[jira] [Updated] (CASSANDRA-13787) RangeTombstoneMarker and PartitionDeletion is not properly included in MV
[
https://issues.apache.org/jira/browse/CASSANDRA-13787?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Josh McKenzie updated CASSANDRA-13787:
--
Bug Category: Parent values: Correctness(12982)Level 1 values: API /
Semantic Implementation(12988)
> RangeTombstoneMarker and PartitionDeletion is not properly included in MV
> -
>
> Key: CASSANDRA-13787
> URL: https://issues.apache.org/jira/browse/CASSANDRA-13787
> Project: Cassandra
> Issue Type: Bug
> Components: Feature/Materialized Views, Legacy/Local Write-Read Paths
>Reporter: ZhaoYang
>Assignee: ZhaoYang
>Priority: Normal
> Fix For: 3.0.15, 3.11.1, 4.0
>
>
> Found two problems related to MV tombstone.
> 1. Range-tombstone-Marker being ignored after shadowing first row, subsequent
> base rows are not shadowed in TableViews.
> If the range tombstone was not flushed, it was used as deleted row to
> shadow new updates. It works correctly.
> After range tombstone was flushed, it was used as RangeTombstoneMarker
> and being skipped after shadowing first update. The bound of
> RangeTombstoneMarker seems wrong, it contained full clustering, but it should
> contain range or it should be multiple RangeTombstoneMarkers for multiple
> slices(aka. new updates)
> -2. Partition tombstone is not used when no existing live data, it will
> resurrect deleted cells. It was found in 11500 and included in that patch.-
> (Merged in CASSANDRA-11500)
> In order not to make 11500 patch more complicated, I will try fix
> range/partition tombstone issue here.
> {code:title=Tests to reproduce}
> @Test
> public void testExistingRangeTombstoneWithFlush() throws Throwable
> {
> testExistingRangeTombstone(true);
> }
> @Test
> public void testExistingRangeTombstoneWithoutFlush() throws Throwable
> {
> testExistingRangeTombstone(false);
> }
> public void testExistingRangeTombstone(boolean flush) throws Throwable
> {
> createTable("CREATE TABLE %s (k1 int, c1 int, c2 int, v1 int, v2 int,
> PRIMARY KEY (k1, c1, c2))");
> execute("USE " + keyspace());
> executeNet(protocolVersion, "USE " + keyspace());
> createView("view1",
>"CREATE MATERIALIZED VIEW view1 AS SELECT * FROM %%s WHERE
> k1 IS NOT NULL AND c1 IS NOT NULL AND c2 IS NOT NULL PRIMARY KEY (k1, c2,
> c1)");
> updateView("DELETE FROM %s USING TIMESTAMP 10 WHERE k1 = 1 and c1=1");
> if (flush)
>
> Keyspace.open(keyspace()).getColumnFamilyStore(currentTable()).forceBlockingFlush();
> String table = KEYSPACE + "." + currentTable();
> updateView("BEGIN BATCH " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 0,
> 0, 0, 0) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 0,
> 1, 0, 1) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 1,
> 0, 1, 0) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 1,
> 1, 1, 1) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 1,
> 2, 1, 2) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 1,
> 3, 1, 3) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 2,
> 0, 2, 0) USING TIMESTAMP 5; " +
> "APPLY BATCH");
> assertRowsIgnoringOrder(execute("select * from %s"),
> row(1, 0, 0, 0, 0),
> row(1, 0, 1, 0, 1),
> row(1, 2, 0, 2, 0));
> assertRowsIgnoringOrder(execute("select k1,c1,c2,v1,v2 from view1"),
> row(1, 0, 0, 0, 0),
> row(1, 0, 1, 0, 1),
> row(1, 2, 0, 2, 0));
> }
> @Test
> public void testExistingParitionDeletionWithFlush() throws Throwable
> {
> testExistingParitionDeletion(true);
> }
> @Test
> public void testExistingParitionDeletionWithoutFlush() throws Throwable
> {
> testExistingParitionDeletion(false);
> }
> public void testExistingParitionDeletion(boolean flush) throws Throwable
> {
> // for partition range deletion, need to know that existing row is
> shadowed instead of not existed.
