request-retry

Example

import { rpRetry } from '@mimik/request-retry';
// or
import rp from '@mimik/request-retry';

request-retry~rpRetry(options) ⇒ Promise.<*>

Make a request with retries.

Kind: inner method of request-retry
Returns: Promise.<*> - A bluebird promise (supports cancellation).
Category: async
Throws:

• Error An error produced by getRichError, wrapping an Axios error or a TimeoutError.

The default retryStrategy is:

defaultRetry = (...args) => {
  const { statusCode } = args[1]; // err
  // Retry on 5xx and 429. If there is no statusCode, retry unless it's an "Invalid URI" error.
  return (statusCode && (Math.floor(statusCode / 100) === 5 || statusCode === 429))
         || (!statusCode && !args[1].message.includes('Invalid URI'));
};

If not already set, the User-Agent header is set to: mimik-{serverType}/{serverVersion}/{serverId} {architecture} {node}.

To ease migration from request-promise, the following adjustments are applied to options:

• uri takes precedence over url.
• If json is an object, body takes precedence over json; both are mapped to Axios data.
• qs is mapped to Axios params.

Requires: module:@mimik/sumologic-winston-logger, module:@mimik/response-helper
Fulfil: object - The Axios response when resolveWithFullResponse is true; otherwise response.data.

| Param | Type | Default | Description | | --- | --- | --- | --- | | options | object | | Options for the request (similar to axios). The validateStatus option is disabled: an error is thrown for status codes outside [200, 300). An additional option, resolveWithFullResponse, controls the return shape: when true, the full Axios response object is returned; when false or omitted, only response.data is returned. | | [options.retry] | object | | Retry configuration. | | [options.retry.retries] | number | 2 | Maximum number of retries, independent of the retryStrategy. If the value is < 0, it is set to 0; if > 15, it is set to 15. | | [options.retry.delay] | number | 1000 | Delay between retries in milliseconds when no delayStrategy is provided. If the value is < 20, it is set to 20; if > 30000, it is set to 30000. | | [options.retry.delayStrategy] | function | | A function that returns the delay (in milliseconds) to wait before the next retry. Signature: (nbRetry, err, options, correlationId). Must return a number > 0 and < 30000; otherwise delay is used. | | [options.retry.retryStrategy] | function | | A function that decides whether to retry. Signature: (nbRetry, err, options, correlationId). Must return a boolean; otherwise it is ignored. | | [options.retry.timeout] | number | 50 | Request timeout (in seconds) applied to the initial request and all retries. If reached, a TimeoutError is thrown. Values < 10 are set to 10; values > 120 are set to 120. | | [options.retry.logLevel] | object | | Controls logging behavior. When omitted, requests and responses are logged at info with full details, and errors at warn. When provided but incomplete, only the provided keys are active; missing keys disable that logging. Non-object values are ignored and defaults apply. | | [options.retry.logLevel.response] | string | | Log level for responses. Invalid values fall back to silly. | | [options.retry.logLevel.error] | string | | Log level for errors. Invalid values fall back to silly. | | [options.retry.logLevel.request] | string | | Log level for requests. Invalid values fall back to silly. | | [options.retry.logLevel.responseDetails] | string | "'type'" | Detail level for the response: count, type, or full. Invalid values fall back to type. When logLevel is omitted entirely, the default is full. | | [options.retry.logLevel.responseName] | string | "'response'" | Label associated with the response. Non-string or empty values fall back to response. | | [options.metrics] | object | | Metrics configuration. | | [options.metrics.HTTPRequestDuration] | function | | A prom-client metric function to measure request duration. | | [options.metrics.url] | string | | Optional URL label for the metric. If omitted, the request URL is used. |