[teiid-issues] [JBoss JIRA] (TEIID-4294) External Materialization MATVIEW_AFTER_LOAD_SCRIPT is not atomic operation
Steven Hawkins (JIRA)
issues at jboss.org
Wed Aug 10 15:01:00 EDT 2016
[ https://issues.jboss.org/browse/TEIID-4294?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13277516#comment-13277516 ]
Steven Hawkins commented on TEIID-4294:
---------------------------------------
To summarize this -
For sources that do not treat ddl transactionally (for example non-XA), or have an isolation level that allows the affect to be seen immediately, then there can be an issue with the documented/default swap strategy such that the table temporarily does not exist.
Possible remedies:
1. With TEIID-4240 for sources that support it, it's possible to send a set of commands to the source rather than as individual native queries.
2. If on error action is wait, then we can temporarily mark the table as invalid during the after/swap - but that would only narrow the possibility of an invalid read and not prevent it.
3. Don't promote/rely on rename - add a column to the status table to indicate whether the primary or the staging table should be accessed (similar to how JDG staging logic functions) so that the swap is atomic. For this to work out-of-the box would require several logic and doc changes.
4. Promote an alternative strategy based upon maintaining just a single table - for sources that support merge this would be straightforward but can require adding an additional column to the materialized table for purging out of date entries. This can be just documented.
> External Materialization MATVIEW_AFTER_LOAD_SCRIPT is not atomic operation
> --------------------------------------------------------------------------
>
> Key: TEIID-4294
> URL: https://issues.jboss.org/browse/TEIID-4294
> Project: Teiid
> Issue Type: Bug
> Components: Query Engine
> Affects Versions: 8.12.5
> Reporter: Jan Stastny
> Assignee: Steven Hawkins
> Priority: Blocker
> Attachments: dv-2016-06-23-mat-view.log, teiid-command.log
>
>
> Firstly, this issue appears even after *TEIID-4283* fix, and is db independent.
> When during materialized view's loading there is a query on the view performed, timing issues appear. In MATVIEW_AFTER_LOAD_SCRIPT there might be more commands separated by semicolon, but these commands are not performed as a single operation. In specific timing a query on such view might fail when you have a MATVIEW_AFTER_LOAD_SCRIPT like this:
> {code:sql}
> "teiid_rel:MATVIEW_AFTER_LOAD_SCRIPT" 'exec Source.native(''RENAME TABLE dv_matviews_mat_view_stage TO dv_matviews_mat_view_temp'');exec Source.native(''RENAME TABLE dv_matviews_mat_view TO dv_matviews_mat_view_stage'');exec Source.native(''RENAME TABLE dv_matviews_mat_view_temp TO dv_matviews_mat_view'');
> {code}
> user might end up querying the view right in the moment, when 'primary' materialized table (dv_matviews_mat_view) was renamed (dv_matviews_mat_view_stage) and thus there's no such table like defined in the materialized view.
