Skip to content

Releases: taosdata/TDengine

3.0.2.3

29 Jan 08:33
4ea7ed5
Compare
Choose a tag to compare

Improvements:

  • pseudo column "_isfilled" can be used with interp() functon to indicate whether the value is original value or not
  • Performance optimization for Windows
  • Error is returned when trying to delete the last tag

Fixed Bugs:

  • taosd crash after stream processing has been running for long time
  • "invalid value in client" is returned when querying the result stable of stream processing
  • taosd crash when "create db" & "drop db" are executed frequently
  • table UID is generated incorrectly in schemaless writing
  • taosd crashed because of race condidtion in LRU Cache
  • WAL file is deleted wrongly after tasod is restarted
  • "create mnode" hangs after creating stream
  • Aggregate query result is wrong because of SMA reading logic error
  • taosd crashed when quering in case of 3 replicas
  • taosbenchmark -y failed to execute

What's Changed

Read more

3.0.2.2

29 Jan 08:20
c09602b
Compare
Choose a tag to compare

Improvements:

  • dnode exit automatically after it is dropped by "drop dnode" command

Fixed Bugs:

  • The result of "Show table distrubted" is not correct
  • Unexpected NULL is returned when querying "partition by tag" with "interval" and "fill" clause
  • Unexpected NULL is returned when querying "percentile" with "interval"
  • Query failed after creating stream
  • taosd crashed when starting on CentOS 7
  • Duplicate values are returned when using nested query with group by
  • taosd crashed in "taosGetCPUCores"
  • taosd crashed when starting
  • show tables/stables failed after granting write privilege
  • Improper return code is returned by taosAdapter
  • taosd crashed when doing "select count(*)"
  • taosd failed to start with "illegal instruction" on CentOS 7
  • taosBenchmark -I stmt failed on Windows 7
  • taosd crashed on ARM64
  • The result of stream processing is not correct
  • taosBenchmark failed when using "-h " to write to TDengine

What's Changed

Full Changelog: ver-3.0.2.1...ver-3.0.2.2

2.6.0.34

23 Dec 07:18
Compare
Choose a tag to compare

Release Notes

from:release/ver-2.6.0.32
to:release/ver-2.6.0.34

FIX BUGS:
fix: change tsMaxTmrCtrlto 20480 (#19100)
fix:max_row constant (#18898)
fix: fixed not last word match bug (#19001)
fix: if all vnode in one group can not connect, return TSDB_CODE_RPC_VGROUP_NOT_REACHED(#18935)

ENHANCEMENTS:
enh: enlarge max wal size from 3 MB to 4 MB (#18886)
enh: support loongArch64

3.0.2.1

26 Dec 11:13
9cf70d8
Compare
Choose a tag to compare

Enhancements:

  • Support "rest" insert mode in the global part of JSON file for taosBenchmark
  • Topic access permission can be controlled specifically
  • Improved logging in case taosd is killed
  • Cleanup connection cache automatically after client is closed for a while
  • Query performance optimization

Fixed Bugs:

  • Output of taosbenchmark is wrong on Windows
  • Cluster state is OK but dnode is shown as offline in log
  • Output of "show vgroups" is incorrect
  • OS type reported from TDengine on macOS is empty
  • Command "alter database replica 3" took too long time
  • Data loss after "alter database replica 3"
  • taos crash in tsdb
  • Resulting table of stream processing can't be queried
  • Memory leak of taosX
  • taos shell "auto tab" can't match "show table distributed"
  • Some taosd crash problems

What's Changed

New Contributors

Full Changelog: ver-3.0.2.0...ver-3.0.2.1

3.0.2.0

12 Dec 01:53
37f4d4f
Compare
Choose a tag to compare

New Features:

  • Replica 3 can be specified when creating a database
  • RAFT protocol is enabled automatically in case of replica 3
  • Replica 1 and 3 can be switched by "alter database" command

Enhancements:

  • Performance optimization for inserting data in SQL mode
  • Performance optimization for replica 3
  • Performance optimization for querying data in a lot of scenarios
  • Stability improvement for replica 3
  • Execute "alter database replica" in async way

Fixed Bugs:

  • Some coredump problems
  • Some OOM issues
  • A series of functional bugs

What's Changed

Read more

2.6.0.32

04 Dec 11:49
Compare
Choose a tag to compare

Release Notes

from:release/ver-2.6.0.30
to:release/ver-2.6.0.32

FIX BUGS:
fix:in some cases, metadata may be received twice (#18609)
feat: support show cluster status command on 2.6 (#18578)
fix: add blank after affected words (#18446)
fix: show result format (#17949)
fix: add help command and fixed fields auto fill (#18433)
fix: core dump due to assert reallocation result when reallocate fails (#18391)
fix: loss of filter (#18317)
fix: variant offset save with int16_t is smaller (#18202)
fix: set connect socket to unblocking call with timeout (#18066)

3.0.1.8

25 Nov 06:41
Compare
Choose a tag to compare

Enhancements:

  • Performance optimization for "select *"
  • Improved error handling if the data to be inserted doesn't include timestamp
  • Performance optimization when there is "IN" clause in both inner and outter query
  • Performance optimization when counting disordered data of NCHAR type
  • Optimization for interp()

