[
https://issues.apache.org/jira/browse/PHOENIX-6083?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17216953#comment-17216953
]
ASF GitHub Bot commented on PHOENIX-6083:
-----------------------------------------
ChinmaySKulkarni commented on a change in pull request #929:
URL: https://github.com/apache/phoenix/pull/929#discussion_r507923226
##########
File path:
phoenix-core/src/it/java/org/apache/phoenix/end2end/index/ViewIndexIT.java
##########
@@ -169,10 +171,67 @@ public void testDeleteViewIndexSequences() throws
Exception {
verifySequenceValue(null, sequenceName, sequenceSchemaName,
Short.MIN_VALUE + 2);
conn1.createStatement().execute("DROP VIEW " + fullViewName);
conn1.createStatement().execute("DROP TABLE "+ fullTableName);
-
+
verifySequenceNotExists(null, sequenceName, sequenceSchemaName);
}
-
+
+ @Test
+ public void testDroppingColumnWhileCreatingIndex2() throws Exception {
+ String schemaName = "S1";
+ String tableName = generateUniqueName();
+ String viewSchemaName = "S1";
+ String fullTableName = SchemaUtil.getTableName(schemaName, tableName);
+ String indexName = "IND_" + generateUniqueName();
+ String viewName = "VIEW_" + generateUniqueName();
+ String fullViewName = SchemaUtil.getTableName(viewSchemaName,
viewName);
+
+ createBaseTable(schemaName, tableName, false, null, null, true);
+ Connection conn = getConnection();
+ conn.setAutoCommit(true);
+ conn.createStatement().execute("CREATE VIEW " + fullViewName + " AS
SELECT * FROM " + fullTableName);
+ conn.commit();
+
+ final String[] failedMsg = new String[1];
+ final CountDownLatch doneSignal = new CountDownLatch(2);
+ Runnable r1 = new Runnable() {
+
+ @Override public void run() {
+ try {
+ conn.createStatement().execute("CREATE INDEX " + indexName
+ " ON " + fullViewName + " (v1)");
+ } catch (SQLException e) {
+ failedMsg[0] = e.getMessage();
Review comment:
nit: Can we store and hence later assert the SQLException code rather
than the exception message?
##########
File path:
phoenix-core/src/it/java/org/apache/phoenix/end2end/index/ViewIndexIT.java
##########
@@ -169,10 +171,67 @@ public void testDeleteViewIndexSequences() throws
Exception {
verifySequenceValue(null, sequenceName, sequenceSchemaName,
Short.MIN_VALUE + 2);
conn1.createStatement().execute("DROP VIEW " + fullViewName);
conn1.createStatement().execute("DROP TABLE "+ fullTableName);
-
+
verifySequenceNotExists(null, sequenceName, sequenceSchemaName);
}
-
+
+ @Test
+ public void testDroppingColumnWhileCreatingIndex2() throws Exception {
+ String schemaName = "S1";
+ String tableName = generateUniqueName();
+ String viewSchemaName = "S1";
+ String fullTableName = SchemaUtil.getTableName(schemaName, tableName);
+ String indexName = "IND_" + generateUniqueName();
+ String viewName = "VIEW_" + generateUniqueName();
+ String fullViewName = SchemaUtil.getTableName(viewSchemaName,
viewName);
+
+ createBaseTable(schemaName, tableName, false, null, null, true);
+ Connection conn = getConnection();
Review comment:
nit: try-with-resources for the connection
##########
File path:
phoenix-core/src/main/java/org/apache/phoenix/schema/MetaDataClient.java
##########
@@ -1689,12 +1689,13 @@ public MutationState createIndex(CreateIndexStatement
statement, byte[][] splits
for (ColumnName colName : requiredCols) {
// acquire the mutex using the global physical table name
to
// prevent this column from being dropped while the index
is being created
+ String colNameSeparatedByDot =
colName.getColumnName().replace(':', '.');
Review comment:
Good to either extract this out as a SchemaUtil API and/or add a comment
here. Is there some existing API which makes the indexed column name have a ':'
instead of '.'?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
> View index creation does a checkAndPut on an incorrect row key
> --------------------------------------------------------------
>
> Key: PHOENIX-6083
> URL: https://issues.apache.org/jira/browse/PHOENIX-6083
> Project: Phoenix
> Issue Type: Bug
> Affects Versions: 5.0.0, 4.15.0
> Reporter: Chinmay Kulkarni
> Assignee: Richard Antal
> Priority: Critical
> Labels: phoenix-hardening, quality-improvement
> Fix For: 5.1.0, 4.16.0
>
> Attachments: image-2020-08-18-11-09-35-080.png,
> image-2020-08-18-11-09-53-816.png
>
>
> When creating a view index, we "lock" on the indexed and included columns by
> doing a checkAndPut on the cell:
> (<physical parent's schema name>, <physical parent's table name>, <name of
> the indexed/included column>)
> This is done to prevent another client from concurrently dropping any of the
> indexed/included columns. This however, is not working as expected since the
> indexed column naming convention is different (<column family>:<column name>)
> rather than just <column name>.
> Thus, concurrent dropping of the column is not prevented and this leads to
> failure in view index creation with either an NPE or IllegalArgumentException.
> !image-2020-08-18-11-09-35-080.png!
> !image-2020-08-18-11-09-53-816.png!
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
