ingestr v1.1.6 was published on July 21, 2026. The main operator concern addressed by this release is an Iceberg staging file leak, which could leave files behind after write activity. The release also updates a tracing dependency, adjusts Fabric timestamp handling, and speeds the MSSQL replace load path.
The full release notes and downloads are on the GitHub release page.
Iceberg staging cleanup targets leaked files ¶
Commit 9afd755efd8edb5d1695fd3db3b7d132f198cb3f fixes Iceberg staging file leaks. For data platform operators, this is the most important correctness change in the release. Files that remain in staging can consume storage outside the lifecycle of the table data that a pipeline is expected to manage.
The release notes do not identify the error path that leaked the files. They also do not say whether the fix removes files left by older ingestr versions. Treat this as a correction to current write behavior, not as an automatic cleanup job for existing residue.
After rollout, monitor the same staging storage used by representative Iceberg jobs. Compare the count and age of residual files before and after a completed run. If a separate cleanup process already handles stale files, keep it in place until the new behavior has been observed under the actual failure and retry patterns of the pipeline.
Tracing dependency update addresses W3C baggage ¶
Commit 0fca18ff6df92ddcf1ece2edbe35058e0955934c bumps dd-trace-go/v2 to 2.8.1 for a W3C baggage denial of service fix. This is a dependency security update, not a new tracing feature. The affected surface named by the changelog is baggage handling.
The notes provide no CVE, attack precondition, or configuration detail. That limits any claim about exposure, but it does not change the deployment requirement: old ingestr processes still contain the old dependency. Replace the running artifact and restart persistent workers as part of the rollout, then confirm the deployed build is v1.1.6.
Fabric timestamp evolution and MSSQL replace throughput ¶
Two entries cover Fabric. The merge record names fabric-schema-evolution-timestamptz, while commit a9f702aa8be4afc7155dc2a8a510b8d6f9b9b077 is labeled fabric-timestamp. The branch name points specifically to timestamp with time zone handling during schema evolution.
The public notes do not describe the resulting type mapping, generated DDL, or any migration requirement. Fabric users should therefore test schema evolution against a table containing timestamp with time zone values. Check both the resulting column type and representative values rather than assuming the terse changelog defines the full contract.
For SQL Server, commit cd105b35356cd14be9fea70b9b8dcbb5c293250a speeds up replace bulk loads. The scope is narrow: the MSSQL bulk load path when replace behavior is used. No benchmark, row count, batch size, or latency figure is included, so the release does not support a general performance claim for other load modes.
Measure this change with the same replace workload and batch shape used in production. The changelog makes no performance claim for append or CDC paths, and those paths should not be included when reporting the effect of this release.
Release scope ¶
This is a regular release, not a prerelease. The changelog contains four concrete change entries and two merge records. It lists no breaking configuration change, new CLI flag, or migration command.
That makes the rollout mechanically small but still worth targeted verification. Inspect Iceberg staging residue, replace old tracing builds, validate Fabric timestamp evolution, and benchmark the MSSQL replace path. Each check maps directly to a stated v1.1.6 change, without assigning broader behavior to a short release note.