HBaseClient_JNI.cpp

Clone Tools
Constraints
Constraints: committers
 
Constraints: files
Constraints: dates
[TRAFODION-2654] Change the location of trafodion-site.xml from $TRAF_HOME/etc to config [TRAFODION-2653] Sort operator loops at times

All HBase client connections from Trafodion uses Trafodion configuration

concept. Any client side property can now be set in trafodion-site.xml.

The scratch files are now open in blocking mode. Hence any scratch file

operation shouldn't return EAGAIN.

The internal table TDDL of table is now created without any table

coprocessors attached to it.

  1. … 23 more files in changeset.
[TRAFODION-2596] Improve the log4j and log4cxx infrastructure in Trafodion

The following changes are made in the way Trafodion logs the messages:

Writes to a log file based on a component or set of components.

C++ part of the code base:

Component Default log file Configuration File

TM tm_<nid>.log log4cxx.trafodion.tm.config

SSCP sscp_<nid>.log log4cxx.trafodion.sscp.config

SSMP ssmp_<nid>.log log4cxx.trafodion.ssmp.config

All SQL processes trafodion.sql_<nid>.log log4cxx.trafodion.sql.config

mxosrvr

sqlci

tdm_arkesp

tdm_arkcmp

tdm_udrserv

Java part of the code base

TM trafodion.dtm.log log4j.dtm.config

SQL trafodion.sql.java.log log4j.sql.config

By default, the log level is set to INFO for most of the cases. When a message

dominates the log file and if doesn't add value in the current level, it will be

changed to the higher level in the hierarchy. This should help to improve the

usefulness of the log file at the default INFO level.

The existence of an environment variable TRAF_MULTIPLE_SQL_LOG_FILE will revert back

to the old way of logging into multiple files. Then, the configuration file

log4cxx.trafodion.masterexe.config will be used.

Currently, RollingFileAppender appender is used for both modes. Different config files

are used to change this appender when the need arises.

log4cxx.trafodion.udr.config and log4cxx.trafodion.lob.config are removed.

log4j.hdfs.config used as the config file for SQL is renamed to log4j.sql.config

Foundation components logging will be revamped later.

This closes #1070

  1. … 19 more files in changeset.
Rework based on Selva's comments

[TRAFODION-2455] More refinements to row count estimation retry logic

  1. … 7 more files in changeset.
[TRAFODION-2440] Add retry to row count estimation logic

  1. … 16 more files in changeset.
TRAFODION-1586 Add support for external Traf tables mappedd to HBase tables

This is the initial support for HBase mapped tables.

A doc for external manuals will be added later.

regress/seabase/TEST022 contains various testcases.

  1. … 77 more files in changeset.
[Trafodion 2351] HBase snapshot concept needs to be contained within HBaseClient in Trafodion

Moved snapshot creation/deletion from SequenceFileWriter to HBaseClient.

  1. … 14 more files in changeset.
[TRAFODION-2356] HBase snapshot concept needs to be contained within HBaseClient in Trafodion

Moved snapshot creation/deletion from SequenceFileWriter to HBaseClient.

  1. … 13 more files in changeset.
[TRAFODION-2356] Trafodion process can dump core at times because JNIEnv is not initialized

  1. … 2 more files in changeset.
jira TRAFIDION-1435 SQL support for region transactions

This checkin is to add support for sql compiler/executor support

for dtm region transactions.

cqd traf_use_region_xn 'ON' enables it.

Off by default for now until dtm support is complete and tested.

Explain (full and short) will also show if region transactions

are being used.

core/test116 has been enhanced with additional tests

  1. … 19 more files in changeset.
[TRAFODION-2129] Trafodion to avoid use of deprecated HBase APIs/Classes

Avoided the caching of Hbase table instances on the client side to avoid

compaction queue size build up while constantly loading the data

  1. … 11 more files in changeset.
[JIRA TRAFODION-2141] Cluster view of hbase stats

-- added column REGION_SERVER to region stats virtual table

-- added cluster view of hbase stats

Syntax: select * from table(cluster stats());

-- added support for 'where' pred in stats queries

Syntax: select * from table(cluster stats()) where schema_name = 'SCH';

  1. … 22 more files in changeset.
[TRAFODION-2129] Trafodion to avoid use of deprecated HBase APIs/Classes

References to the deprecated classes HConnection, HConnectionManager and

HBaseAdmin are reworked to use Connection, ConnectionFactory and Admin classes

respectively.

