Mintplex-Labs/anything-llm · error

Internal Server Error

Error message

Internal Server Error

What it means

AnythingLLM's OpenAI-compatible GET /v1/openai/models handler returns this 500 from its catch-all after `response.sendStatus(500)` when building the model list from `Workspace.where()` (a Prisma query over the workspaces table) throws. The handler maps each workspace row to `{id: workspace.slug, created: Math.floor(Number(new Date(workspace.createdAt))/1000), owned_by}`, so any database read failure or an unexpected row shape aborts the whole list. The real exception is printed server-side by `console.error(e.message, e)` — the HTTP body carries no detail.

Solutions

1. Check the server console — the actual exception is logged right before the 500 is sent; identify whether it is a Prisma connection error, a lock, or a schema error
2. Probe database health with GET /v1/system (system/index.js:70): if that also returns 500, the fault is the database, not this endpoint
3. Verify STORAGE_DIR exists, is writable, and persists across container restarts so the SQLite DB is not recreated mid-request
4. If schema drift is indicated, re-run the server so boot-time migrations apply, or restore the DB from a backup taken on the same AnythingLLM version

Example fix

// before
const res = await fetch('https://host/api/v1/openai/models', {headers:{Authorization:'Bearer key'}});
const models = await res.json(); // throws on 500 'Internal Server Error' body

// after
const res = await fetch('https://host/api/v1/openai/models', {headers:{Authorization:'Bearer key'}});
if (res.status === 500) throw new Error('models list failed - check AnythingLLM server logs (DB likely unreachable)');
if (!res.ok) throw new Error(`models list failed: HTTP ${res.status}`);
const models = await res.json();

const health = await fetch(`${base}/api/v1/system`, opts);
if (health.status !== 200) throw new Error('AnythingLLM DB layer unhealthy - /v1/openai/models will 500');

null

try { const r = await fetch(modelsUrl, opts); if (r.status===500) throw new Error('DB read failed server-side'); return await r.json(); } catch (e) { logger.error('list models failed', e.message); return []; }

When it happens

Trigger: Calling GET /v1/openai/models with a valid API key while the database is unreachable, locked, or schema-drifted (e.g. SQLite file on a stale volume, Postgres container down). Also triggered by a workspace row with a null slug/createdAt if mapping code path throws during serialization.

Common situations: STORAGE_DIR on a read-only or full Docker volume; container killed mid-write corrupting anythingllm.db; upgrading/downgrading AnythingLLM without running Prisma migrations so the workspaces table no longer matches the entity model.

Understand the failure class

Background: Database query failed: Internal Server Error 500s wrapping SQL, Prisma, and connection failures — what to check first — this error's family across 16 libraries.

• HTTP status errors: handling 4xx and 5xx responses — how to handle 4xx and 5xx responses properly.

Related errors

• Failed to fetch workspaces
• Internal Server Error
• Internal Server Error
• Internal Server Error
• Could not find a document by id

AI-assisted analysis of Mintplex-Labs/anything-llm@f92433b4ea (2026-08-18). Data as JSON: /api/errors/b6368d21a082879f. Report an issue: GitHub.

Appendix: source

Thrown at server/endpoints/api/openai/index.js:72

    */
    try {
      const data = [];
      const workspaces = await Workspace.where();
      for (const workspace of workspaces) {
        data.push({
          id: workspace.slug,
          object: "model",
          created: Math.floor(Number(new Date(workspace.createdAt)) / 1000),
          owned_by: workspace?.chatProvider || process.env.LLM_PROVIDER,
        });
      }
      return response.status(200).json({
        object: "list",
        data,
      });
    } catch (e) {
      console.error(e.message, e);
      response.sendStatus(500).end();
    }
  });

  app.post(
    "/v1/openai/chat/completions",
    [validApiKey],
    async (request, response) => {
      /*
      #swagger.tags = ['OpenAI Compatible Endpoints']
      #swagger.description = 'Execute a chat with a workspace with OpenAI compatibility. Supports streaming as well. Model must be a workspace slug from /models.'
      #swagger.requestBody = {
          description: 'Send a prompt to the workspace with full use of documents as if sending a chat in AnythingLLM. Only supports some values of OpenAI API. See example below.',
          required: true,
          content: {
            "application/json": {
              example: {
                messages: [
                {"role":"system", content: "You are a helpful assistant"},

View on GitHub (pinned to f92433b4ea)