Netherite 1.3.5

What's Changed

This release includes several high-priority bug fixes:

• Fix excessive memory use introduced in release 1.3.3 #266 (can cause OOM and performance degradation)
• Fix double application of partition events #252 (can cause incorrect instance counts and possibly other state corruptions)
• fix handling of expired groups in ReassemblyState #256

As well as several improvements to reliability, including optimizations and lower-priority bugs:

• Optimize host shutdown #249
• Fix race condition in OutboxState #250
• Fix race in memory tracker #254
• Fix null metrics exceptions #248
• Improve trace messages for non-executable instances #258
• update fatal exception handling #259
• fix LoadPublisher.FlushAsync #262
• limit frequency of failing partition starts #261
• Fixes to azure storage calls #260

Full Changelog: 1.3.4...1.3.5

Netherite 1.3.4

What's Changed

This patch release contains two fixes:

• Fix ignored tuning parameter #243
• Fix instance count assertion crashes #247

lt also contains a new example that shows how to use an entity to implement a semaphore:

• Add semaphore example to PerformanceBenchmarks #242

Full Changelog: 1.3.3...1.3.4

Netherite 1.3.3

What's Changed

This patch release fixes several reliability issues. We recommend that everyone upgrade as soon as possible.

• Fix fragmentation/reassembly logic #233. This PR fixed a bug that could under some circumstances cause messages received by partitions to get corrupted and lost during fragmentation/reassembly.
• fix various issues with client shutdown #237. Fixes null reference exception observed when using a client on a host that is shutting down.
• Fix recovery failures caused by changing FASTER's memory size parameter #235. Fixes a problem introduced in release 1.3.2 which could cause partitions to go into an infinitely failing recovery cycle.

Full Changelog: 1.3.2...1.3.3

Netherite 1.3.2

What's Changed

This patch release includes some bug fixes.
We also updated the FASTER tuning parameters so that data is deleted from storage more quickly after it is no longer needed.

• Fix bug in BatchTimer #226
• Overhaul FASTER page and segment size parameters #230
• Add FabricBot config for automatic GitHub triaging #227

Full Changelog: 1.3.1...1.3.2

Netherite 1.3.1

What's Changed

The primary purpose of this patch release is to fix a nasty bug that can cause errors in the instance index (#219).
Also, we added the Functions isolated extension package (#212, #216).

Full Changelog: 1.3.0...1.3.1

Netherite 1.3.0

What's Changed

The most important change in this release is that it fixes a bug that can lead to data corruption under some circumstances (as observed when running tests in a consumption plan with heavily oversubscribed CPU). Since data corruption in the task hub is irreversible, we recommend that everyone updates to this version to avoid this risk.

Other changes include:

• Fix query timeouts in #202
• update benchmarks for corpus connection, periodic test in #206
• update deployment scripts for consumption plan #204
• Validate partition table name on consumption plan #207
• Move Faster to v12 blob SDK, and implement e-tag protection #208

Full Changelog: 1.2.0...1.3.0

Netherite 1.2.0

What's Changed

• send unsolicited partition load information to LoadMonitor #187
• update tooling to latest Functions version in #190
• add taskhub name validation #193
• overhaul internal layers and single-host emulation #194
• support implicit state deletion #192
• update to latest Azure Storage Tables client #195
• add support for authentication using an Azure.Core.TokenCredential #197
• fix and revise queries implementation #201
• update samples to latest version #200

Netherite 1.1.1

This small-patch release fixes a deployment issue caused by using FunctionsStartup class (#186).

Netherite 1.1.0

This 1.1.0 release includes many reliability improvements and bug fixes. Since the last release 1.0.1 there were also a number of small patches prior to this release

1.1.0

• Update to FASTER 2.0.14 (#181)
• Add a sample for running an orchestration with DTFx. (#180)
• Remove obsolete uses of ConnectionStringResolver (#171)
• Fix io notification (#179)
• Measure cache size periodically (#170)
• Remove serialization cache (#173)
• Update packages (#175)
• Fix cachesize test (#172)
• Set the default for MaxEntityOperationBatchSize (#165)
• Add support for specialized purge interface (#167)
• Fix minor bug in AzureStorageDevice (#166)
• Enforce persistence barrier for self-messages (#164)
• Fix race condition on OrchestrationRuntimeState (#161)
• Add runtime check for batch workers that are stuck (#163)
• Fix cachedebugger test (#162)

1.0.4

• Improve lease management (#157)
• Update to FASTER 2.0.3, DurableTask 2.9.0, Durable Functions 2.7.1 (#158)
• Generalize explicit placement of instances (#159)

1.0.3

• Remove Microsoft.Extensions.Logging.Abstractions dependency (#156)

1.0.2

• Fix empty batch in outbox (#154)
• Fix suppression of load publishing (#153)
• Suppress warning for netcoreapp2.2 (#152)
• Update to FASTER 2.0.2 (#147)
• Fix/Improve tracing of workitems, partition activity, and compaction (#146)

Netherite 1.0.1

This 1.0.1 release is a minor patch for the 1.0.0 release that fixes a few last-minute issues:

• fix errors in logging statements (#142)
• update to patch 2.0.1 of Faster which has a few last-minute fixes (#143)
• add more targets, including a dummy netcoreapp2.2. target, to fix package dependency issues (#144)