A connection object is created and it is passed around to ensure that the

same connection object is used within the process. Connection object is thread safe.

Admin object is created and closed after its use immediately.

Creation of admin object is lightweight and the object is not thread safe

Also fixed the random hang at the time of rollback transaction waiting for the response from the TM process.

The TM process hung because it assumed that RS didn't respond back when there is an exception in DDL

rollback or end-point coprocessor call.

  1. … 28 more files in changeset.
[TRAFODION-1955] JNI optimization at the time of compilation Additional changes and bug fix Removed ByteArrayList JNI class

  1. … 2 more files in changeset.
[TRAFODION-1955] JNI optimization for the methods used at the time of query compilation Changes as per review comments. Added back ByteArrayList JNI but without any references to it to ensure other branches are not broken. Contributors are advised to modify their code to avoid use of this class.

  1. … 5 more files in changeset.
[TRAFODION-1955] JNI optimization at the time of compilation

Removed ByteArrayList class and used array of byte array instead.

Only one JNI to java transistion is done to getRegionStartKeys

and getRegionEndKeys at the time of query compilation

Also fixed an issue with lib_mgmt Makefile.

  1. … 15 more files in changeset.
[TRAFODION-1421] Implement parallel scanner primitive new CQD: hbase_dop_parallel_scanner - default 0.0 = disabled - from 0.x to 1.0 -> the DOP will be dynamically assessed as a percentage of region to scan. If 10 region, and cqd is .3 (30%), then 3 threads will be used... - from 2.0 and above, will take ceiling, and use it as fixed number of threads. Advantage of parallel scanner parallelism over ESP is on resource consumption. Parallel scanner is just creating threads, while ESP are full proceses wth high memory consumption. For now, this feature is just a primitive that can be manually exercised. The next obvious step is to make the compiler aware of it and pick it over ESP DOP when appropriate...

  1. … 21 more files in changeset.
[TRAFODION-1918] Remove auto flush concept in Trafodion

The auto flush is on by default in insertRows method. But insertRows was

incorrectly called with the wrong parameter. In fixing this issue, it was

observed that auto flush is deprecated in HBase 1.x. It is also incorrect

to disable auto flush in Trafodion. Hence removed the code related to

auto flush, flush table and flush all tables in Trafodion.

All these are dormant code in Trafodion.

Also disabled executor/TEST106 for now because it seems to cause a hang

at the time of drop table.

  1. … 13 more files in changeset.
Merge remote branch 'origin/master' into deliver2_dt_br

-- resolved merge conflicts

Conflicts:

core/sql/executor/HBaseClient_JNI.h

core/sql/regress/executor/EXPECTED013.SB

core/sql/regress/seabase/EXPECTED011

core/sql/sqlcomp/CmpSeabaseDDL.h

core/sql/sqlcomp/CmpSeabaseDDLcommon.cpp

core/sql/sqlcomp/CmpSeabaseDDLtable.cpp

  1. … 18 more files in changeset.
JIRA TRAFODION-1798 (ddl xns) and few other fixes, details below.

-- support for sql part of ddl xns. Section 1 of JIRA TRAFODION-1798

-- cqd ddl_transactions to enable or disable ddl xns.

Default is currently off. Once it is tested, it will be turned on.

Dev regressions are run with cqd set to ON

-- get stmts run with read committed to get changes in current xns

-- support for where preds with get stmts

-- scan to pass in transid even if running with read uncommitted access.

This enables rows modified in current xn to be returned.

-- cleanup no longer return multiple duplicate error messages if

objects id is not found.

-- cleanup no longer includes internallay created schemas (_HV_ , _HB_)

during cleanup operations.