> createTable("CREATE TABLE %s (a int, b int, c int, d int, PRIMARY KEY
> (a))");
> execute("USE " + keyspace());
> executeNet(protocolVersion, "USE " + keyspace());
> createView("mv_test1",
>"CREATE MATERIALIZED VIEW %s AS SE
[jira] [Updated] (CASSANDRA-13787) RangeTombstoneMarker and PartitionDeletion is not properly included in MV
[
https://issues.apache.org/jira/browse/CASSANDRA-13787?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
ZhaoYang updated CASSANDRA-13787:
-
Fix Version/s: 4.0
> RangeTombstoneMarker and PartitionDeletion is not properly included in MV
> -
>
> Key: CASSANDRA-13787
> URL: https://issues.apache.org/jira/browse/CASSANDRA-13787
> Project: Cassandra
> Issue Type: Bug
> Components: Local Write-Read Paths, Materialized Views
>Reporter: ZhaoYang
>Assignee: ZhaoYang
> Fix For: 3.0.15, 3.11.1, 4.0
>
>
> Found two problems related to MV tombstone.
> 1. Range-tombstone-Marker being ignored after shadowing first row, subsequent
> base rows are not shadowed in TableViews.
> If the range tombstone was not flushed, it was used as deleted row to
> shadow new updates. It works correctly.
> After range tombstone was flushed, it was used as RangeTombstoneMarker
> and being skipped after shadowing first update. The bound of
> RangeTombstoneMarker seems wrong, it contained full clustering, but it should
> contain range or it should be multiple RangeTombstoneMarkers for multiple
> slices(aka. new updates)
> -2. Partition tombstone is not used when no existing live data, it will
> resurrect deleted cells. It was found in 11500 and included in that patch.-
> (Merged in CASSANDRA-11500)
> In order not to make 11500 patch more complicated, I will try fix
> range/partition tombstone issue here.
> {code:title=Tests to reproduce}
> @Test
> public void testExistingRangeTombstoneWithFlush() throws Throwable
> {
> testExistingRangeTombstone(true);
> }
> @Test
> public void testExistingRangeTombstoneWithoutFlush() throws Throwable
> {
> testExistingRangeTombstone(false);
> }
> public void testExistingRangeTombstone(boolean flush) throws Throwable
> {
> createTable("CREATE TABLE %s (k1 int, c1 int, c2 int, v1 int, v2 int,
> PRIMARY KEY (k1, c1, c2))");
> execute("USE " + keyspace());
> executeNet(protocolVersion, "USE " + keyspace());
> createView("view1",
>"CREATE MATERIALIZED VIEW view1 AS SELECT * FROM %%s WHERE
> k1 IS NOT NULL AND c1 IS NOT NULL AND c2 IS NOT NULL PRIMARY KEY (k1, c2,
> c1)");
> updateView("DELETE FROM %s USING TIMESTAMP 10 WHERE k1 = 1 and c1=1");
> if (flush)
>
> Keyspace.open(keyspace()).getColumnFamilyStore(currentTable()).forceBlockingFlush();
> String table = KEYSPACE + "." + currentTable();
> updateView("BEGIN BATCH " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 0,
> 0, 0, 0) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 0,
> 1, 0, 1) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 1,
> 0, 1, 0) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 1,
> 1, 1, 1) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 1,
> 2, 1, 2) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 1,
> 3, 1, 3) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 2,
> 0, 2, 0) USING TIMESTAMP 5; " +
> "APPLY BATCH");
> assertRowsIgnoringOrder(execute("select * from %s"),
> row(1, 0, 0, 0, 0),
> row(1, 0, 1, 0, 1),
> row(1, 2, 0, 2, 0));
> assertRowsIgnoringOrder(execute("select k1,c1,c2,v1,v2 from view1"),
> row(1, 0, 0, 0, 0),
> row(1, 0, 1, 0, 1),
> row(1, 2, 0, 2, 0));
> }
> @Test
> public void testExistingParitionDeletionWithFlush() throws Throwable
> {
> testExistingParitionDeletion(true);
> }
> @Test
> public void testExistingParitionDeletionWithoutFlush() throws Throwable
> {
> testExistingParitionDeletion(false);
> }
> public void testExistingParitionDeletion(boolean flush) throws Throwable
> {
> // for partition range deletion, need to know that existing row is
> shadowed instead of not existed.
> createTable("CREATE TABLE %s (a int, b int, c int, d int, PRIMARY KEY
> (a))");
> execute("USE " + keyspace());
> executeNet(protocolVersion, "USE " + keyspace());
> createView("mv_test1",
>"CREATE MATERIALIZED VIEW %s AS SELECT * FROM %%s WHERE a
> IS NOT NULL AND b IS NOT NULL PRIMARY KEY (a, b)");
> Keyspace ks = Keyspace.open(keyspace());
>
[jira] [Updated] (CASSANDRA-13787) RangeTombstoneMarker and PartitionDeletion is not properly included in MV
[
https://issues.apache.org/jira/browse/CASSANDRA-13787?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Sylvain Lebresne updated CASSANDRA-13787:
-
Resolution: Fixed
Fix Version/s: 3.11.1
3.0.15
Status: Resolved (was: Ready to Commit)
Alright, committed, thanks.
