chore(deps): bump github.com/redpanda-data/connect/public/bundle/free/v4 from 4.99.0 to 4.101.0 in /collector in the benthos group across 1 directory

[dependabot]

Bumps the benthos group with 1 update in the /collector directory: github.com/redpanda-data/connect/public/bundle/free/v4.

Updates github.com/redpanda-data/connect/public/bundle/free/v4 from 4.99.0 to 4.101.0

Release notes

Sourced from github.com/redpanda-data/connect/public/bundle/free/v4's releases.

v4.101.0

For installation instructions check out the getting started guide.

Added

• aws_dynamodb_cdc: Added optional checkpoint_namespace field to allow multiple independent pipelines to safely share a single DynamoDB checkpoint table without overwriting each other's state. (@​squiidz, #4602)
• oracledb_cdc: Added snapshot_filters configuration to enable per-table SQL SELECT overrides during initial snapshot capture. (@​josephwoodward, #4606)

Fixed

• snowflake_streaming: Fixed potential silent data corruption by verifying parquet row counts match actual serialized rows before upload to Snowflake. (@​squiidz, #4605)

The full change log can be found here.

v4.100.0

For installation instructions check out the getting started guide.

Added

• jira: Added a new Jira input component that streams issues, comments, and changelog events into Connect pipelines with cursor-based incremental polling and deduplication of boundary issues. (@​squiidz, #4484)

Fixed

• general: Embedded the IANA timezone database in the CGO-enabled build to ensure timezone-aware features work correctly in minimal container runtimes. (@​squiidz, #4583)
• iceberg: Made Iceberg commit retries idempotent to prevent duplicate data files when a commit lands server-side but the client observes a transient failure. (@​Jeffail, #4591)
• oracledb_cdc: Added support for bare rowid-based predicates in Oracle LogMiner SQL_REDO parsing and improved debug logging for query failures. (@​josephwoodward, #4585)
• oracledb_cdc: Hardened Oracle LogMiner SQL_REDO parser and LOB assembly against panics and data corruption by fixing hex decoding, adding UNISTR national character support, validating LOB offsets, and preventing empty primary key sets from matching vacuously. (@​Jeffail, #4586)
• postgres_cdc: Fixed data loss during crashes by adding an acknowledgement barrier that ensures snapshot rows are fully acknowledged before promoting the replication slot. (@​squiidz, #4584)
• splunk: Fixed Splunk integration test TLS configuration and startup timeout to properly verify certificates and allow sufficient time for Splunk to boot. (@​squiidz, #4582)

Changed

• iceberg: Optimized Iceberg sink performance by caching schema field metadata to reduce per-record allocations, and added tuning documentation with recommended configuration for throughput optimization. (@​Jeffail, #4590)

Unreleased

Fixed

• general: The CGO-enabled distribution binary now embeds the IANA time zone database via the timetzdata build tag, matching the other distributions, so time.LoadLocation works in minimal runtimes without system tzdata instead of silently falling back to UTC (which shifts JQL date predicates in the jira input). (@​squiidz, #4583)

The full change log can be found here.

Changelog

Sourced from github.com/redpanda-data/connect/public/bundle/free/v4's changelog.

4.101.0 - 2026-07-17

Added

• aws_dynamodb_cdc: Added optional checkpoint_namespace field to allow multiple independent pipelines to safely share a single DynamoDB checkpoint table without overwriting each other's state. (@​squiidz, #4602)
• oracledb_cdc: Added snapshot_filters configuration to enable per-table SQL SELECT overrides during initial snapshot capture. (@​josephwoodward, #4606)

Fixed

• snowflake_streaming: Fixed potential silent data corruption by verifying parquet row counts match actual serialized rows before upload to Snowflake. (@​squiidz, #4605)

4.100.0 - 2026-07-09

Added

• jira: Added a new Jira input component that streams issues, comments, and changelog events into Connect pipelines with cursor-based incremental polling and deduplication of boundary issues. (@​squiidz, #4484)

Fixed

• general: Embedded the IANA timezone database in the CGO-enabled build to ensure timezone-aware features work correctly in minimal container runtimes. (@​squiidz, #4583)
• iceberg: Made Iceberg commit retries idempotent to prevent duplicate data files when a commit lands server-side but the client observes a transient failure. (@​Jeffail, #4591)
• oracledb_cdc: Added support for bare rowid-based predicates in Oracle LogMiner SQL_REDO parsing and improved debug logging for query failures. (@​josephwoodward, #4585)
• oracledb_cdc: Hardened Oracle LogMiner SQL_REDO parser and LOB assembly against panics and data corruption by fixing hex decoding, adding UNISTR national character support, validating LOB offsets, and preventing empty primary key sets from matching vacuously. (@​Jeffail, #4586)
• postgres_cdc: Fixed data loss during crashes by adding an acknowledgement barrier that ensures snapshot rows are fully acknowledged before promoting the replication slot. (@​squiidz, #4584)
• splunk: Fixed Splunk integration test TLS configuration and startup timeout to properly verify certificates and allow sufficient time for Splunk to boot. (@​squiidz, #4582)

Changed

• iceberg: Optimized Iceberg sink performance by caching schema field metadata to reduce per-record allocations, and added tuning documentation with recommended configuration for throughput optimization. (@​Jeffail, #4590)

Unreleased

Added

• aws_dynamodb_cdc: DynamoDB CDC now supports an optional checkpoint_namespace field, allowing multiple independent pipelines to share a single checkpoint table without overwriting each other's checkpoints. (@​squiidz, #4602)

Fixed

• general: The CGO-enabled distribution binary now embeds the IANA time zone database via the timetzdata build tag, matching the other distributions, so time.LoadLocation works in minimal runtimes without system tzdata instead of silently falling back to UTC (which shifts JQL date predicates in the jira input). (@​squiidz, #4583)
• snowflake_streaming: BDEC parquet files are now verified for internally consistent row counts (footer, row groups, and column chunks) before upload, so an inconsistent file fails the batch instead of being registered with Snowflake. (@​squiidz)

Commits

• a25598a Update changelog for v4.101.0 (#4608)
• d736bd6 aws_dynamodb_cdc: add checkpoint_namespace for checkpoint table sharing (#4602)
• c9136a9 oracledb_cdc: support snapshot filtering (#4606)
• 9b0281b docs(redpanda_migrator): fix stale on-demand schema sync claim
• e60ee50 snowflake_streaming: verify parquet row counts before upload
• e76f14f jira: register integration test package in packages.json (#4603)
• 090f2aa Update changelog for v4.100.0
• 5800e63 Bump to Go 1.26.5
• f1c9314 postgres_cdc: wait for snapshot acks before promoting the replication… (#4584)
• c724d25 iceberg: make commit retries idempotent to prevent duplicate data files
• Additional commits viewable in compare view