（可选）您可以将此重试策略用于返回错误代码 404 的 IAM API 请求。IAM 读取最终一致；因此，资源在创建后可能不会立即显示，而这可能会导致 404 错误。
此外，针对返回错误代码 409 和状态 ABORTED 的所有 IAM API 请求，请使用此重试策略的修改版本。此类错误表示出现并发问题；例如，您可能尝试更新另一个客户端已覆盖的允许政策。对于此类错误，您应该始终使用引入了抖动的截断指数退避算法来重试请求的整个读取-修改-写入系列。如果仅重试写入操作，则请求将继续失败。
[[["易于理解","easyToUnderstand","thumb-up"],["解决了我的问题","solvedMyProblem","thumb-up"],["其他","otherUp","thumb-up"]],[["很难理解","hardToUnderstand","thumb-down"],["信息或示例代码不正确","incorrectInformationOrSampleCode","thumb-down"],["没有我需要的信息/示例","missingTheInformationSamplesINeed","thumb-down"],["翻译问题","translationIssue","thumb-down"],["其他","otherDown","thumb-down"]],["最后更新时间 (UTC)：2025-08-21。"],[[["\u003cp\u003eThis document outlines best practices for retrying failed requests to the Identity and Access Management (IAM) API, recommending the use of truncated exponential backoff with jitter for safe-to-retry requests.\u003c/p\u003e\n"],["\u003cp\u003eUsing truncated exponential backoff with jitter helps prevent overwhelming the IAM API with a large number of retries, which can lead to exceeding project quotas and limits.\u003c/p\u003e\n"],["\u003cp\u003eThe retry algorithm involves exponentially increasing wait times between retries, with the addition of a random delay (jitter) to avoid synchronized retries from multiple clients, and a cap on the wait time, followed by a final deadline.\u003c/p\u003e\n"],["\u003cp\u003eThe recommended retry strategy should be used for IAM API requests that return \u003ccode\u003e500\u003c/code\u003e, \u003ccode\u003e502\u003c/code\u003e, \u003ccode\u003e503\u003c/code\u003e, or \u003ccode\u003e504\u003c/code\u003e error codes, and optionally \u003ccode\u003e404\u003c/code\u003e errors due to eventual consistency, and a modified version should be used for \u003ccode\u003e409\u003c/code\u003e \u003ccode\u003eABORTED\u003c/code\u003e errors.\u003c/p\u003e\n"],["\u003cp\u003eFor \u003ccode\u003e409\u003c/code\u003e errors, which typically signify concurrency issues, you should retry the entire read-modify-write series of requests, as retrying only the write operation will not resolve the issue.\u003c/p\u003e\n"]]],[],null,["# Retry failed requests\n\nThis page describes best practices for retrying failed requests to the\nIdentity and Access Management (IAM) API.\n\nFor requests that are safe to retry, we recommend using truncated exponential backoff with introduced jitter.\n\nOverview of truncated exponential backoff\n-----------------------------------------\n\n\nEach request to the IAM API can succeed or fail. If your application retries\nfailed requests without waiting, it might send a large number of retries to IAM\nin a short period of time. As a result, you might exceed [quotas and limits](/iam/quotas) that apply to every\nIAM resource in your Google Cloud project.\n\n\nTo avoid triggering this issue, we strongly recommend that you use\n[truncated exponential backoff](https://en.wikipedia.org/wiki/Exponential_backoff) with\nintroduced [jitter](https://en.wikipedia.org/wiki/Jitter), which is a standard\nerror-handling strategy for network applications. In this approach, a client periodically retries\na failed request with exponentially increasing delays between retries. A small, random delay,\nknown as jitter, is also added between retries. This random delay helps prevent a synchronized\nwave of retries from multiple clients, also known as the\n[thundering herd problem](https://en.wikipedia.org/wiki/Thundering_herd_problem).\n\nExponential backoff algorithm\n-----------------------------\n\n\nThe following algorithm implements truncated exponential backoff with jitter:\n\n1. Send a request to IAM.\n2. If the request fails, wait 1 + `random-fraction` seconds, then retry the request.\n3. If the request fails, wait 2 + `random-fraction` seconds, then retry the request.\n4. If the request fails, wait 4 + `random-fraction` seconds, then retry the request.\n5. Continue this pattern, waiting 2^n^ + `random-fraction` seconds after each retry, up to a `maximum-backoff` time.\n6. After `deadline` seconds, stop retrying the request.\n\n\nUse the following values as you implement the algorithm:\n\n- Before each retry, the wait time is `min((2`^n^` + random-fraction), maximum-backoff)`, with `n` starting at 0 and incremented by 1 for each retry.\n- Replace `random-fraction` with a random fractional value less than or equal to 1. Use a different value for each retry. Adding this random value prevents clients from becoming synchronized and sending large numbers of retries at the same time.\n- Replace `maximum-backoff` with the maximum amount of time, in seconds, to wait between retries. Typical values are 32 or 64 (2^5^ or 2^6^) seconds. Choose the value that works best for your use case.\n- Replace `deadline` with the maximum number of seconds to keep sending retries. Choose a value that reflects your use case. For example, in a continuous integration/continuous deployment (CI/CD) pipeline that is not highly time-sensitive, you might set `deadline` to 300 seconds (5 minutes).\n\nTypes of errors to retry\n------------------------\n\nUse this retry strategy for all requests to the IAM API that\nreturn the error codes `500`, `502`, `503`, or `504`.\n\nOptionally, you can use this retry strategy for requests to the\nIAM API that return the error code `404`.\n[IAM reads are eventually consistent](/iam/docs/overview#consistency); as a\nresult, resources might not be visible immediately after you create them, which\ncan lead to `404` errors.\n\nIn addition, use a modified version of this retry strategy for all requests to\nthe IAM API that return the error code `409` and the status\n`ABORTED`. This type of error indicates a concurrency issue; for example, you\nmight be trying to update an [allow policy](/iam/docs/allow-policies) that another client has\nalready overwritten. For this type of error, you should always retry the entire\n[read-modify-write](/iam/docs/allow-policies#etag) series of requests, using\ntruncated exponential backoff with introduced jitter. If you retry only the write operation, the request will\ncontinue to fail.\n\nWhat's next\n-----------\n\n- Learn [how concurrency issues are managed](/iam/docs/allow-policies#etag) in allow policies.\n- Understand how to [implement the read-modify-write pattern](/iam/docs/granting-changing-revoking-access#programmatic) for updating allow policies."]]