> RangeTombstoneMarker and PartitionDeletion is not properly included in MV
> -
>
> Key: CASSANDRA-13787
> URL: https://issues.apache.org/jira/browse/CASSANDRA-13787
> Project: Cassandra
> Issue Type: Bug
> Components: Local Write-Read Paths, Materialized Views
>Reporter: ZhaoYang
>Assignee: ZhaoYang
> Fix For: 3.0.15, 3.11.1
>
>
> Found two problems related to MV tombstone.
> 1. Range-tombstone-Marker being ignored after shadowing first row, subsequent
> base rows are not shadowed in TableViews.
> If the range tombstone was not flushed, it was used as deleted row to
> shadow new updates. It works correctly.
> After range tombstone was flushed, it was used as RangeTombstoneMarker
> and being skipped after shadowing first update. The bound of
> RangeTombstoneMarker seems wrong, it contained full clustering, but it should
> contain range or it should be multiple RangeTombstoneMarkers for multiple
> slices(aka. new updates)
> -2. Partition tombstone is not used when no existing live data, it will
> resurrect deleted cells. It was found in 11500 and included in that patch.-
> (Merged in CASSANDRA-11500)
> In order not to make 11500 patch more complicated, I will try fix
> range/partition tombstone issue here.
> {code:title=Tests to reproduce}
> @Test
> public void testExistingRangeTombstoneWithFlush() throws Throwable
> {
> testExistingRangeTombstone(true);
> }
> @Test
> public void testExistingRangeTombstoneWithoutFlush() throws Throwable
> {
> testExistingRangeTombstone(false);
> }
> public void testExistingRangeTombstone(boolean flush) throws Throwable
> {
> createTable("CREATE TABLE %s (k1 int, c1 int, c2 int, v1 int, v2 int,
> PRIMARY KEY (k1, c1, c2))");
> execute("USE " + keyspace());
> executeNet(protocolVersion, "USE " + keyspace());
> createView("view1",
>"CREATE MATERIALIZED VIEW view1 AS SELECT * FROM %%s WHERE
> k1 IS NOT NULL AND c1 IS NOT NULL AND c2 IS NOT NULL PRIMARY KEY (k1, c2,
> c1)");
> updateView("DELETE FROM %s USING TIMESTAMP 10 WHERE k1 = 1 and c1=1");
> if (flush)
>
> Keyspace.open(keyspace()).getColumnFamilyStore(currentTable()).forceBlockingFlush();
> String table = KEYSPACE + "." + currentTable();
> updateView("BEGIN BATCH " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 0,
> 0, 0, 0) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 0,
> 1, 0, 1) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 1,
> 0, 1, 0) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 1,
> 1, 1, 1) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 1,
> 2, 1, 2) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 1,
> 3, 1, 3) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 2,
> 0, 2, 0) USING TIMESTAMP 5; " +
> "APPLY BATCH");
> assertRowsIgnoringOrder(execute("select * from %s"),
> row(1, 0, 0, 0, 0),
> row(1, 0, 1, 0, 1),
> row(1, 2, 0, 2, 0));
> assertRowsIgnoringOrder(execute("select k1,c1,c2,v1,v2 from view1"),
> row(1, 0, 0, 0, 0),
> row(1, 0, 1, 0, 1),
> row(1, 2, 0, 2, 0));
> }
> @Test
> public void testExistingParitionDeletionWithFlush() throws Throwable
> {
> testExistingParitionDeletion(true);
> }
> @Test
> public void testExistingParitionDeletionWithoutFlush() throws Throwable
> {
> testExistingParitionDeletion(false);
> }
> public void testExistingParitionDeletion(boolean flush) throws Throwable
> {
> // for partition range deletion, need to know that existing row is
> shadowed instead of not existed.
