The UPDATE statement wrote NULL when the column set is DATETIME and
value is '0000-00-00 00:00:00'. The problem was inside WriteEngine's
handling of UPDATE statements and this is where heart of change is.
Other changes are related to some obsolete data structures in DML/DDL
handling that just hanging around there, doing nothing.
This patch improves handling of NULLs in textual fields in ColumnStore.
Previously empty strings were considered NULLs and it could be a problem
if data scheme allows for empty strings. It was also one of major
reasons of behavior difference between ColumnStore and other engines in
MariaDB family.
Also, this patch fixes some other bugs and incorrect behavior, for
example, incorrect comparison for "column <= ''" which evaluates to
constant True for all purposes before this patch.
This reverts commit f4e3022fbdecc25a22ae6eaf072e462aa6695f35.
The commit apparently caused MCOL-5318 and MCOL-5319 which involve the
internal ColumnStore batch insert mechanism passing through the SQL
layer. The code block involved in this change is a predicate checking
for the HWM extent in WriteEngineServer at the end of the batch insert.
This is done in WE_DMLCommandProc::processBatchInsertHwm(). The original
predicate check in this function for the HWM extent is restored until
further investigation.
In the implementation of MCOL-5021, an assert was added in
`WE_DMLCommandProc::processBatchInsertHwm()` that assumed the
`WriteEngine::TableMetaData` cache is uniform across the cluster.
However, this assumption is incorrect.
This bug caused undefined behaviour in ColumnStore resulting in bugs
such as MCOL-5367. In MCOL-5367, in a multi-node ColumnStore cluster,
an INSERT ... SELECT in a transaction with system variable
`columnstore_use_import_for_batchinsert=OFF/ON` did not show inserted
records when a SELECT query was issued. Assuming a 3-node cluster setup,
DMLProc only sends a given batch of records to be inserted to one of the
3 nodes, and not all nodes. As a result, the `WriteEngine::TableMetaData`
cache is only populated for that one node and is not uniform across the
cluster, causing the assert to fail.
As a fix, we simply remove this assert as it is redundant and should not
have been added in the first place.
DMLProc starts ROLLBACK when SELECT part of UPDATE fails b/c EM facility in PP were restarted.
Unfortunately this ROLLBACK stuck if EM/PP are not yet available.
DMLProc must have a t/o with re-try doing ROLLBACK.
The idea is relatively simple - encode prefixes of collated strings as
integers and use them to compute extents' ranges. Then we can eliminate
extents with strings.
The actual patch does have all the code there but miss one important
step: we do not keep collation index, we keep charset index. Because of
this, some of the tests in the bugfix suite fail and thus main
functionality is turned off.
The reason of this patch to be put into PR at all is that it contains
changes that made CHAR/VARCHAR columns unsigned. This change is needed in
vectorization work.
Part 1:
As part of MCOL-3776 to address synchronization issue while accessing
the fTimeZone member of the Func class, mutex locks were added to the
accessor and mutator methods. However, this slows down processing
of TIMESTAMP columns in PrimProc significantly as all threads across
all concurrently running queries would serialize on the mutex. This
is because PrimProc only has a single global object for the functor
class (class derived from Func in utils/funcexp/functor.h) for a given
function name. To fix this problem:
(1) We remove the fTimeZone as a member of the Func derived classes
(hence removing the mutexes) and instead use the fOperationType
member of the FunctionColumn class to propagate the timezone values
down to the individual functor processing functions such as
FunctionColumn::getStrVal(), FunctionColumn::getIntVal(), etc.
(2) To achieve (1), a timezone member is added to the
execplan::CalpontSystemCatalog::ColType class.
Part 2:
Several functors in the Funcexp code call dataconvert::gmtSecToMySQLTime()
and dataconvert::mySQLTimeToGmtSec() functions for conversion between seconds
since unix epoch and broken-down representation. These functions in turn call
the C library function localtime_r() which currently has a known bug of holding
a global lock via a call to __tz_convert. This significantly reduces performance
in multi-threaded applications where multiple threads concurrently call
localtime_r(). More details on the bug:
https://sourceware.org/bugzilla/show_bug.cgi?id=16145
This bug in localtime_r() caused processing of the Functors in PrimProc to
slowdown significantly since a query execution causes Functors code to be
processed in a multi-threaded manner.
As a fix, we remove the calls to localtime_r() from gmtSecToMySQLTime()
and mySQLTimeToGmtSec() by performing the timezone-to-offset conversion
(done in dataconvert::timeZoneToOffset()) during the execution plan
creation in the plugin. Note that localtime_r() is only called when the
time_zone system variable is set to "SYSTEM".
This fix also required changing the timezone type from a std::string to
a long across the system.
data types TEXT, CHAR, VARCHAR, FLOAT and DOUBLE are not yet supported by vectorized path
This patch introduces an example for Google benchmarking suite to measure a perf diff
b/w legacy scan/filtering code and the templated version
This patch:
1. Removes the option to declare uncompressed columns (set columnstore_compression_type = 0).
2. Ignores [COMMENT '[compression=0] option at table or column level (no error messages, just disregard).
3. Removes the option to set more than 2 extents per file (ExtentsPreSegmentFile).
4. Updates rebuildEM tool to support up to 10 dictionary extent per dictionary segment file.
5. Adds check for `DBRootStorageType` for rebuildEM tool.
6. Renamed rebuildEM to mcsRebuildEM.
Logs for research purposes
Keep progress - may not build
Good interface to collect LBIDs and CPInfo's
Write Engine compiles with new interface
New interface breaks things the least way and allows for new features to be added gradually.
Still ironing design - rewriting parts of WE
Keep progress commit
Write Engine compiles, going to test
I could introduce crashes there. Let's see.
Disable logging for tests
Fixing build problems - keep progress commit
Changed related to new interface
Add back accidentally removed m_txnLBIDMap.find
Remove printf/cout; up-to-date comment for AddLBIDtoList
Add "auto" type annotation
Work on PR comments
Descriptive vector emptines check
MCOL-4409 This patch combines VDecimal and Decimal and makes
IDB_Decimal an alias for the result class
MCOL-4409 More boilerplate reduction in Func_mod
Removed couple TSInt128::toType() methods
Removed uint128 from joblist/lbidlist.*
Another toString() method for wide-decimal that is EMPTY/NULL aware
Unified decimal processing in WF functions
Fixed a potential issue in EqualCompData::operator() for
wide-decimal processing
Fixed some signedness warnings
WF::percentile runtime threw an exception b/c of wrong DT deduced from its argument
Replaced literals with constants
Tought WF_sum_avg::checkSumLimit to use refs instead of values
2. Set Decimal precision in SimpleColumn::evaluate().
3. Add support for int128_t in ConstantColumn.
4. Set IDB_Decimal::s128Value in buildDecimalColumn().
5. Use width 16 as first if predicate for branching based on decimal width.
There are multiple overloaded version of the low level DML write methods to
push down CSC column type. WE needs the type to convert values correctly.
Replaced WE_INT128 with CSC data type that is more informative.
Removed commented and obsolete code.
Replaced switch-case blocks with oneliners.