-- Correct error msg was not getting returned if an invalid index

existed in table and the same index was created again.

-- init traf, drop md views was giving an error if views didnt exist.

That has been fixed.

-- regressions with -diff option now show original file timestamps

instead of the timestamp when the diff command was run.

  1. … 72 more files in changeset.
JIRA 1844: support for 'alter table alter column datatype'

  1. … 26 more files in changeset.
Implement TRAFODION-1420 Use ClientSmallScanner for small scans to improve perdormance Hbase implements an optimization for small scan (defined as scanning less than a data block ie 64Kb) resulting in 3X performance improvement. The underlying trick is about cutting down RPC calls from 3 (OpenScan/Next/Close) to 1, and use pread stateless instead of seek/read state-full and locking method to read data. This JIRA is about improving the compiler who can be aware if a scan will be acting on single data block (small) or not, and pass this data to executor so that it can use the right parameter for scan. (scan.setSmall(boolean)). reference: https://issues.apache.org/jira/browse/HBASE-9488 https://issues.apache.org/jira/browse/HBASE-7266

  1. … 27 more files in changeset.
Merge [TRAFODION-1655] PR 205 RMS iotime and OperCpuTime fix

[TRAFODION-1655] RMS reports incorrect hbase io time and 0 for OperCpuTime for operators hosted in ESP Isolate the timer for Operator Cpu Time from HBase IO Timer. Statistics enabling flag was not set though statistics area was created and some of the metrics were collected. Operator Cpu time metrics collection requires statistics to be enabled. It is now enabled when the operators are hosted in ESPs.

  1. … 4 more files in changeset.
JIRA Trafodion-1645 SQL commands to retrieve region stats.

Details in JIRA.

  1. … 23 more files in changeset.
Another set of changes to take care reviewer comments

[TRAFODION-1633] Fix for HBaseCient.getHbaseTableInfo throwing fpe

Fix for the failure in seabase/TEST027 with [TRAFODION-1423]

Disabled creating aligned format index with multi-column family.

  1. … 2 more files in changeset.
Most of the Trafodion Java source files are built through Maven, using projects DCS, REST, HBase-trx and SQL. A few files remain in the core/sql/executor and core/sql/ustat directories that are built through javac commands in the Makefile and helper files. Also, these files don't conform to common naming standards and therefore are a bit harder to debug with tools like Eclipse.

I would like to do the following:

1. Move these remaining Java source files under core/sql/src,

where the UDR-related files already are.

2. Include the class files in trafodion-sql-1.2.0.jar, the

deliverable of the core/sql Maven project.

3. Change the package name to org.trafodion.sql for all these files.

4. Remove the trafodion-HBaseAccess-1.2.0.jar file (content

will be included in trafodion-sql-1.2.0.jar). This jar file

used to be built with a Makefile.

5. Make some source code changes to reflect the new package names

and to avoid calling private method of class

org.apache.hadoop.hive.ql.io.orc.OrcStruct.

6. Remove Makefile rules for compiling Java files.

Also, this change includes a change of the component name for the

event logger. We use SQL.COMP now, to make analyzing log messages

for Trafodion easier. Venkat asked for this change.

  1. … 44 more files in changeset.
[TRAFODION-9] Non-blocking hbase operation to smoothen the data flow in trafodion engine

The non-blocking hbase operation all index maintenance operators is now enabled by

default for IUD, merge and update commands.

The hbase operations put, checkAndPut and delete operations from the trafodion engine

for the index maintenance is done using Executor Services to make it non-blocking and the

data flow in the trafodion operators can continue up to the point till it needs to pause for completion.

checkAndDelete is never used for index maintenance now.

Also, ensured that index maintenance uses rowset or single row operators when it should.

  1. … 13 more files in changeset.
[TRAFODION-1448] Leak in HTableClient objects

Refactored the code to ensure that HTableClient object in java

is released for every batch of gets.

In the process removed unused code in HTableClient_JNI.cpp and

ExpHbaseInterface.cpp

  1. … 6 more files in changeset.