> createTable("CREATE TABLE %s (a int, b int, c int, d int, PRIMARY KEY
> (a))");
> execute("USE " + keyspace());
> executeNet(protocolVersion, "USE " + keyspace());
> createView("mv_test1",
>"CREATE MATERIALIZED VIEW
[jira] [Updated] (CASSANDRA-13787) RangeTombstoneMarker and PartitionDeletion is not properly included in MV
[
https://issues.apache.org/jira/browse/CASSANDRA-13787?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Anonymous updated CASSANDRA-13787:
--
Status: Ready to Commit (was: Patch Available)
> RangeTombstoneMarker and PartitionDeletion is not properly included in MV
> -
>
> Key: CASSANDRA-13787
> URL: https://issues.apache.org/jira/browse/CASSANDRA-13787
> Project: Cassandra
> Issue Type: Bug
> Components: Local Write-Read Paths, Materialized Views
>Reporter: ZhaoYang
>Assignee: ZhaoYang
>
> Found two problems related to MV tombstone.
> 1. Range-tombstone-Marker being ignored after shadowing first row, subsequent
> base rows are not shadowed in TableViews.
> If the range tombstone was not flushed, it was used as deleted row to
> shadow new updates. It works correctly.
> After range tombstone was flushed, it was used as RangeTombstoneMarker
> and being skipped after shadowing first update. The bound of
> RangeTombstoneMarker seems wrong, it contained full clustering, but it should
> contain range or it should be multiple RangeTombstoneMarkers for multiple
> slices(aka. new updates)
> -2. Partition tombstone is not used when no existing live data, it will
> resurrect deleted cells. It was found in 11500 and included in that patch.-
> (Merged in CASSANDRA-11500)
> In order not to make 11500 patch more complicated, I will try fix
> range/partition tombstone issue here.
> {code:title=Tests to reproduce}
> @Test
> public void testExistingRangeTombstoneWithFlush() throws Throwable
> {
> testExistingRangeTombstone(true);
> }
> @Test
> public void testExistingRangeTombstoneWithoutFlush() throws Throwable
> {
> testExistingRangeTombstone(false);
> }
> public void testExistingRangeTombstone(boolean flush) throws Throwable
> {
> createTable("CREATE TABLE %s (k1 int, c1 int, c2 int, v1 int, v2 int,
> PRIMARY KEY (k1, c1, c2))");
> execute("USE " + keyspace());
> executeNet(protocolVersion, "USE " + keyspace());
> createView("view1",
>"CREATE MATERIALIZED VIEW view1 AS SELECT * FROM %%s WHERE
> k1 IS NOT NULL AND c1 IS NOT NULL AND c2 IS NOT NULL PRIMARY KEY (k1, c2,
> c1)");
> updateView("DELETE FROM %s USING TIMESTAMP 10 WHERE k1 = 1 and c1=1");
> if (flush)
>
> Keyspace.open(keyspace()).getColumnFamilyStore(currentTable()).forceBlockingFlush();
> String table = KEYSPACE + "." + currentTable();
> updateView("BEGIN BATCH " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 0,
> 0, 0, 0) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 0,
> 1, 0, 1) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 1,
> 0, 1, 0) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 1,
> 1, 1, 1) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 1,
> 2, 1, 2) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 1,
> 3, 1, 3) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 2,
> 0, 2, 0) USING TIMESTAMP 5; " +
> "APPLY BATCH");
> assertRowsIgnoringOrder(execute("select * from %s"),
> row(1, 0, 0, 0, 0),
> row(1, 0, 1, 0, 1),
> row(1, 2, 0, 2, 0));
> assertRowsIgnoringOrder(execute("select k1,c1,c2,v1,v2 from view1"),
> row(1, 0, 0, 0, 0),
> row(1, 0, 1, 0, 1),
> row(1, 2, 0, 2, 0));
> }
> @Test
> public void testExistingParitionDeletionWithFlush() throws Throwable
> {
> testExistingParitionDeletion(true);
> }
> @Test
> public void testExistingParitionDeletionWithoutFlush() throws Throwable
> {
> testExistingParitionDeletion(false);
> }
> public void testExistingParitionDeletion(boolean flush) throws Throwable
> {
> // for partition range deletion, need to know that existing row is
> shadowed instead of not existed.
