Retries after timeout errors immediately run into timeout after 0 ms

[holzerch]

Retries after timeout errors immediately run into timeout after 0 ms

It looks like the SDK is reusing the AbortController from the initial request. But this is already timed out and will no longer work. Possible solution: Create a new AbortController for every request/retry

Here is a screenshot from our logs when the issue happens: The logging is done by wrapper around the fetch function. NETWORK_ERROR means there was no HTTP response status code

here the init of the client

const ctpClient = new ClientBuilder()
            .withProjectKey(config.projectKey)
            .withClientCredentialsFlow({
                host: `https://auth.${config.region}.gcp.commercetools.com`,
                httpClient: fetchWithDefaults,
                projectKey: config.projectKey,
                credentials: {
                    clientId: config.clientId,
                    clientSecret: config.clientSecret,
                },
            })
            .withHttpMiddleware({
                host: `https://api.${config.region}.gcp.commercetools.com`,
                httpClient: fetchWithDefaults,
                enableRetry: true,
                timeout: 5000,
                retryConfig: {
                    retryOnAbort: true,
                    maxRetries: 4,
                    retryDelay: 400,
                    retryCodes: [504, 503, 502, 500],
                },
            })
            .withConcurrentModificationMiddleware()
            .build()

[holzerch]

Fixed with the latest version. Thank you