Sometimes our scraping might fail consistently due to intense global load. It happens sometimes and your best bet is to implement exponential backoff and retry. See https://nubela.co/proxycurl/docs#rate-limit