[jira] [Commented] (CASSANDRA-5614) W/O specified columns ASPCSI does not get notified of deletes
[ https://issues.apache.org/jira/browse/CASSANDRA-5614?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13718111#comment-13718111 ] Sam Tunnicliffe commented on CASSANDRA-5614: I realised I'd missed the case where columns are added after a RT, so attaching a fourth patch which handles that. My github branch is https://github.com/beobal/cassandra/tree/5614 > W/O specified columns ASPCSI does not get notified of deletes > - > > Key: CASSANDRA-5614 > URL: https://issues.apache.org/jira/browse/CASSANDRA-5614 > Project: Cassandra > Issue Type: Bug >Affects Versions: 1.2.0 >Reporter: Benjamin Coverston >Assignee: Sam Tunnicliffe >Priority: Minor > Fix For: 1.2.7 > > Attachments: > 0001-CASSANDRA-5614-PreCompactedRow-updates-2i-correctly.patch, > 0002-CASSANDRA-5614-LazilyCompactedRow-outputs-SSTables-a.patch, > 0003-CASSANDRA-5614-Memtable-updates-with-RowTombstone-up.patch, > 0004-CASSANDRA-5614-Consider-timestamps-when-checking-col.patch > > > I'm working on a secondary index implementation based on the composite index > type. > AbstractSimplePerColumnSecondaryIndex.java#delete is not called when CQL > delete statements do not specify columns. > When I specify columns it is called. Pretty sure this is a bug. > Setup: > {code} > cqlsh> create KEYSPACE foo WITH replication = {'class': 'SimpleStrategy' , > 'replication_factor': 1}; > cqlsh> use foo; > cqlsh:foo> CREATE TABLE albums (artist text, album text, rating int, release > int, PRIMARY KEY (artist, album)); > cqlsh:foo> CREATE INDEX ON albums (rating); > {code} > {code} > cqlsh:foo> insert into albums (artist, album, rating, release) VALUES > ('artist', 'album', 1, 2); > {code} > Does not get called here: > {code} > cqlsh:foo> DELETE FROM albums where artist='artist' and album='album'; > {code} > {code} > cqlsh:foo> insert into albums (artist, album, rating, release) VALUES > ('artist', 'album', 1, 2); > {code} > gets called here: > {code} > cqlsh:foo> DELETE rating FROM albums where artist='artist' and album='album'; > {code} -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (CASSANDRA-5614) W/O specified columns ASPCSI does not get notified of deletes
[ https://issues.apache.org/jira/browse/CASSANDRA-5614?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13717914#comment-13717914 ] Jonathan Ellis commented on CASSANDRA-5614: --- Thanks! Do you have a github branch for this? > W/O specified columns ASPCSI does not get notified of deletes > - > > Key: CASSANDRA-5614 > URL: https://issues.apache.org/jira/browse/CASSANDRA-5614 > Project: Cassandra > Issue Type: Bug >Affects Versions: 1.2.0 >Reporter: Benjamin Coverston >Assignee: Sam Tunnicliffe >Priority: Minor > Fix For: 1.2.7 > > Attachments: > 0001-CASSANDRA-5614-PreCompactedRow-updates-2i-correctly.patch, > 0002-CASSANDRA-5614-LazilyCompactedRow-outputs-SSTables-a.patch, > 0003-CASSANDRA-5614-Memtable-updates-with-RowTombstone-up.patch > > > I'm working on a secondary index implementation based on the composite index > type. > AbstractSimplePerColumnSecondaryIndex.java#delete is not called when CQL > delete statements do not specify columns. > When I specify columns it is called. Pretty sure this is a bug. > Setup: > {code} > cqlsh> create KEYSPACE foo WITH replication = {'class': 'SimpleStrategy' , > 'replication_factor': 1}; > cqlsh> use foo; > cqlsh:foo> CREATE TABLE albums (artist text, album text, rating int, release > int, PRIMARY KEY (artist, album)); > cqlsh:foo> CREATE INDEX ON albums (rating); > {code} > {code} > cqlsh:foo> insert into albums (artist, album, rating, release) VALUES > ('artist', 'album', 1, 2); > {code} > Does not get called here: > {code} > cqlsh:foo> DELETE FROM albums where artist='artist' and album='album'; > {code} > {code} > cqlsh:foo> insert into albums (artist, album, rating, release) VALUES > ('artist', 'album', 1, 2); > {code} > gets called here: > {code} > cqlsh:foo> DELETE rating FROM albums where artist='artist' and album='album'; > {code} -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (CASSANDRA-5614) W/O specified columns ASPCSI does not get notified of deletes
