openfaas/faas · error

log request failed

Error message

log request failed

What it means

The log streaming proxy in gateway/handlers/logs.go forwards the client's log request to the function provider's /system/logs endpoint using http.DefaultTransport.RoundTrip. If that round trip fails at the transport level — connection refused, DNS failure, TLS error, timeout — the gateway logs 'LogProxy: forwarding request failed' and returns 500 with 'log request failed'. The provider address comes from the gateway's URL resolver, typically the functions_provider_url/upstream configuration.

Source

Thrown at gateway/handlers/logs.go:65

		wf, ok := w.(writerFlusher)
		if !ok {
			log.Println("LogHandler: response is not a Flusher, required for streaming response")
			http.NotFound(w, r)
			return
		}

		if writeRequestURI {
			log.Printf("LogProxy: proxying request to %s %s\n", logRequest.Host, logRequest.URL.String())
		}

		ctx, cancel := context.WithCancel(ctx)
		logRequest = logRequest.WithContext(ctx)
		defer cancel()

		logResp, err := http.DefaultTransport.RoundTrip(logRequest)
		if err != nil {
			log.Printf("LogProxy: forwarding request failed: %s\n", err.Error())
			http.Error(w, "log request failed", http.StatusInternalServerError)
			return
		}
		defer logResp.Body.Close()

		switch logResp.StatusCode {
		case http.StatusNotFound, http.StatusNotImplemented:
			w.WriteHeader(http.StatusNotImplemented)
			return
		case http.StatusOK:
			// watch for connection closures and stream data
			// connections and contexts should have cancel methods deferred already
			select {
			case err := <-copyNotify(&unbufferedWriter{wf}, logResp.Body):
				if err != nil {
					log.Printf("LogProxy: error while copy: %s", err.Error())
					return
				}
			case <-cn.CloseNotify():

View on GitHub (pinned to 8d803bf9e2)

Solutions

1. From inside the gateway container, verify the provider answers: curl the provider /healthz or /system/functions endpoint
2. Correct functions_provider_url so it points directly at the provider service
3. Check the provider pod status and logs (kubectl get pods -n openfaas, kubectl logs on the provider deployment)
4. Review NetworkPolicies or service mesh rules between gateway and provider namespaces
5. If DNS fails, confirm the provider service exists and the namespace suffix is correct

# preflight: provider reachable before streaming logs
curl -fsS "$FUNCTIONS_PROVIDER_URL/healthz" || echo 'provider unreachable: log streaming will fail'

resp, err := client.Get(logsURL)
if err != nil || resp.StatusCode == http.StatusInternalServerError {
    // transport failure to the provider: back off and retry
    time.Sleep(backoff)
    continue
}

When it happens

Trigger: Calling GET /system/logs?name=myfn&follow=true while the provider (faas-netes, faasd, or another faas-provider implementation) cannot be reached: wrong upstream host or port, provider pod not running or not ready, NetworkPolicy blocking gateway-to-provider traffic, or a TLS certificate mismatch on the upstream.

Common situations: functions_provider_url pointing at a stale ClusterIP or wrong port after a reinstall; provider crash-looping; CNI/NetworkPolicy changes blocking namespace-local traffic; faasd address or socket changed.

Related errors

• unknown log request error (%v)
• string(upstreamBody)
• A body is required for this endpoint
• err.Error()
• Error queuing request: %s

AI-assisted analysis of openfaas/faas@8d803bf9e2 (2026-08-16). Data as JSON: /api/errors/a7819099ce2bba98. Report an issue: GitHub.