weaviate/weaviate · warning

init graphql provider

Error message

init graphql provider

What it means

text2vec-nvidia wraps any error from initNearText() with the message "init graphql provider" during InitExtension. initNearText() wires the nearText GraphQL searcher/argument provider (modules/text2vec-nvidia/nearText.go:19) and currently always returns nil, so this wrap is purely defensive: it would only surface if the wiring fails or if m.vectorizer is nil (InitExtension called before Init), which would cause a nil dereference rather than this wrapped error. If you see it, it means module startup sequencing broke.

Source

Thrown at modules/text2vec-nvidia/module.go:96

	}

	return nil
}

func (m *NvidiaModule) InitExtension(modules []modulecapabilities.Module) error {
	for _, module := range modules {
		if module.Name() == m.Name() {
			continue
		}
		if arg, ok := module.(modulecapabilities.TextTransformers); ok {
			if arg != nil && arg.TextTransformers() != nil {
				m.nearTextTransformer = arg.TextTransformers()["nearText"]
			}
		}
	}

	if err := m.initNearText(); err != nil {
		return errors.Wrap(err, "init graphql provider")
	}
	return nil
}

func (m *NvidiaModule) initVectorizer(ctx context.Context, timeout time.Duration,
	logger logrus.FieldLogger,
) error {
	apiKey := os.Getenv("NVIDIA_APIKEY")
	client := clients.New(apiKey, timeout, logger)

	m.vectorizer = text2vecbase.New(client,
		batch.NewBatchVectorizer(client, 50*time.Second, batchSettings, logger, m.Name()),
		batch.ReturnBatchTokenizer(batchSettings.TokenMultiplier, m.Name(), ent.LowerCaseInput),
	)
	m.metaProvider = client

	return nil
}

View on GitHub (pinned to 75aa4b6d11)

Solutions

1. Ensure all modules' Init() is called before InitExtension() (standard startup does this; fix custom wiring if you call these yourself).
2. Check the wrapped inner error in the log line — it names the actual failure inside initNearText.
3. If you modified nearText.go or module.go, verify m.vectorizer is initialized before initNearText() runs.
4. Update/rebuild Weaviate if this reproduces on an unmodified binary; on stock code this path cannot fail.

Example fix

// before (custom embedding code)
module.InitExtension(allModules) // Init not called yet -> nil vectorizer
// after
module.Init(ctx, initParams)
module.InitExtension(allModules)

// operator-side preflight: ensure module init order and env
if os.Getenv("NVIDIA_APIKEY") == "" {
  log.Println("warning: NVIDIA_APIKEY not set; text2vec-nvidia will fail at vectorization time")
}
// in custom wiring, enforce lifecycle order:
if err := mod.Init(ctx, params); err != nil { return err }
if err := mod.InitExtension(allModules); err != nil { return fmt.Errorf("init graphql provider: %w", err) }

// Go: unwrap and classify at startup
if err := mod.InitExtension(all); err != nil {
  return fmt.Errorf("nvidia module startup failed: %w", err)
}

When it happens

Trigger: Server startup calls InitExtension on the text2vec-nvidia module and initNearText() returns a non-nil error; or InitExtension is invoked before Init so nearText.NewSearcher receives a nil vectorizer, producing an error wrapped here.

Common situations: Custom module registration order in configure_api.go, embedded/custom Weaviate builds that call module lifecycle methods manually, or code changes to initNearText that introduce error returns.

Related errors

• init graphql provider
• read initial version from file
• init %s
• init graphql provider
• init graphql provider

AI-assisted analysis of weaviate/weaviate@75aa4b6d11 (2026-09-04). Data as JSON: /api/errors/d6531fdfa007ba89. Report an issue: GitHub.