Atlas Online Archive sharding identity safety and orphan recovery
Parent: MongoDB Atlas Online Archive · Published reference · snapshot 2026-10-29
↓ Facts as markdownall context files
Source-anchored research for Atlas Online Archive sharding identity safety and orphan recovery. Run status: COMPLETED.
These notes link each claim to its source. Some citations identify archived research reports that are no longer hosted on this site. A published reference means the content is available; it does not certify independent review or accuracy.Read the editorial policy and follow the sources before relying on a claim.
How it works
- When resharding finishes for a collection archived by Atlas Online Archive, its online archive files are marked Orphaned. Orphaning therefore does not require a missing cluster: collection deletion or resharding can orphan an archive associated with an active or paused archive configuration. (high confidence) (verified as of 2026-10-14) [source]
- Moving an archived collection with moveCollection can orphan its Online Archive files when the move completes. The move tutorial, database command, and mongosh method all document this effect. (medium confidence) (verified as of 2026-10-14) [source]
- Both the unsharding tutorial and `unshardCollection` reference say associated Online Archive files become `Orphaned` when unsharding completes. These are two pages from one vendor, not independent organisational corroboration. (high confidence) (verified as of 2026-10-14) [source]
- The selective-restore example uses explicit `on` fields with `whenMatched: keepExisting` and `whenNotMatched: insert`; it warns about duplicate archived `_id` values and requires a compound unique index when several `on` fields identify a match. These are identity-resolution considerations, not a guaranteed orphan-reattachment recipe. (low confidence) [source]
- Resharding fails when `_id` values are not globally unique, to avoid corrupting collection data; duplicate IDs can also prevent chunk migration. The manual directs operators to copy duplicate documents into new documents and then remove the duplicates. (low confidence) [source]
Problems, failure modes and limitations
- MongoDB documents that Online Archive deletes documents from the live cluster using _id and any shard keys. Its safety warning directs operators to pause the archive before sharding operations. (low confidence) (verified as of 2026-10-14) [source]
- Historical identity-risk context: in its February 21, 2014 TokuMX 1.4 engineering post, Percona explains how manually duplicated _id values can reside on different shards when the shard key differs, and why migration uniqueness checks matter. Progress separately warns about duplicate cross-shard identifiers. These sources support identity validation as a concern, not an Atlas Online Archive incident attribution. (medium confidence) [source]
- When `_id` is not the shard key, its default unique index does not enforce uniqueness across shards; applications must ensure global uniqueness. Progress independently warns about duplicate identifiers in sharded writes, and Percona describes the corresponding duplicate-ID migration hazard in historical TokuMX 1.4. Those sources support the identity risk, not Atlas archive-state behavior. (high confidence) [source]
Comparisons and alternatives
- Pausing is a safety prerequisite for sharding operations, but the reviewed documentation provides no exemption from the documented Orphaned outcome after completed resharding. A runbook should check the resulting archive association instead of assuming a pause preserves it. (medium confidence) (verified as of 2026-10-14) [source]
Facts and statements
- MongoDB directs applications to refrain from updating documents once they become eligible for archival. Archive eligibility therefore creates a documented application-level mutation constraint. (low confidence) (verified as of 2026-10-14) [source]
- Applications must ensure _id uniqueness across documents used with Online Archive; a sharded deployment must not assume cluster-wide uniqueness merely from an _id field. Progress independently warns that cross-shard identifier uniqueness requires appropriate configuration. (medium confidence) (verified as of 2026-10-14) [source]
- After resharding, another archive can use the same database, collection, and fields provided no archive for that combination is `Active`; moveCollection and unsharding documentation state the same condition. This describes replacement creation, not reattachment of the old archive. (high confidence) (verified as of 2026-10-14) [source]
- MongoDB states that an Orphaned archive does not automatically lose its archived data and directs operators to delete orphaned archives manually. Operational inference: treat that cleanup as a separate data-retention decision before creating or using a replacement; it is not evidence that historical data has been reattached. (low confidence) (verified as of 2026-10-14) [source]
- Online Archive identifies documents for deletion from the live cluster using `_id` together with any shard keys. (low confidence) [source]
- MongoDB instructs operators to pause archiving before sharding operations, avoid updating archive-eligible documents, and ensure application-wide `_id` uniqueness to prevent data loss. (low confidence) [source]
- MongoDB documents `$merge` for rehydrating archived documents to an Atlas cluster and instructs users to verify restored data before deleting the archive. This general restore procedure does not explicitly guarantee that every orphaned archive remains usable as its source. (high confidence) [source]
Children
- No children recorded.