[ https://issues.apache.org/jira/browse/CASSANDRA-5614?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13716905#comment-13716905 ] Sam Tunnicliffe commented on CASSANDRA-5614: First, we aren't doing the right thing in compaction either. Neither PCR or LCR clean up the index in the face of range tombstones, relying on repairs at read time to purge obsolete entries. Looking into this, it also seems that PCR & LCR actually generate different SSTables. Although they're functionally equivalent, those that come from PCR don't have any columns covered by a RangeTombstone. SSTables generated by LCR do contain these redundant extra columns. eg: if we have two SSTables: {code} [{"key": "6e697276616e61","columns": [ ["bleach:","",1374246211348000], ["bleach:rating","4",1374246211348000], ["bleach:release","1",1374246211348000], ["nevermind:","",1374246211365000], ["nevermind:rating","5",1374246211365000], ["nevermind:release","2",1374246211365000] ] }] [{"key": "6e697276616e61","columns": [ ["nevermind","nevermind:!",1374246212053000,"t",1374246212] ] }] {code} When compacted with PCR we end up with in : {code} [{ "key": "6e697276616e61", "columns": [ ["bleach:","",1374246658577000], ["bleach:rating","4",1374246658577000], ["bleach:release","1",1374246658577000], ["nevermind","nevermind:!",1374246659274000,"t",1374246659] ] }] {code} But with LCR we get: {code} [{ "key": "6e697276616e61", "columns": [ ["bleach:","",1374246211348000], ["bleach:rating","4",1374246211348000], ["bleach:release","1",1374246211348000], ["nevermind","nevermind:!",1374246212053000,"t",1374246212], ["nevermind:","",1374246211365000], ["nevermind:rating","5",1374246211365000], ["nevermind:release","2",1374246211365000] ] }] {code} The first patch fixes PCR to clean up indexes properly. The second fixes LCR so that it generates SSTables like PCR & fixes its index cleanup. The third adds the index cleanup for memtable updates with RT. > W/O specified columns ASPCSI does not get notified of deletes > - > > Key: CASSANDRA-5614 > URL: https://issues.apache.org/jira/browse/CASSANDRA-5614 > Project: Cassandra > Issue Type: Bug >Affects Versions: 1.2.0 >Reporter: Benjamin Coverston >Assignee: Sam Tunnicliffe >Priority: Minor > Fix For: 1.2.7 > > Attachments: > 0001-CASSANDRA-5614-PreCompactedRow-updates-2i-correctly.patch, > 0002-CASSANDRA-5614-LazilyCompactedRow-outputs-SSTables-a.patch, > 0003-CASSANDRA-5614-Memtable-updates-with-RowTombstone-up.patch > > > I'm working on a secondary index implementation based on the composite index > type. > AbstractSimplePerColumnSecondaryIndex.java#delete is not called when CQL > delete statements do not specify columns. > When I specify columns it is called. Pretty sure this is a bug. > Setup: > {code} > cqlsh> create KEYSPACE foo WITH replication = {'class': 'SimpleStrategy' , > 'replication_factor': 1}; > cqlsh> use foo; > cqlsh:foo> CREATE TABLE albums (artist text, album text, rating int, release > int, PRIMARY KEY (artist, album)); > cqlsh:foo> CREATE INDEX ON albums (rating); > {code} > {code} > cqlsh:foo> insert into albums (artist, album, rating, release) VALUES > ('artist', 'album', 1, 2); > {code} > Does not get called here: > {code} > cqlsh:foo> DELETE FROM albums where artist='artist' and album='album'; > {code} > {code} > cqlsh:foo> insert into albums (artist, album, rating, release) VALUES > ('artist', 'album', 1, 2); > {code} > gets called here: > {code} > cqlsh:foo> DELETE rating FROM albums where artist='artist' and album='album'; > {code} -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (CASSANDRA-5614) W/O specified columns ASPCSI does not get notified of deletes
[ https://issues.apache.org/jira/browse/CASSANDRA-5614?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13710358#comment-13710358 ] Jonathan Ellis commented on CASSANDRA-5614: --- bq. When a column is deleted by name and that column is present in the memtable, we have the value and so can do that. When the named column isn't in the memtable, then we don't have the old value Right, but we know the old value is in the sstable and will be cleaned out on compaction. So I still don't see what would break here if we "expand" the tombstone to the indexed columns -- the same logic applies, we either get it from the memtable or at compaction time. (I *think* we apply row tombstones to indexed data correctly in compaction already -- if so, all we need to do is handle the memtable case, and not actually create extra tombstones.) > W/O specified columns ASPCSI does not get notified of deletes > - > > Key: CASSANDRA-5614 > URL: https://issues.apache.org/jira/browse/CASSANDRA-5614 > Project: Cassandra > Issue Type: Bug >Affects Versions: 1.2.0 >Reporter: Benjamin Coverston >Assignee: Sam Tunnicliffe >Priority: Minor > Fix For: 1.2.7 > > > I'm working on a secondary index implementation based on the composite index > type. > AbstractSimplePerColumnSecondaryIndex.java#delete