Known Limitations for Self-managed mongot

This page describes limitations and unsupported behavior for self-managed mongot deployments.

Limitations by Category

The following sections describe known limitations for self-managed mongot deployments, grouped by category.

Deployment, Topology, and Platform Support

Limitation	Applies to	Description
mongot Standalone Tarball or Container Isn't a Supported Enterprise Production Path	Enterprise Advanced.	Running `mongot` from a standalone tarball or container is not a supported production path for Enterprise Advanced. Enterprise production support assumes MCK-managed `mongot`, with the related operational tooling, monitoring integration, and support coverage.
atlas-local Docker Image Is Not a Production Deployment Target	The `atlas-local` Docker image.	The `atlas-local` Docker image is a single-container, single-node setup that is intended for local development and testing, and is not suitable for production. For production, use a supported production deployment path.
mongot Isn't Supported on IBM Power (ppc64le) or IBM Z (s390x)	All editions and deployment paths.	`mongot` isn't supported on IBM Power (`ppc64le`) or IBM Z (`s390x`). `mongot` is supported only on Linux for `x86_64` and `aarch64`.
Windows and macOS Aren't Supported as Native Deployment Targets	All editions and deployment paths.	MongoDB doesn't provide native Windows or macOS `mongot` binaries, because `mongot` is built for Linux. For local development and testing, deploy `mongot` using the `atlas-local` Docker image. For more complex deployments, manage `mongot` using the MongoDB Controllers for Kubernetes Operator or configure the `mongodb-community-search` container image to connect to your MongoDB environment.

Packaging and Installation

Limitation	Applies to	Description
The Linux Tarball Doesn't Install a systemd Service	Community Edition tarball installs.	The Linux tarball doesn't include a `systemd` unit. You can author your own `systemd` unit or use the container image for process lifecycle management.
Package Manager Installation Isn't Available for Community Edition	Community Edition.	`apt` and `yum` installation isn't available for Community Edition. To install `mongot`, use the tarball or container image.

Authentication and Security

Limitation	Applies to	Description
X.509 Certificate Rotation Requires a mongot Restart	All deployment paths.	`mongot` reads X.509 certificates at startup. To pick up new certificates, you must restart `mongot` after rotating X.509 certificates. To minimize interruptions, schedule certificate rotations during maintenance windows.
SCRAM Password Rotation Requires a mongot Restart	All deployment paths.	`mongot` reads SCRAM credentials at startup. To pick up new credentials, you must restart `mongot` after rotating SCRAM credentials. To minimize interruptions, schedule SCRAM credential rotations during maintenance windows.
mongot Doesn't Validate OCSP or CRL Revocation	All deployment paths that use X.509 or TLS authentication.	`mongot` doesn't validate X.509 or TLS certificate revocation through OCSP or CRLs. To handle revocation, remove the corresponding `$external` user on `mongod`. After removal, the next handshake rejects the revoked certificate.
mongot Doesn't Provide Application-Level Encryption at Rest	All deployment paths.	`mongot` doesn't provide native application-level encryption at rest. Instead, you can encrypt the underlying filesystem or block device, for example with LUKS or `dm-crypt` on Linux, cloud-provider-managed encrypted volumes, or KMS-backed disk encryption.
FIPS-Validated TLS Mode Is Not Supported	All deployment paths.	`mongot` doesn't support running its TLS stack in a FIPS-validated cryptographic mode.

Observability

Limitation	Applies to	Description
Atlas Search Metrics UI Isn't Available for Self-Managed Deployments	All self-managed deployments.	The MongoDB Search Metrics UI isn't available with self-managed `mongot`. Use Prometheus and Grafana, or another monitoring stack, with the `mongot` `/metrics` endpoint. To learn more, see Monitoring Tool Integrations.
Ops Manager Visibility Not Comprehensive for `mongot`	Enterprise Advanced deployments that use Ops Manager and MongoDB Controllers for Kubernetes Operator.	In Enterprise Advanced, Ops Manager ingests and displays a limited set of metrics through an MongoDB Controllers for Kubernetes Operator-deployed OpenTelemetry pipeline. These metrics are curated for support triage and health visibility, but not for deep diagnostics.

Operations and Upgrades

Limitation	Applies to	Description
One mongot Instance Can't Serve Multiple mongod Deployments	All deployment paths.	A single `mongot` instance can't be shared across multiple unrelated MongoDB deployments because a `mongot` instance is bound to exactly one replica set, or to one shard in a sharded cluster. To serve queries for multiple MongoDB deployments, deploy a separate `mongot` instance for each replica set or shard.
Major-Version Downgrades Require Re-sync	All deployment paths.	Downgrading `mongot` across major versions requires a re-sync between `mongod` and `mongot` because major versions aren't rollback-safe.
mongot Upgrades Require a Process Restart	All deployment paths.	`mongot` upgrades require a process restart. To minimize interruptions, schedule maintenance windows or use a rolling restart pattern with multiple `mongot` instances under MongoDB Controllers for Kubernetes Operator.
mongot Configuration Changes Require a Restart	All deployment paths.	`mongot` reads its configuration at startup. To apply configuration changes, you must restart `mongot` after editing the `mongot` configuration file.

Atlas Versus Self-Managed Feature Comparison

The following sections compare support for MongoDB Search and MongoDB Vector Search features between Atlas and self-managed mongot. Use the tables in this section to compare options for a migration or hybrid deployment.

Search and Vector Search Behavior

Search and Vector Search share the same search engine on Atlas and self-managed mongot, so query behavior, scoring, and index semantics are generally consistent across platforms.

The following table compares which Search and Vector Search features are available between Atlas-managed and self-managed mongot.