> {code:xml}
> CREATE VIEW external_long_ttl (
> customer_id integer NOT NULL,
> total_amount integer
> ) OPTIONS (MATERIALIZED 'TRUE', UPDATABLE 'FALSE',
> MATERIALIZED_TABLE 'Source.JSTASTNY.dv_matviews_mat_view',
> "teiid_rel:ALLOW_MATVIEW_MANAGEMENT" 'true',
> "teiid_rel:MATVIEW_STATUS_TABLE" 'Source.JSTASTNY.dv_matviews_statustable',
> "teiid_rel:ON_VDB_START_SCRIPT" 'MERGE INTO dv_matviews_check_table(id,vdb_create) SELECT id, vdb_create+1 FROM dv_matviews_check_table WHERE id=''external_long_ttl'';',
> "teiid_rel:ON_VDB_DROP_SCRIPT" 'MERGE INTO dv_matviews_check_table(id,vdb_drop) SELECT id, vdb_drop+1 FROM dv_matviews_check_table WHERE id=''external_long_ttl'';',
> "teiid_rel:MATVIEW_LOAD_SCRIPT" 'INSERT INTO dv_matviews_mat_view_stage(customer_id,total_amount) SELECT CONVERT(c.id,integer) AS customer_id, CONVERT(SUM(o.amount),integer) AS total_amount FROM dv_matviews_customers c INNER JOIN dv_matviews_orders o ON c.id = o.customer_id GROUP BY c.id;',
> "teiid_rel:MATVIEW_BEFORE_LOAD_SCRIPT" 'exec Source.native(''truncate table dv_matviews_mat_view_stage'');MERGE INTO dv_matviews_check_table(id,before_load) SELECT id, before_load+1 FROM dv_matviews_check_table WHERE id=''external_long_ttl'';',
> "teiid_rel:MATVIEW_AFTER_LOAD_SCRIPT" 'exec Source.native(''RENAME TABLE dv_matviews_mat_view_stage TO dv_matviews_mat_view_temp'');exec Source.native(''RENAME TABLE dv_matviews_mat_view TO dv_matviews_mat_view_stage'');exec Source.native(''RENAME TABLE dv_matviews_mat_view_temp TO dv_matviews_mat_view'');MERGE INTO dv_matviews_check_table(id,after_load) SELECT id, after_load+1 FROM dv_matviews_check_table WHERE id=''external_long_ttl'';',
> "teiid_rel:MATVIEW_ONERROR_ACTION" 'IGNORE',
> "teiid_rel:MATVIEW_TTL" 20000)
> AS SELECT CONVERT(c.id,integer) AS customer_id, CONVERT(SUM(o.amount),integer) AS total_amount FROM dv_matviews_customers c INNER JOIN dv_matviews_orders o ON c.id = o.customer_id GROUP BY c.id;
> {code}
> Then for some queries I get:
> {code:plain}
> 10:40:19,252 WARN [org.teiid.CONNECTOR] (Worker2_QueryProcessorQueue130) Connector worker process failed for atomic-request=+k/MyS3sUhQX.14.7.113: org.teiid.translator.jdbc.JDBCExecutionException: 259 TEIID11008:TEIID11004 Error executing statement(s): [Prepared Values: [] SQL: SELECT g_0."CUSTOMER_ID", g_0."TOTAL_AMOUNT" FROM "JSTASTNY".dvqe_X_XjdkY_YDV_MATVIEWS_MAT_VIEW AS g_0]
> at org.teiid.translator.jdbc.JDBCQueryExecution.execute(JDBCQueryExecution.java:131) [translator-jdbc-8.12.5.redhat-5.jar:8.12.5.redhat-5]
> at org.teiid.dqp.internal.datamgr.ConnectorWorkItem.execute(ConnectorWorkItem.java:364)
> at sun.reflect.GeneratedMethodAccessor158.invoke(Unknown Source) [:1.7.0_79]
> at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43) [rt.jar:1.7.0_79]
> at java.lang.reflect.Method.invoke(Method.java:606) [rt.jar:1.7.0_79]
> at org.teiid.dqp.internal.datamgr.ConnectorManager$1.invoke(ConnectorManager.java:211)
> at com.sun.proxy.$Proxy80.execute(Unknown Source)
> at org.teiid.dqp.internal.process.DataTierTupleSource.getResults(DataTierTupleSource.java:306)
> at org.teiid.dqp.internal.process.DataTierTupleSource$1.call(DataTierTupleSource.java:112)
> at org.teiid.dqp.internal.process.DataTierTupleSource$1.call(DataTierTupleSource.java:108)
