What causes "stream API request failed: %w" in micro/go-micro?

Verify BaseURL is reachable: curl -v /v1/chat/completions from the same host.

stream API request failed: %w

Error message

stream API request failed: %w

What it means

streamOpenAI (used for Ollama Cloud / OpenAI-compatible endpoints) failed to execute the streaming HTTP POST to /v1/chat/completions. The error wraps the transport-level error from http.DefaultClient.Do, so the message contains the underlying net/http cause (DNS, TCP, TLS, timeout, context cancellation). No HTTP response was received at all.

Source

Thrown at ai/ollama/ollama.go:349

	reqBody, err := json.Marshal(apiReq)
	if err != nil {
		return nil, fmt.Errorf("failed to marshal stream request: %w", err)
	}

	apiURL := strings.TrimRight(p.opts.BaseURL, "/") + p.streamPath()
	httpReq, err := http.NewRequestWithContext(ctx, http.MethodPost, apiURL, bytes.NewReader(reqBody))
	if err != nil {
		return nil, fmt.Errorf("failed to create stream request: %w", err)
	}
	httpReq.Header.Set("Content-Type", "application/json")
	httpReq.Header.Set("Accept", "text/event-stream")
	if p.opts.APIKey != "" {
		httpReq.Header.Set("Authorization", "Bearer "+p.opts.APIKey)
	}

	httpResp, err := http.DefaultClient.Do(httpReq)
	if err != nil {
		return nil, fmt.Errorf("stream API request failed: %w", err)
	}
	if httpResp.StatusCode != http.StatusOK {
		defer httpResp.Body.Close()
		respBody, _ := io.ReadAll(httpResp.Body)
		return nil, fmt.Errorf("stream API error (%s): %s", httpResp.Status, string(respBody))
	}

	return &sseStream{body: httpResp.Body, scanner: bufio.NewScanner(httpResp.Body)}, nil
}

// buildOpenAIMessages converts an ai.Request into the OpenAI chat message format.
func buildOpenAIMessages(req *ai.Request) []map[string]any {
	messages := []map[string]any{}
	if req.SystemPrompt != "" {
		messages = append(messages, map[string]any{"role": "system", "content": req.SystemPrompt})
	}
	for _, m := range req.Messages {
		messages = append(messages, map[string]any{"role": m.Role, "content": m.Content})

View on GitHub (pinned to 24529f1404)

Solutions

Verify BaseURL is reachable: curl -v <BaseURL>/v1/chat/completions from the same host.
Check that ctx is not already cancelled and no overly tight timeout is applied.
Confirm DNS/proxy settings (HTTP_PROXY/HTTPS_PROXY) and that ollama.com is reachable.
Inspect the wrapped %w error for the concrete cause (connection refused vs timeout vs TLS).

Example fix

// before
p := ai.NewProvider("ollama", ai.WithBaseURL("https://ollama.wrong-host.example/v1"))
// after
p := ai.NewProvider("ollama", ai.WithBaseURL("https://ollama.com/v1"), ai.WithAPIKey(os.Getenv("OLLAMA_API_KEY")))

Defensive patterns

Strategy: retry

Validate before calling

u, err := url.Parse("https://ollama.com/v1")
if err != nil || u.Host == "" {
    return fmt.Errorf("invalid base URL: %w", err)
}
conn, err := net.DialTimeout("tcp", u.Hostname()+":443", 3*time.Second)
if err != nil {
    return fmt.Errorf("ollama cloud unreachable: %w", err)
}
conn.Close()
if ctx.Err() != nil {
    return ctx.Err()
}

Try / catch

stream, err := provider.Stream(ctx, req)
if err != nil {
    var netErr net.Error
    if errors.As(err, &netErr) && (errors.Is(err, context.DeadlineExceeded) || isTransient(err)) {
        // retry with backoff
    }
    return fmt.Errorf("stream setup failed: %w", err)
}

Prevention

Health-check the endpoint at startup with a short plain (non-stream) request.
Always pass a context with a sane but generous deadline for long streams.
Log the wrapped error chain with %v/%w so the root cause is visible.
Avoid aggressive proxies/VPNs between the app and ollama.com in production.

When it happens

Trigger: Calling Stream() when isCloud() is true (BaseURL contains ollama.com or cloudOverride set) and the HTTP request fails at the transport layer: server unreachable, connection refused, DNS failure, TLS error, or ctx cancelled before response headers.

Common situations: Wrong BaseURL or port, Ollama Cloud API key set but network/VPN blocking ollama.com, local proxy down, request context cancelled/expired mid-request, DNS misconfiguration in containers.

Understand the failure class

Background: 'Something went wrong' / 'Request failed (500)' / 'HTTP error! status: 404' — what failed HTTP requests actually mean and how to find the real cause — this error's family across 28 libraries.

Related errors

AI-assisted analysis of micro/go-micro@24529f1404 (2026-09-01). Data as JSON: /api/errors/7690997c94374527. Report an issue: GitHub.