# Automated rolling builds in Atlas, Cloud Manager and Ops Manager > Source-anchored research for Automated rolling builds in Atlas, Cloud Manager and Ops Manager. Run status: COMPLETED. Parent: [Rolling index builds](https://llms-explorer.com/tree/rolling-index-builds/) · 3 facets · 19 facts · page: https://llms-explorer.com/tree/automated-rolling-builds-in-atlas-cloud-manager-and-ops-manager/ ## How it works - Rolling builds can reduce the index-build load on serving members, but they reduce cluster resiliency and increase build duration. MongoDB recommends them only when ordinary builds do not meet the workload's needs; Percona independently describes their value for systems already close to capacity. (medium confidence) — [source](https://www.mongodb.com/docs/atlas/atlas-ui/indexes.md) - Do not overlap rolling and replicated/non-rolling builds. Cloud Manager and Ops Manager also require successive rolling requests to wait for the current build to finish; Atlas warns concurrent rolling and replicated builds can cause broken builds and crash loops. (high confidence) (verified as of 2026-10-31) — [source](https://www.mongodb.com/docs/atlas/atlas-ui/indexes.md) - SERVER-71769 documents historical Atlas incidents where secondaries had an index missing from the primary, which engineers suspected came from interrupted rolling builds. A subsequent replicated build could stall because those secondaries skipped commit-quorum voting. The issue records adjusting setIndexCommitQuorum as a workaround; it does not establish that every current version still exhibits the defect. (low confidence) — [source](https://jira.mongodb.org/browse/SERVER-71769) - Percona's July 28, 2025 engineering walkthrough independently warns that a rolling build needs an oplog window that covers the member's offline indexing interval, with extra margin recommended for safety. This is an underlying replication constraint relevant when assessing managed automation; its Kubernetes-specific commands are not an Atlas, Cloud Manager or Ops Manager procedure. (low confidence) — [source](https://www.percona.com/blog/how-to-perform-rolling-index-builds-with-percona-operator-for-mongodb/) - Rolling builds can reduce build-time performance impact, but increase total build time and reduce cluster resiliency. MongoDB recommends them only when regular index builds do not meet workload needs. (high confidence) (verified as of 2026-10-31) — [source](https://www.mongodb.com/docs/atlas/atlas-ui/indexes.md) ## Problems, failure modes and limitations - The historical SERVER-68925 warning covers automated builds with affected agents on MongoDB 4.2.19–4.2.22, 4.4.13–4.4.16, 5.0.6–5.0.11 and 6.0.0–6.0.1. Documented fixes are Cloud Manager Agent 12.4.0.7703+ or Ops Manager Agent 12.0.11.7606+, or server upgrades to 4.2.23+, 4.4.17+, 5.0.12+ or 6.0.2+ within those release lines; this is historical remediation, not a recommendation to deploy obsolete branches. (high confidence) — [source](https://www.mongodb.com/docs/cloud-manager/data-explorer/indexes.md) ## Facts and statements - Atlas, Cloud Manager and Ops Manager expose rolling index creation through their index-management interfaces. Cloud Manager and Ops Manager remove one node at a time, beginning with a secondary; Atlas explicitly documents taking each member offline, building in standalone mode, and returning it to the replica set. (medium confidence) — [source](https://www.mongodb.com/docs/atlas/atlas-ui/indexes.md) - The Atlas CLI exposes the Admin API through `atlas api rollingIndex createRollingIndex`, providing a programmatic entry point for creating an index on a named cluster in a rolling manner. (low confidence) (verified as of 2026-10-31) — [source](https://www.mongodb.com/docs/atlas/cli/current/command/atlas-api-rollingIndex.md) - Atlas UI documentation limits rolling index builds to M10+ clusters and rejects the combination of a rolling build with the unique option. (low confidence) (verified as of 2026-10-31) — [source](https://www.mongodb.com/docs/atlas/atlas-ui/indexes.md) - Cloud Manager and Ops Manager Data Explorer do not support rolling builds for standalone deployments. Their rolling-build option lists unique, storageEngine, textIndexVersion and 2dsphereIndexVersion as incompatible and says those options are ignored if supplied. (medium confidence) (verified as of 2026-10-31) — [source](https://www.mongodb.com/docs/cloud-manager/data-explorer/indexes.md) - Cloud Manager and Ops Manager automatically cancel a rolling build that fails on some nodes, remove successfully built copies, publish an activity event and email the project owner. Their docs nevertheless require listIndexes on every node after an aborted build to verify cleanup and avoid inconsistent indexes. (high confidence) (verified as of 2026-10-31) — [source](https://www.mongodb.com/docs/cloud-manager/data-explorer/indexes.md) - MongoDB's September 19, 2022 advisory records an agent defect that could introduce index inconsistencies specifically during automated rolling builds in Atlas, Ops Manager and Cloud Manager. It says Atlas received a fix, while previously affected Atlas clusters could retain damage; the advisory is historical evidence, not evidence that today's Atlas service remains vulnerable. (medium confidence) — [source](https://www.mongodb.com/resources/products/alerts) - COMPASS-10970, updated August 24, 2026, states that Atlas 'infinite' clusters with disaggregated storage do not support rolling index builds even though they are dedicated clusters. Its scope is UI capability gating; this limits any inference that every dedicated Atlas deployment supports rolling builds. (low confidence) (verified as of 2026-10-31) — [source](https://jira.mongodb.org/browse/COMPASS-10970) - A 2020 Stack Overflow question reports production performance problems during background builds on a large sharded dataset and asks for automation; the 2021 answer points to Ops Manager Data Explorer. This is practitioner motivation, not a benchmark or proof of current product reliability. (low confidence) — [source](https://stackoverflow.com/questions/62272251/mongodb-rolling-index-build) - Atlas, Cloud Manager and Ops Manager expose an optional rolling index build through their index-creation UI; Ops Manager's Data Explorer workflow is also identified in a 2021 community answer. (high confidence) (verified as of 2026-10-31) — [source](https://www.mongodb.com/docs/atlas/atlas-ui/indexes.md) - Atlas automates removing each replica-set node, starting it as a standalone to build the index, and returning it to the replica set. Cloud Manager and Ops Manager describe removing one node at a time, starting with a secondary, for replica sets and sharded clusters. (high confidence) (verified as of 2026-10-31) — [source](https://www.mongodb.com/docs/atlas/atlas-ui/indexes.md) - Atlas documents rolling index builds for M10+ clusters. Cloud Manager and Ops Manager Data Explorer exclude standalone deployments from their rolling workflow. (high confidence) (verified as of 2026-10-31) — [source](https://www.mongodb.com/docs/atlas/atlas-ui/indexes.md) - Atlas rejects a rolling build configured with unique. Cloud Manager and Ops Manager instead document ignoring incompatible unique, storageEngine, textIndexVersion and 2dsphereIndexVersion options in the rolling-build Options pane. Operators must not assume requested uniqueness was enforced. (high confidence) (verified as of 2026-10-31) — [source](https://www.mongodb.com/docs/atlas/atlas-ui/indexes.md) - Percona's 2022 manual walkthrough illustrates the work automated orchestration replaces: restart a secondary standalone, build locally, restore replication, wait for catch-up, repeat, and step down the original primary before processing it. It also warns that the oplog must retain enough history for catch-up. (low confidence) — [source](https://www.percona.com/blog/mongodb-index-building-on-replicaset-and-shard-cluster/)