Search and Vector Search Feature	Feature Availability
`$search`, `$searchMeta`, and `$vectorSearch` aggregation stages	Same on Atlas and self-managed
Index definition syntax	`nestedRoot` syntax not supported on self-managed `mongot`. To learn more, see How to Index Fields for Vector Search.
`createSearchIndex`, `dropSearchIndex`, `updateSearchIndex`, and `getSearchIndexes` helper methods	Same on Atlas and self-managed
`$listSearchIndexes` aggregation stage	Output for self-managed `mongot` differs from Atlas Search. See `$listSearchIndexes` for the Atlas Search output shape. Self-managed `mongot` output differs in the following fields: No top-level `type` field `synonymMappingStatusDetail` is a flat list of status objects, not a nested document
Lucene-based scoring and ranking	Same on Atlas and self-managed
Synonyms, analyzers, and custom mappings	Same on Atlas and self-managed
Vector quantization options	Same on Atlas and self-managed
`$rerank`	Available only on Atlas.
Automated Embedding through Voyage AI	Preview on both platforms. To learn which Automated Embedding features are supported for self-managed `mongot`, see Automated Embedding.

Automated Embedding

The following table compares Automated Embedding feature support between Atlas and self-managed mongot. To learn more about automated embedding, see Automated Embedding Overview.

Automated Embedding Feature	Atlas	Self-Managed `mongot`
Voyage AI integration	Available in Preview with the same Voyage AI models and options as self-managed `mongot`.	Available in Preview with the same Voyage AI models and options as Atlas.
Rate-limit visibility	Atlas exposes RPM and TPM limits in the UI.	Rate-limit visibility isn't surfaced in the product. Limits are enforced on the Voyage AI side.
Billing path	Atlas can bill embeddings through the MongoDB proxy endpoint.	Billing depends on whether you use the MongoDB proxy endpoint or connect directly to Voyage AI.

Deployment and Lifecycle Management

The following table compares how Atlas and self-managed deployments handle operational aspects of deploying and managing the lifecycle of mongot.

Operational Aspect	Atlas	Self-managed `mongot`
Provisioning	MongoDB provisions and runs `mongot`.	You provision `mongot` by using a tarball, container, MongoDB Controllers for Kubernetes Operator, or `atlas-local`.
`mongot` version visibility	Atlas exposes release timing rather than direct `mongot` version strings.	`mongot` packages are versioned with explicit `MAJOR.MINOR.PATCH` versions.
Upgrades	MongoDB performs Atlas upgrades on its own cadence.	Upgrades are entirely self-managed.
Security patching	MongoDB applies patches on Atlas.	Security patching requires self-managed upgrades.
Multi-instance scaling	MongoDB Search Nodes handle scale-out within the managed service.	For multiple search nodes or sharded topologies, use MongoDB Controllers for Kubernetes Operator or manage the topology yourself.
Sharded clusters	Supported in Atlas.	Supported only with the MongoDB Controllers for Kubernetes Operator for Enterprise Advanced. The MongoDB Controllers for Kubernetes Operator doesn't support sharded architectures with Community Edition.

Monitoring and Observability

The following table compares which monitoring tools and observability options are available for Atlas and self-managed mongot.

Monitoring and Observability Option	Atlas	Self-managed `mongot`
MongoDB Search Metrics UI	Available in the Atlas UI.	Not available. To learn more, see Observability.
MongoDB-managed alerts	Curated alerts with default thresholds.	You define alerting yourself. To learn more, see Recommended Alerts for mongot.
FTDC retention	Managed by MongoDB.	You manage rotation and retention.
Log retention	Managed by MongoDB.	You forward and retain logs.
`mongot`-metric dashboards	Available in the Atlas UI.	Not provided in a UI component. To learn how to build dashboards from `mongot` metrics, see Metrics Reference for mongot.
Ops Manager `mongot` visibility	Not available in Atlas.	In Enterprise Advanced, Ops Manager exposes curated `mongot` metrics through an MongoDB Controllers for Kubernetes Operator-deployed OpenTelemetry pipeline. To learn more, see Observability.

Reliability and Availability

The following table compares how Atlas and self-managed deployments handle reliability and availability capabilities for mongod and mongot.

Reliability and Availability Capability	Atlas-managed Deployment	Self-managed Deployment
`mongot` instance availability	Atlas manages recovery behavior.	You are responsible for supervision and recovery. Use MongoDB Controllers for Kubernetes Operator-managed replicas or your own process supervision.
Backup and restore	Atlas Cloud Backup manages `mongod` snapshots. Search indexes in `mongot` can be rebuilt from data restored from MongoDB database snapshots.	You are responsible for backing up `mongod`. Search indexes in `mongot` can be rebuilt from data restored from MongoDB database snapshots.
Disaster recovery	Atlas provides regional and cloud failover patterns.	You design the disaster recovery topology yourself.
Capacity scaling	Atlas provides managed Search Node scaling.	You size capacity up front and rescale by changing `mongot` resources and restarting as needed.

Security

The following table compares how Atlas-managed deployments and self-managed deployments handle security for mongod and mongot.

Security Feature	Atlas-managed Deployment	Self-managed Deployment
Authentication between `mongod` and `mongot`	Managed internally by Atlas.	You configure SCRAM or X.509 authentication.
TLS for `mongod` to `mongot`	Always on and managed by Atlas.	You configure TLS yourself.
Encryption at rest	Atlas provides managed encryption-at-rest options, including KMS integrations.	`mongot` doesn't provide native encryption at rest. To learn more, see Authentication and Security.

Back

Troubleshoot

Release Notes