is not called when CQL > delete statements do not specify columns. > When I specify columns it is called. Pretty sure this is a bug. > Setup: > {code} > cqlsh> create KEYSPACE foo WITH replication = {'class': 'SimpleStrategy' , > 'replication_factor': 1}; > cqlsh> use foo; > cqlsh:foo> CREATE TABLE albums (artist text, album text, rating int, release > int, PRIMARY KEY (artist, album)); > cqlsh:foo> CREATE INDEX ON albums (rating); > {code} > {code} > cqlsh:foo> insert into albums (artist, album, rating, release) VALUES > ('artist', 'album', 1, 2); > {code} > Does not get called here: > {code} > cqlsh:foo> DELETE FROM albums where artist='artist' and album='album'; > {code} > {code} > cqlsh:foo> insert into albums (artist, album, rating, release) VALUES > ('artist', 'album', 1, 2); > {code} > gets called here: > {code} > cqlsh:foo> DELETE rating FROM albums where artist='artist' and album='album'; > {code} -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (CASSANDRA-5614) W/O specified columns ASPCSI does not get notified of deletes
[ https://issues.apache.org/jira/browse/CASSANDRA-5614?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13707501#comment-13707501 ] Sam Tunnicliffe commented on CASSANDRA-5614: To delete from the 2i cf you need the old column value to derive the 2i cf key. When a column is deleted by name and that column is present in the memtable, we have the value and so can do that. When the named column isn't in the memtable, then we don't have the old value so we skip the delete and treat the operation as an insert. In SIM.StandardUpdater this is a no-op when isMarkedForDelete() == true, so even when the column is named if it's not in the memtable ASPCSI.delete is never called. > W/O specified columns ASPCSI does not get notified of deletes > - > > Key: CASSANDRA-5614 > URL: https://issues.apache.org/jira/browse/CASSANDRA-5614 > Project: Cassandra > Issue Type: Bug >Affects Versions: 1.2.0 >Reporter: Benjamin Coverston >Assignee: Sam Tunnicliffe >Priority: Minor > Fix For: 1.2.7 > > > I'm working on a secondary index implementation based on the composite index > type. > AbstractSimplePerColumnSecondaryIndex.java#delete is not called when CQL > delete statements do not specify columns. > When I specify columns it is called. Pretty sure this is a bug. > Setup: > {code} > cqlsh> create KEYSPACE foo WITH replication = {'class': 'SimpleStrategy' , > 'replication_factor': 1}; > cqlsh> use foo; > cqlsh:foo> CREATE TABLE albums (artist text, album text, rating int, release > int, PRIMARY KEY (artist, album)); > cqlsh:foo> CREATE INDEX ON albums (rating); > {code} > {code} > cqlsh:foo> insert into albums (artist, album, rating, release) VALUES > ('artist', 'album', 1, 2); > {code} > Does not get called here: > {code} > cqlsh:foo> DELETE FROM albums where artist='artist' and album='album'; > {code} > {code} > cqlsh:foo> insert into albums (artist, album, rating, release) VALUES > ('artist', 'album', 1, 2); > {code} > gets called here: > {code} > cqlsh:foo> DELETE rating FROM albums where artist='artist' and album='album'; > {code} -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (CASSANDRA-5614) W/O specified columns ASPCSI does not get notified of deletes
[ https://issues.apache.org/jira/browse/CASSANDRA-5614?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13707412#comment-13707412 ] Jonathan Ellis commented on CASSANDRA-5614: --- I don't see how this is different from "pretend the user had deleted the indexed columns by name" as far as read-before-write goes. > W/O specified columns ASPCSI does not get notified of deletes > - > > Key: CASSANDRA-5614 > URL: https://issues.apache.org/jira/browse/CASSANDRA-5614 > Project: Cassandra > Issue Type: Bug >Affects Versions: 1.2.0 >Reporter: Benjamin Coverston >Assignee: Sam Tunnicliffe >Priority: Minor > Fix For: 1.2.7 > > > I'm working on a secondary index implementation based on the composite index > type. > AbstractSimplePerColumnSecondaryIndex.java#delete is not called when CQL > delete statements do not specify columns. > When I specify columns it is called. Pretty sure this is a bug. > Setup: > {code} > cqlsh> create KEYSPACE foo WITH replication = {'class': 'SimpleStrategy' , > 'replication_factor': 1}; > cqlsh> use foo; > cqlsh:foo> CREATE TABLE albums (artist text, album text, rating int, release > int, PRIMARY KEY (artist, album)); > cqlsh:foo> CREATE INDEX ON albums (rating); > {code} > {code} > cqlsh:foo> insert into albums (artist, album, rating, release) VALUES > ('artist', 'album', 1, 2); > {code} > Does not get called here: > {code} > cqlsh:foo> DELETE FROM albums where artist='artist' and album='album'; > {code} > {code} > cqlsh:foo> insert into albums (artist, album, rating, release) VALUES > ('artist', 'album', 1, 2); > {code} > gets called here: > {code} > cqlsh:foo> DELETE rating FROM albums where artist='artist' and album='album'; > {code} -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (CASSANDRA-5614) W/O specified columns ASPCSI does not get notified of deletes
