When the OCSP subsystem attempts to reach out to a remote system, it might find that the MongoDB Server which it is running within is shutting down. It should never be possible for this error to be confused with a network fault and result in "soft failure". Instead, OCSPManager::requestStatus should explicitly propagate ErrorCodes::ShutdownInProgress to its caller. dispatchOCSPRequests should consider this error to be a "strong" error, and propagate it up. The OCSPCache should permit the error to propagate up to SSLManagerOpenSSL::ocspClientVerification which should consider the result to indicate that it should reject the certificate it is validating.