Skip to content

FAQ and troubleshooting

General

Is NamiDB Cloud the same engine as the open-source NamiDB?

Yes. Cloud runs the open-source engine, so the Cypher reference, the value encoding and the error codes are the same. Cloud adds the managed parts: accounts, organizations, keys, limits, backups, the console and billing.

Can I move between Cloud and self-hosted?

Yes. An export is JSON Lines plus an import.cypher script that loads into a self-hosted server (or Neo4j). The other way, load your data with the SDK bulk helpers, the CLI’s import, or UNWIND $rows batches.

Do I pay for idle databases?

Not for compute. Idle databases stop using compute and wake on the next query; you pay for the storage they use and the requests they serve. See Scale to zero.

Where is my data stored?

In a storage bucket dedicated to your organization, in the storage location assigned when the organization was created. See the Subprocessors and Security pages.

Is there a free tier?

Yes: the Free plan, with 2 databases, 2 GB, 100,000 writes and 1 million reads a month, hard-capped so it is never billed. It pauses after 7 days without activity.

Keys and authentication

401 unauthorized

  • The key is wrong, revoked or expired: check API keys in the console.
  • You sent a management token to the data API (or an API key to the management API). Each credential works only with its own API.
  • The header is Authorization: Bearer nmdb_live_… (with Bearer ).

403 database_not_in_scope

The key covers other databases. Create a key that covers this one, or use All databases.

403 read_only_key

A write (CREATE, MERGE, SET, DELETE, schema changes) was sent with a read-only key.

403 ip_not_allowed

The organization or the key has an IP allowlist and your address isn’t on it. The message names the address NamiDB saw. Add it under API keys → IP allowlist. See IP allowlists.

I lost my API key

Keys are shown once and can’t be retrieved. Revoke it and create a new one, or Rotate it to keep the old one working for a while.

Performance

The first query takes longer

The database was idle and its engine had stopped (after 10, 30 or 60 minutes on Free, Pro or Scale). The first request waits while it starts; the next ones are fast. On Scale, turn on Keep warm. The SDKs and drivers wait for it; if you see 503 engine_starting, retry after Retry-After.

429 rate_limited

More requests per second or concurrent queries than the plan allows. Back off for Retry-After, batch writes with UNWIND $rows, and reuse one client per process. See Rate limits.

504 timeout

The query ran past the plan’s timeout (10 s / 30 s / 60 s). Use EXPLAIN to look at the plan, add an index or constraint on the properties you filter or MERGE on, add LIMIT, and narrow variable-length patterns. The Activity tab’s slow-query log groups slow statements by shape.

413 result_too_large

The result is over 64 MiB. Return fewer rows (LIMIT, SKIP) or fewer properties.

413 query_memory_limit

The query needs more working memory than your plan’s database size allows for one query. It fails the same way every time, so don’t retry it as is: filter and add LIMIT earlier, paginate, return fewer or smaller values, and aggregate instead of collect()ing raw rows. See Errors.

503 memory_pressure

The database is briefly short of memory for queries, usually because several large queries are running at once. Retry with exponential backoff (the Neo4j drivers do this for you). If it persists, make large queries smaller or run fewer of them at the same time.

Limits and billing

429 plan_limit

An allowance is used up with the spend cap on (writes are blocked when storage or write requests run out; everything when reads run out), or the database is read-only after a downgrade. Check Usage. Upgrade, turn the cap off (paid plans), delete data, or wait for the period to reset.

403 org_paused

A Free organization paused after 7 days without activity. An owner or admin presses Resume in the console.

403 org_suspended

NamiDB suspended the organization; the message says why. Contact support. Your data stays exportable.

Drivers and tools

My Neo4j driver can’t connect

  • Use neo4j+s://db.namidb.com:7687 or bolt+s://…; TLS is required.
  • Username: anything; password: the API key.
  • Name the database when the key reaches more than one.
  • Port 7687 must be open outbound on your network.

MCP client shows no tools

Check the URL includes the database (/v1/<database>/mcp) and the Authorization: Bearer header is sent. The client must support remote (Streamable HTTP) servers.

Integers lose precision in JavaScript

JSON numbers beyond 2^53 lose precision in JavaScript. Store large identifiers as strings.

Data

How do I undo a mistake?

Restore to a point in time before it: database page → Restore to a point in time. It creates a new database holding the original as it was then: exactly, to the second, within the last 48 hours on Pro and 7 days on Scale; from the newest backup before that time otherwise. See Backups for the recovery point per plan.

My import is slow

Create a uniqueness constraint on the key you MERGE on before importing, so each row is a lookup instead of a scan. Import nodes before relationships.

Can I restore over an existing database?

No. Restores always create a new database, so live data is never overwritten. Point your application at the new one, or delete the old one and restore again under its name.