> at java.util.concurrent.FutureTask.run(FutureTask.java:262) [rt.jar:1.7.0_79]
> at org.teiid.dqp.internal.process.FutureWork.run(FutureWork.java:65)
> at org.teiid.dqp.internal.process.DQPWorkContext.runInContext(DQPWorkContext.java:276)
> at org.teiid.dqp.internal.process.ThreadReuseExecutor$RunnableWrapper.run(ThreadReuseExecutor.java:119)
> at org.teiid.dqp.internal.process.ThreadReuseExecutor$3.run(ThreadReuseExecutor.java:210)
> at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1145) [rt.jar:1.7.0_79]
> at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:615) [rt.jar:1.7.0_79]
> at java.lang.Thread.run(Thread.java:745) [rt.jar:1.7.0_79]
> Caused by: com.sap.db.jdbc.exceptions.JDBCDriverException: SAP DBTech JDBC: [259]: invalid table name: Could not find table/view DVQE_X_XJDKY_YDV_MATVIEWS_MAT_VIEW in schema JSTASTNY: line 1 col 62 (at pos 61)
> at com.sap.db.jdbc.exceptions.SQLExceptionSapDB.createException(SQLExceptionSapDB.java:345)
> at com.sap.db.jdbc.exceptions.SQLExceptionSapDB.generateDatabaseException(SQLExceptionSapDB.java:185)
> at com.sap.db.jdbc.packet.ReplyPacket.buildExceptionChain(ReplyPacket.java:100)
> at com.sap.db.jdbc.ConnectionSapDB.execute(ConnectionSapDB.java:1130)
> at com.sap.db.jdbc.ConnectionSapDB.execute(ConnectionSapDB.java:877)
> at com.sap.db.jdbc.CallableStatementSapDB.execute(CallableStatementSapDB.java:410)
> at com.sap.db.jdbc.CallableStatementSapDB.execute(CallableStatementSapDB.java:266)
> at com.sap.db.jdbc.CallableStatementSapDB.executeQuery(CallableStatementSapDB.java:756)
> at com.sap.db.jdbc.trace.PreparedStatement.executeQuery(PreparedStatement.java:161)
> at org.jboss.jca.adapters.jdbc.WrappedPreparedStatement.executeQuery(WrappedPreparedStatement.java:462)
> at org.teiid.translator.jdbc.JDBCQueryExecution.execute(JDBCQueryExecution.java:123) [translator-jdbc-8.12.5.redhat-5.jar:8.12.5.redhat-5]
> ... 17 more
> {code}
> From the logs it seems that the after load script has completed, but probably not in the source db. In logs just before the exception I see:
> {code:plain}
> 12:34:59,247 DEBUG [org.teiid.CONNECTOR] (Worker1_QueryProcessorQueue79) Source-specific command: RENAME TABLE dv_matviews_mat_view_temp TO dv_matviews_mat_view
> 12:34:59,663 DEBUG [org.teiid.CONNECTOR] (Worker1_QueryProcessorQueue79) ORTEkGreciAL.0.85.84 Executed command
> 12:34:59,663 DEBUG [org.teiid.CONNECTOR] (Worker1_QueryProcessorQueue79) ORTEkGreciAL.0.85.84 Processing MORE request
> 12:34:59,663 DEBUG [org.teiid.CONNECTOR] (Worker1_QueryProcessorQueue79) ORTEkGreciAL.0.85.84 Getting results from connector
> 12:34:59,663 DEBUG [org.teiid.CONNECTOR] (Worker1_QueryProcessorQueue79) ORTEkGreciAL.0.85.84 Obtained last batch, total row count: 1
> 12:34:59,663 DEBUG [org.teiid.CONNECTOR] (Worker1_QueryProcessorQueue79) ORTEkGreciAL.0.85.84 Remove State
> 12:34:59,664 DEBUG [org.teiid.CONNECTOR] (Worker1_QueryProcessorQueue79) ORTEkGreciAL.0.85.84 Processing Close : EXEC Source.native('RENAME TABLE dv_matviews_mat_view_temp TO dv_matviews_mat_view')
> {code}
> I attached the logs to this issue.
--
This message was sent by Atlassian JIRA
(v6.4.11#64026)
More information about the teiid-issues
mailing list