Weekly update in TiDB

Last week, we landed 67 PRs in the TiDB repositories.

Added

• CreateIndex supports the LOCK option.
• Add GoVersion info for tidb_version
• Add a session variable to show the configuration
• Support show stats_healthy to check if a table needs to be analyzed

Removed

• Remove the useless field in jsonColumn
• Clean up the abandoned storage engine

Fixed

• Check the CreateTable statement charset option
• Fix the bug of show index printing non-public index when add index operation is not finished
• Treat a decimal truncate as a warning in update

Improved

• Run GC workers parallelly
• Pass the operator label from Plan to Executor
• Prepare the candidate index to improve performance
• Use pseudo estimation when the stats of some table is outdated
• Ignore the error and keep GC always working
• Set a min count for AutoAnalyze to avoid the auto analysis of small tables
• Improve importer tools:
  • Support randDate by statistics
  • Support generating other types of columns randomly by statistics
  • Generate a string by statistics
  • Support VARCHAR in set
  • Generate the integer data by Histogram
• Refine metrics in TiDB:
  • Rename _count to _num
  • Unify metrics naming
  • Add a metric for pseudo estimation
  • Make metrics content clearer and compacter
  • Move domain metrics and add the privilege load counter
  • Add metrics for DDL and the server
  • Add metrics for the DDL worker
  • Add metrics for expensive executors and statement nodes
  • Add metrics for stats
  • Fix inconsistent labels for the panic counter
  • Move DDL metrics from the ddl package to the metrics package
  • Refine TiKV client metrics
  • Add metrics for the DDL owner
  • Add metrics and logs for ticlient

Weekly update in TiSpark

Last week, we landed 6 PRs in the TiSpark repositories.

Improved

• Change the remote repository for the test data
• Add JDBC write guide usage

Fixed

• Fix key range corner cases
• Fix the bug of group by without aggregates
• Fix the resetFilters method
• Fix the character decoding charset

Weekly update in TiKV and PD

Last week, we landed 19 PRs in the TiKV and PD repositories.

Added

• Add the read pool
• Introduce the label property and the label scheduler
• Support configuring the label scheduler and the label property

Fixed

• Fix the bug caused by limit=0 or by no limit

Improved

• Add the file name and the line number for the box error
• Record the leader missing information
• Filter Store when counting the average leader and score
• Add stale_command in ASYNC_REQUESTS_COUNTER_VEC
• Check whether range.start < range.end
• Change the unit of heartbeat from ts to ms
• Limit the maximum total size of snapshot
• Collect gRPC logs
• Save the critical metadata in metrics
• Use local metrics
• Adjust the status of the store

New contributor (Thanks!)

• TiDB: caojiafeng