tikv-server v5.3.1

Feature enhancements

• Update the proc filesystem (procfs) to v0.12.0 #11702
• Improve the error log report in the Raft client #11959
• Increase the speed of inserting SST files by moving the verification process to the Import thread pool from the Apply thread pool #11239

Bug fixes

• Fix a bug that TiKV cannot delete a range of data (unsafe_destroy_range cannot be executed) when the GC worker is busy #11903
• Fix the issue that destroying a peer might cause high latency #10210
• Fix a bug that the any_value function returns a wrong result when regions are empty #11735
• Fix the issue that deleting an uninitialized replica might cause an old replica to be recreated #10533
• Fix the metadata corruption issue when Prepare Merge is triggered after a new election is finished but the isolated peer is not informed #11526
• Fix the deadlock issue that happens occasionally when coroutines run too fast #11549
• Fix the potential deadlock and memory leak issues when profiling flame graphs #11108
• Fix the rare data inconsistency issue when retrying a prewrite request in pessimistic transactions #11187
• Fix a bug that the configuration resource-metering.enabled does not work #11235
• Fix the issue that some coroutines leak in resolved_ts #10965
• Fix the issue of reporting false "GC can not work" alert under low write flow #9910
• Fix a bug that tikv-ctl cannot return the correct Region-related information #11393
• Fix the issue that a down TiKV node causes the resolved timestamp to lag #11351
• Fix a panic issue that occurs when Region merge, ConfChange, and Snapshot happen at the same time in extreme conditions #11475
• Fix the issue that TiKV cannot detect the memory lock when TiKV performs a reverse table scan #11440
• Fix the issue of negative sign when the decimal divide result is zero #29586
• Fix a memory leak caused by the monitoring data of statistics threads #11195
• Fix the issue of TiCDC panic that occurs when the downstream database is missing #11123
• Fix the issue that TiCDC adds scan retries frequently due to the Congest error #11082
• Fix the issue that batch messages are too large in Raft client implementation #9714
• Collapse some uncommon storage-related metrics in Grafana dashboard #11681

tikv-server v5.1.4

Feature enhancements

• Update the proc filesystem (procfs) to v0.12.0 #11702
• Improve the error log report in the Raft client #11959
• Increase the speed of inserting SST files by moving the verification process to the Import thread pool from the Apply thread pool #11239

Bug fixes

• Fix a bug that TiKV cannot delete a range of data (unsafe_destroy_range cannot be executed) when the GC worker is busy #11903
• Fix the issue that destroying a peer might cause high latency #10210
• Fix a bug that the any_value function returns a wrong result when regions are empty #11735
• Fix the issue that deleting an uninitialized replica might cause an old replica to be recreated #10533
• Fix the metadata corruption issue when Prepare Merge is triggered after a new election is finished but the isolated peer is not informed #11526
• Fix the deadlock issue that happens occasionally when coroutines run too fast #11549
• Fix the potential deadlock and memory leak issues when profiling flame graphs #11108
• Fix the rare data inconsistency issue when retrying a prewrite request in pessimistic transactions #11187
• Fix a bug that the configuration resource-metering.enabled does not work #11235
• Fix the issue that some coroutines leak in resolved_ts #10965
• Fix the issue of reporting false "GC can not work" alert under low write flow #9910
• Fix a bug that tikv-ctl cannot return the correct Region-related information #11393
• Fix the issue that a down TiKV node causes the resolved timestamp to lag #11351
• Fix a panic issue that occurs when Region merge, ConfChange, and Snapshot happen at the same time in extreme conditions #11475
• Fix the issue that TiKV cannot detect the memory lock when TiKV performs a reverse table scan #11440
• Fix the issue of negative sign when the decimal divide result is zero #29586
• Fix a memory leak caused by the monitoring data of statistics threads #11195
• Fix the issue of TiCDC panic that occurs when the downstream database is missing #11123
• Fix the issue that TiCDC adds scan retries frequently due to the Congest error #11082
• Fix the issue that batch messages are too large in Raft client implementation #9714
• Collapse some uncommon storage-related metrics in Grafana dashboard #11681

tikv-server v5.4.0

Improvements

• Coprocessor supports paging API to process requests in a stream-like way #11448
• Support read-through-lock so that read operations do not need to wait for secondary locks to be resolved #11402
• Add a disk protection mechanism to avoid panic caused by disk space drainage #10537
• Support archiving and rotating logs #11651
• Reduce the system call by the Raft client and increase CPU efficiency #11309
• Coprocessor supports pushing down substring to TiKV #11495
• Improve the scan performance by skip reading locks in the Read Committed isolation level #11485
• Reduce the default thread pool size used by backup operations and limit the use of thread pool when the stress is high #11000
• Support dynamically adjusting the sizes of the Apply thread pool and the Store thread pool #11159
• Support configuring the size of the snap-generator thread pool #11247
• Optimize the issue of global lock race that occurs when there are many files with frequent reads and writes #250

Bug fixes

• Fix the issue that the MVCC deletion records are not cleared by GC #11217
• Fix the issue that retrying prewrite requests in the pessimistic transaction mode might cause the risk of data inconsistency in rare cases #11187
• Fix the issue that GC scan causes memory overflow #11410
• Fix the issue that RocksDB flush or compaction causes panic when the disk capacity is full #11224

tikv-server v5.0.6

Improvements

• Increase the speed of inserting SST files by moving the verification process to the Import thread pool from the Apply thread pool #11239
• Add more metrics for the garbage collection module of Raft logs to locate performance problems in the module #11374
• Collapse some uncommon storage-related metrics in Grafana dashboard #11681

Bug fixes

• Fix the issue that a down TiKV node causes the resolved timestamp to lag #11351
• Fix the issue that TiKV cannot detect the memory lock when TiKV perform a reverse table scan #11440
• Fix the issue that the accumulation of GC tasks might cause TiKV to be OOM (out of memory) #11410
• Fix the issue of TiKV panic that occurs when the files do not exist when TiDB Lightning imports data #10438
• Fix the issue that the node of a TiKV replica is down after the node gets snapshots because TiKV cannot modify the metadata accurately #10225
• Fix the leak issue of the backup thread pool #10287
• Fix the issue of casting illegal strings into floating-point numbers #23322

tikv-server v4.0.16

Compatibility changes

• Before v4.0.16, when TiDB converts an illegal UTF-8 string to a Real type, an error is reported directly. Starting from v4.0.16, TiDB processes the conversion according to the legal UTF-8 prefix in the string #11466

Improvements

• Reduce disk space consumption by adopting the zstd algorithm to compress SST files when restoring data using Backup & Restore or importing data using Local-backend of TiDB Lightning #11469

Bug fixes

• Fix a panic issue that occurs when Region merge, ConfChange, and Snapshot happen at the same time in extreme conditions #11475
• Fix the issue of negative sign when the decimal divide result is zero #29586
• Fix the issue that the average latency of the by-instance gRPC requests is inaccurate in TiKV metrics #11299
• Fix the issue of TiCDC panic that occurs when the downstream database is missing #11123
• Fix the issue that the Raft connection is broken when the channel is full #11047
• Fix the issue that TiDB cannot correctly identify whether the Int64 types in Max/Min functions are a signed integer or not, which causes the wrong calculation result of Max/Min #10158
• Fix the issue that CDC adds scan retries frequently due to the Congest error #11082

tikv-server v5.2.3

Bug fix

• Fix the issue that the GcKeys task does not work when it is called by multiple keys. Caused by this issue, compaction filer GC might not drop the MVCC deletion information. #11217

tikv-server v5.0.5

Bug fix

• Fix the issue that the GcKeys task does not work when it is called by multiple keys. Caused by this issue, compaction filer GC might not drop the MVCC deletion information. #11217

tikv-server v5.1.3

Bug fix

• Fix the issue that the GcKeys task does not work when it is called by multiple keys. Caused by this issue, compaction filer GC might not drop the MVCC deletion information. #11217

tikv-server v5.3.0

Improvements

• Enhance disk space protection to improve storage stability
• Simplify the algorithm of L0 flow control #10879
• Improve the error log report in the raft client module #10944
• Improve logging threads to avoid them becoming a performance bottleneck #10841
• Add more statistics types of write queries #10507

Bug Fixes

• Fix the issue of unavailable TiKV caused by Raftstore deadlock when migrating Regions. The workaround is to disable the scheduling and restart the unavailable TiKV. #10909
• Fix the issue that CDC adds scan retries frequently due to the Congest error #11082
• Fix the issue that the Raft connection is broken when the channel is full #11047
• Fix the issue that batch messages are too large in Raft client implementation #9714
• Fix the issue that some coroutines leak in resolved_ts #10965
• Fix a panic issue that occurs to the coprocessor when the size of response exceeds 4 GiB #9012
• Fix the issue that snapshot Garbage Collection (GC) misses GC snapshot files when snapshot files cannot be garbage collected #10813
• Fix a panic issue caused by timeout when processing Coprocessor requests #10852
• Fix a memory leak caused by monitoring data of statistics threads #11195
• Fix a panic issue caused by getting the cgroup information from some platforms #10980

tikv-server v5.2.2

Improvements

• Simplify the algorithm of L0 flow control #10879
• Improve the error log report in raft client module #10983
• Make the slow log of TiKV coprocessor only consider the time spent on processing requests #10841
• Drop log instead of blocking threads when the slogger thread is overloaded and the queue is filled up #10841
• Add more statistics types of write queries #10507

Bug Fixes

• Fix the issue that CDC add scan retries frequently due to Congest error #11082
• Fix that the raft connection is broken when the channel is full #11047
• Fix the issue that batch messages are too large in Raft client implementation #9714
• Fix the issue that concurrent leaks in resolved_ts #10965
• Fix a panic issue that occurs to coprocessor when response size exceeds 4 GiB #9012
• Fix the issue that snapshot Garbage Collection (GC) misses GC snapshot files when snapshot files cannot be garbage collected #10813
• Fix a panic issue that occurs when processing coprocessor requests times out #10852