> createTable("CREATE TABLE %s (a int, b int, c int, d int, PRIMARY KEY
> (a))");
> execute("USE " + keyspace());
> executeNet(protocolVersion, "USE " + keyspace());
> createView("mv_test1",
>"CREATE MATERIALIZED VIEW %s AS SELECT * FROM %%s WHERE a
> IS NOT NULL AND b IS NOT NULL PRIMARY KEY (a, b)");
> Keyspace ks = Keyspace.open(keyspace());
> ks.getColumn
[jira] [Updated] (CASSANDRA-13787) RangeTombstoneMarker and PartitionDeletion is not properly included in MV
[
https://issues.apache.org/jira/browse/CASSANDRA-13787?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jeremy Hanna updated CASSANDRA-13787:
-
Summary: RangeTombstoneMarker and PartitionDeletion is not properly
included in MV (was: RangeTombstoneMarker and ParitionDeletion is not properly
included in MV)
> RangeTombstoneMarker and PartitionDeletion is not properly included in MV
> -
>
> Key: CASSANDRA-13787
> URL: https://issues.apache.org/jira/browse/CASSANDRA-13787
> Project: Cassandra
> Issue Type: Bug
> Components: Local Write-Read Paths, Materialized Views
>Reporter: ZhaoYang
>Assignee: ZhaoYang
>
> Found two problems related to MV tombstone.
> 1. Range-tombstone-Marker being ignored after shadowing first row, subsequent
> base rows are not shadowed in TableViews.
> If the range tombstone was not flushed, it was used as deleted row to
> shadow new updates. It works correctly.
> After range tombstone was flushed, it was used as RangeTombstoneMarker
> and being skipped after shadowing first update. The bound of
> RangeTombstoneMarker seems wrong, it contained full clustering, but it should
> contain range or it should be multiple RangeTombstoneMarkers for multiple
> slices(aka. new updates)
> -2. Partition tombstone is not used when no existing live data, it will
> resurrect deleted cells. It was found in 11500 and included in that patch.-
> (Merged in CASSANDRA-11500)
> In order not to make 11500 patch more complicated, I will try fix
> range/partition tombstone issue here.
> {code:title=Tests to reproduce}
> @Test
> public void testExistingRangeTombstoneWithFlush() throws Throwable
> {
> testExistingRangeTombstone(true);
> }
> @Test
> public void testExistingRangeTombstoneWithoutFlush() throws Throwable
> {
> testExistingRangeTombstone(false);
> }
> public void testExistingRangeTombstone(boolean flush) throws Throwable
> {
> createTable("CREATE TABLE %s (k1 int, c1 int, c2 int, v1 int, v2 int,
> PRIMARY KEY (k1, c1, c2))");
> execute("USE " + keyspace());
> executeNet(protocolVersion, "USE " + keyspace());
> createView("view1",
>"CREATE MATERIALIZED VIEW view1 AS SELECT * FROM %%s WHERE
> k1 IS NOT NULL AND c1 IS NOT NULL AND c2 IS NOT NULL PRIMARY KEY (k1, c2,
> c1)");
> updateView("DELETE FROM %s USING TIMESTAMP 10 WHERE k1 = 1 and c1=1");
> if (flush)
>
> Keyspace.open(keyspace()).getColumnFamilyStore(currentTable()).forceBlockingFlush();
> String table = KEYSPACE + "." + currentTable();
> updateView("BEGIN BATCH " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 0,
> 0, 0, 0) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 0,
> 1, 0, 1) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 1,
> 0, 1, 0) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 1,
> 1, 1, 1) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 1,
> 2, 1, 2) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 1,
> 3, 1, 3) USING TIMESTAMP 5; " +
> "INSERT INTO " + table + " (k1, c1, c2, v1, v2) VALUES (1, 2,
> 0, 2, 0) USING TIMESTAMP 5; " +
> "APPLY BATCH");
> assertRowsIgnoringOrder(execute("select * from %s"),
> row(1, 0, 0, 0, 0),
> row(1, 0, 1, 0, 1),
> row(1, 2, 0, 2, 0));
> assertRowsIgnoringOrder(execute("select k1,c1,c2,v1,v2 from view1"),
> row(1, 0, 0, 0, 0),
> row(1, 0, 1, 0, 1),
> row(1, 2, 0, 2, 0));
> }
> @Test
> public void testExistingParitionDeletionWithFlush() throws Throwable
> {
> testExistingParitionDeletion(true);
> }
> @Test
> public void testExistingParitionDeletionWithoutFlush() throws Throwable
> {
> testExistingParitionDeletion(false);
> }
> public void testExistingParitionDeletion(boolean flush) throws Throwable
> {
> // for partition range deletion, need to know that existing row is
> shadowed instead of not existed.
> createTable("CREATE TABLE %s (a int, b int, c int, d int, PRIMARY KEY
> (a))");
> execute("USE " + keyspace());
> executeNet(protocolVersion, "USE " + keyspace());
> createView("mv_test1",
>"CREATE MATERIALIZED VIEW %s AS SELECT * FROM %%s WHERE a
> I