Fixed Bugs:

  • Crash caused by querying last_row() on a table of 4,000 columns
  • Crash caused by "select count(1)"
  • Crash caused by tag length exceeding the schema definition
  • Crash casued by commiting NONE or NULL to disk
  • Crash caused by fill(NULL) + interval + partition by tbname
  • Crash caused by fill_history + partition by tbname + state_window/session
  • Crash caused by stream processing in crash_gen
  • Crash at executor/src/sortoperator.c
  • Crash at libs/executor/src/tfill.c
  • Crash at executor/src/sortoperator.c
  • Crash happened in stability testing with replica=3
  • Crash caused by creating stream on child table with fill_history and data filtering
  • Overflow when converting time decision
  • SIGABORT received if the data to be inserted doesn't incldue timestamp
  • "Use DB" got error: Database is in dropping status
  • fill(value, '1') should NOT succeed for a column of integer type
  • "Invalid read of size 1" reported by Valgrind
  • Duplicate child table name is generating when creating child table automatically
  • Incorrect result returned by fill_history + partition by tbname + interval
  • Database should not be created if its name exceeds the upper limit
  • Other dnodes go to offline if one node is down and restarted
  • User should not be created if the password length exceeds upper limit
  • A lot of issues found by Address Sanitier
  • Bugs found by 3rd party database verification tool

What's Changed

Read more

3.0.1.7

18 Nov 06:06
Compare
Choose a tag to compare

Enhancements:

  • Future timestamp can be inserted
  • Performance optimization for query in a lot of scenarios
  • Performance optimization for data inserting
  • Significantly reduced the number of open file descriptions
  • Schemalss inserting can handle data including "\0"
  • Mode() function can be used with specific columns

Fixed Bugs:

  • taosd crashed when there is out of order data
  • taosd crashed when querying last_row()
  • taosd crash caused by creating stream
  • taosd crash caused by "CASE WHEN THEN NULL"
  • Client programs crashed because of taosc
  • Some other crashed in taosd
  • Memory leak issues in taosd
  • duplicate timestamp returned when querying with interval() fill(prev)
  • NULL retruned when querying with interval() fill(prev)
  • show create table \G doesn't show the full column list
  • FQDN over 32 bytes can't be handled properly
  • result returned by "last_row() group by tbname" is not consistent with that returned by "last group by tag"
  • child tables are not accumulated in the count of time series
  • show tables or select from ins_tables doesn't work after the license expires
  • The timezone specified when establishing connection doesn't work

What's Changed

Read more

2.6.0.30

11 Nov 13:54
Compare
Choose a tag to compare

Release Notes

from:release/ver-2.6.0.28
to:release/ver-2.6.0.30

FIX BUGS:
fix: fixed three problems about corrupted wal would be repaired (#17904)
fix: taosbenchmark query concurrent segfault for 2.6 (#17796)
fix(taosAdapter): fix cors config (#910)
fix: last function optimize error (#17747)

ENHANCEMENTS:
fix: show result format (#17949)
fix: taosbenchmark query interval unit (#17879)
feat: taosbenchmark support query fails then continue for 2.6 (#17981)
feat(schemaless): support escape character back-slash tab and back-slash return (#17776)

3.0.1.6

07 Nov 00:32
Compare
Choose a tag to compare

Enhancements:

  • TSMA can handle historical and out-of-order data
  • Performance optimization in a lot of querying scenarios on a super table: last_row(), last(), cound(*), avg(), max(), min(), limit
  • Security improvement by RPC integrity check
  • Optimized error message when client version doesn't match with the server
  • Performance improvement for select * from ins_tables
  • "show dnode variables" can be used with like clause
  • "show cluster variables" can show the cluster level configurations
  • "show vgroup" added a new state: restore
  • Table length is checked when adding a new column
  • Enhanced the atomicity of vnode commit
  • Improved stability

Fixed OOM & Crash:

  • OOM in case of "select *" from a super table
  • taosd crash causedby interp()
  • taosd hangs in cloud service
  • taosd OOM after long time data writing
  • OOM caused by taosc in taosBenchmark and taosAdapter
  • Some crashes occurred in case of replica=3
  • mnode failed to restart because of corrupted metadata
  • taosd crash caused by not deleting the index of deleted tables
  • taosd crash caused by last_row()
  • taosd crash caused by querying with group by and where condition
  • taosd crash caused by querying data from deleted tables
  • taosd crash caused by "show create table" on a table of 4,000 columns
  • taosd crash caused by count(*)
  • taos crash caused by broken network
  • taos crash caused by completing the command automatically
  • Some other crash problems

Fixed functional bugs:

  • last() doesn't return result even though there is data
  • Pseudo column _irowts returns zero timestamp
  • The number of rows returned by count(*) is larger than actual data size by one
  • "show create database" doesn't show cachesize parameter
  • Wrong result returned by apercentile()
  • Reesults are different when querying with "group by" and "where" on same tag
  • Incorrect sequence returned by order by
  • Two mnodes are in "candidate" and one is in "leader"