ingestr v1.1.10 - CDC State and Streaming Metrics

ingestr v1.1.10 was published on July 27, 2026. The central operator change is stronger CDC correctness: SQL Server gains hardened CDC behavior and destination managed state, while MongoDB incremental reads receive a fix for an interval gap. This is a stable release, not a prerelease.

The full release notes and downloads are on the GitHub release page.

The MongoDB change fixes a gap between incremental intervals. The notes do not state which boundary condition caused the gap, so they do not support claims about an exact failure mode. Even with that limitation, this is a data completeness fix. Teams that run checkpointed MongoDB ingestion should compare adjacent source intervals during rollout and verify that expected records appear once.

SQL Server receives a broader set of CDC changes. The release hardens CDC operation, adds destination managed state, fixes the MSSQL handoff, and hardens CDC transactions. Together, these changes concentrate on continuity across state transitions and transactional boundaries. The release notes do not document the state schema, persistence location, or handoff sequence, so existing recovery assumptions still need validation against a representative destination.

The project also standardized and parallelized CDC stress tests. That is test infrastructure work rather than a documented throughput improvement. It gives maintainers more coverage around the same transaction and handoff paths, but operators should not read it as a performance claim.

Destination managed state changes where CDC progress can be coordinated. For platform teams, the practical concern is restart behavior: confirm that a stopped job resumes from the expected point and that a destination restore does not silently separate data from its CDC state. The notes provide no migration procedure, compatibility matrix, or state format details.

Streaming metrics are now exposed in Prometheus format. This gives operators a scrape compatible surface for long running transfers. The release notes do not list metric names, labels, an endpoint, or a configuration switch, so dashboard and alert work should begin by inspecting the emitted series in a test deployment. Pay particular attention to label cardinality before sending the series to a shared Prometheus service.

The formatting and license audit item associated with metrics is internal maintenance. It does not describe another observable metrics feature.

The truncate insert strategy has been removed. This is the clearest compatibility risk in v1.1.10, especially for scheduled jobs that select a load strategy explicitly. The notes do not name a replacement and include no migration steps. Search deployment manifests, scheduler parameters, and generated job definitions for that strategy before upgrading, then test the resulting write behavior on a disposable destination.

One Snowflake change disables minicore in test targets. Based on the wording in the notes, this is a test scope adjustment, not a Snowflake runtime feature. The job limit was also increased, but the old and new values are not stated, so capacity plans should not assume a specific concurrency increase.

Read the complete v1.1.10 release entry before rollout. It is short, and it is the authoritative boundary for what this tag claims to change.