v25.1.6

Bug Fixes

• Fix Avro translation to Iceberg when root Avro schema is a primitive type. by @nvartolomei in #26461
• Fix an issue where audit log could lock down a cluster, if miss-configured. Now it is always possible to disable it. by @IoannisRP in #26652
• Iceberg integration: Encode avro record field names to avoid using disallowed characters (like dots with default partitioning hour(redpanda.timestamp). This makes it possible to read Iceberg data with latest version of DuckDB. by @nvartolomei in #26535
• prevents Redpanda from crashing when reading invalid record data by @mmaslankaprv in #26492

Improvements

• Adds support for the Iceberg table properties write.metadata.path and write.data.path. When an Iceberg catalog defines these properties, Redpanda will use them to determine where to write Iceberg table metadata and data, respectively, instead of using default locations based on the table location. by @wdberkeley in #26440
• Allows direct uploading of debug bundles collected with rpk debug remote-bundle by @JFlath in #26515
• Cut down the amount of time spent in fstat() syscalls during storage layer housekeeping & cut down the amount of time spent in fstat() syscalls in the storage layer EVEN MORE IN GENERAL! by @WillemKauf in #26656
• Fixed large allocation issues when handling OffsetCommits by @mmaslankaprv in #26414
• Fixes an issue in which users could experience oversized allocations during a DescribeGroup request. by @WillemKauf in #26532
• ability to control batch cache settings for __consumer_offsets topic by @mmaslankaprv in #26558

Full Changelog: v25.1.5...v25.1.6