[ https://issues.apache.org/jira/browse/CASSANDRA-5614?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13706971#comment-13706971 ] Sam Tunnicliffe commented on CASSANDRA-5614: I'm tempted to say that this is not a problem because it works as expected with both KeysIndex & CompositesIndex. Yes, the presence of range rather than individual column tombstones in the DeletionInfo means that the index isn't updated directly at update time, but the index repair-on-read compensates for that at query time. We could take the deletion info from the update & use that in some new deleteAll method on SIM.Updater which iterates through the indexed columns issuing deletes, but AFAICT that would require reads to aquire the existing indexed values. > W/O specified columns ASPCSI does not get notified of deletes > - > > Key: CASSANDRA-5614 > URL: https://issues.apache.org/jira/browse/CASSANDRA-5614 > Project: Cassandra > Issue Type: Bug >Affects Versions: 1.2.0 >Reporter: Benjamin Coverston >Assignee: Sam Tunnicliffe >Priority: Minor > Fix For: 1.2.7 > > > I'm working on a secondary index implementation based on the composite index > type. > AbstractSimplePerColumnSecondaryIndex.java#delete is not called when CQL > delete statements do not specify columns. > When I specify columns it is called. Pretty sure this is a bug. > Setup: > {code} > cqlsh> create KEYSPACE foo WITH replication = {'class': 'SimpleStrategy' , > 'replication_factor': 1}; > cqlsh> use foo; > cqlsh:foo> CREATE TABLE albums (artist text, album text, rating int, release > int, PRIMARY KEY (artist, album)); > cqlsh:foo> CREATE INDEX ON albums (rating); > {code} > {code} > cqlsh:foo> insert into albums (artist, album, rating, release) VALUES > ('artist', 'album', 1, 2); > {code} > Does not get called here: > {code} > cqlsh:foo> DELETE FROM albums where artist='artist' and album='album'; > {code} > {code} > cqlsh:foo> insert into albums (artist, album, rating, release) VALUES > ('artist', 'album', 1, 2); > {code} > gets called here: > {code} > cqlsh:foo> DELETE rating FROM albums where artist='artist' and album='album'; > {code} -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (CASSANDRA-5614) W/O specified columns ASPCSI does not get notified of deletes
[ https://issues.apache.org/jira/browse/CASSANDRA-5614?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13676057#comment-13676057 ] Benjamin Coverston commented on CASSANDRA-5614: --- Instead of expanding them, at least for the purposes of what I'm doing I would be OK with just getting keybytes, and column if column had a special marker for a row tombstone. > W/O specified columns ASPCSI does not get notified of deletes > - > > Key: CASSANDRA-5614 > URL: https://issues.apache.org/jira/browse/CASSANDRA-5614 > Project: Cassandra > Issue Type: Bug >Affects Versions: 1.2.0 >Reporter: Benjamin Coverston >Assignee: Sam Overton >Priority: Minor > Fix For: 1.2.6 > > > I'm working on a secondary index implementation based on the composite index > type. > AbstractSimplePerColumnSecondaryIndex.java#delete is not called when CQL > delete statements do not specify columns. > When I specify columns it is called. Pretty sure this is a bug. > Setup: > {code} > cqlsh> create KEYSPACE foo WITH replication = {'class': 'SimpleStrategy' , > 'replication_factor': 1}; > cqlsh> use foo; > cqlsh:foo> CREATE TABLE albums (artist text, album text, rating int, release > int, PRIMARY KEY (artist, album)); > cqlsh:foo> CREATE INDEX ON albums (rating); > {code} > {code} > cqlsh:foo> insert into albums (artist, album, rating, release) VALUES > ('artist', 'album', 1, 2); > {code} > Does not get called here: > {code} > cqlsh:foo> DELETE FROM albums where artist='artist' and album='album'; > {code} > {code} > cqlsh:foo> insert into albums (artist, album, rating, release) VALUES > ('artist', 'album', 1, 2); > {code} > gets called here: > {code} > cqlsh:foo> DELETE rating FROM albums where artist='artist' and album='album'; > {code} -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira