Weekly update in TiDB

Last week, we landed 39 PRs in the TiDB repository.

Added

• Add different labels for restricted SQL and general SQL metrics
• Add the USR1 signal handler to dump goroutine
• Use the point get plan for the UPDATE statement
• Support load data with IgnoreLines
• Support the json_contains builtin function

Improved

• Make Analyze Buckets number configurable
• Set HIGH_PRIORITY for the bootstrap SQL statements
• Only check the range typed partition when creating a partitioned table
• Make a tiny refactor in the reorganization stage of adding indices
• Relax the tso backoff limit
• Reduce chunk’s iterator function call
• Improve compatibility for the MariaDB client
• Do AutoAnalyze on a certain period of a day
• Change the logic of converting the logical join to the index join
• Read the inner table and build hash table parallel in hash join
• Add batch copy to the inner join and the left and right outer join
• Add a ctxPool field to the worker struct to make executing the SQL statement in the DDL package possible

Fixed

• Fix some datetime related cases which are inconsistent with MySQL
• Fix a compatibility problem of analyzing period variables
• Fix an error in the parser when parsing a single line comment ended with a newline character
• Fix an issue that the gc_delete_range table is queried with a wrong form of timestamp
• Fix an issue that the bit type can use null as its default value
• Return the correct column name and column label name
• Fix a panic when logging detailed statistics
• Fix a wrong count output of explain for the TableScan plan
• Fix the update join result when the join table order is changed
• Fix the issue that creating a partitioned table with bigint columns fails

Weekly update in TiSpark

Last week, we released a new version 1.1 and landed 10 PRs in the TiSpark repository.

Fixed

• Remove the experimental split function and related parameters
• Fix the SharedState behavior in ThriftServer
• Fix the issue that start-tithriftserver.sh loses extra options

Weekly update in TiKV and PD

Last week, we landed 50 PRs in the TiKV and PD repositories.

Added

• Add new built-in functions:

  • builtin-Greatest*/Least*
  • builtin-BitCount
  • builtin-LTrim/RTrim
  • builtin-math.Sin
  • builtin-Rand/RandWithSeed
  • builtin-Reverse/ReverseBinary)
  • builtin-CharLength
  • builtin-HexIntArg/HexStrArg
  • builtin-Asin/Acos)
  • builtin-Inet6Ntoa
  • builtin-MD5
  • builtin-Cot/Degrees
  • builtin-Elt
  • builtin-LastDay
  • builtin-Month

Fixed

• Always update read delegate’s Regions to avoid stale Region information in LocalReader
• Remove read delegate on conf change RemoveNode
• Do not drop MsgRequestPreVote messages from newly split Regions

Improved

• Reduce key clone in GC
• Limit the garbage cleanup speed to avoid blocking snapshot application
• Remove extra Region clone when handling Region heartbeats
• Clean up residual Region clone
• Start schedulers based on the proportion of Regions

New contributors (Thanks!)

TiKV:

• WPH95
• chux0519
• intellild
• liufuyang
• malc0lm
• mtunique

TiDB:

• llvim
• xiangyuf