information.
Do not substitute one visitor’s private result for another merely to reduce calls. The data’s meaning and audience still need to be preserved when the request sequence is simplified.
Compare the design with the current plan allowance
The applicable Snippet allowance should be checked in the current product documentation and account. Workers and Snippets are different products, so a number taken from a Workers tutorial should not be assumed to describe a Snippet.
If the task legitimately needs more requests than the available allowance, the responsible developer can assess another implementation. That decision should include the feature’s requirements and costs, rather than treating a hosting upgrade as a way to expand an unrelated edge limit.
After the correction, repeat both the original failing input and the fallback case. Count the resulting sequence and verify the feature’s output. A successful normal page does not test the branch that previously exhausted the allowance.
Cloudflare defines 1202 and explains redirect counting in its Snippet troubleshooting guide. Keep that reference alongside a request sequence from the actual feature when choosing the correction.
Verify the worst supported path
Retest the original failure and fallback case, then compare the request count and expected output. Keep user-specific data and freshness requirements intact if introducing reuse. A normal first-response success alone does not test the branch that exhausted the allowance.