[jira] [Commented] (CASSANDRA-9484) Inconsistent select count
[ https://issues.apache.org/jira/browse/CASSANDRA-9484?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14959124#comment-14959124 ] Philip Thompson commented on CASSANDRA-9484: I can't reproduce this at all anymore, even on base 2.2 and 3.0. We can probably close this ticket. I'm going to merge the test into CI, and we can re-open this ticket if it starts failing again. > Inconsistent select count > - > > Key: CASSANDRA-9484 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9484 > Project: Cassandra > Issue Type: Bug >Reporter: Philip Thompson >Assignee: Benjamin Lerer > Fix For: 3.x, 2.2.x > > > I am running the dtest simultaneous_bootstrap_test located at > https://github.com/riptano/cassandra-dtest/compare/cassandra-7069 and finding > that at the final data verification step, the query {{SELECT COUNT (*) FROM > keyspace1.standard1}} alternated between correctly returning 500,000 rows and > returning 500,001 rows. Running cleanup or compaction does not affect the > behavior. I have verified with sstable2json that there are exactly 500k rows > on disk between the two nodes in the cluster. > I am reproducing this on trunk currently. It is not happening on 2.1-head. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (CASSANDRA-9484) Inconsistent select count
[ https://issues.apache.org/jira/browse/CASSANDRA-9484?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14958663#comment-14958663 ] Stefania commented on CASSANDRA-9484: - [~philipthompson] can you see if you can still reproduce it with the 2.2 patch of 10509? > Inconsistent select count > - > > Key: CASSANDRA-9484 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9484 > Project: Cassandra > Issue Type: Bug >Reporter: Philip Thompson >Assignee: Benjamin Lerer > Fix For: 3.x, 2.2.x > > > I am running the dtest simultaneous_bootstrap_test located at > https://github.com/riptano/cassandra-dtest/compare/cassandra-7069 and finding > that at the final data verification step, the query {{SELECT COUNT (*) FROM > keyspace1.standard1}} alternated between correctly returning 500,000 rows and > returning 500,001 rows. Running cleanup or compaction does not affect the > behavior. I have verified with sstable2json that there are exactly 500k rows > on disk between the two nodes in the cluster. > I am reproducing this on trunk currently. It is not happening on 2.1-head. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (CASSANDRA-9484) Inconsistent select count
[ https://issues.apache.org/jira/browse/CASSANDRA-9484?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14958502#comment-14958502 ] Stefania commented on CASSANDRA-9484: - This looks similar to CASSANDRA-10509. > Inconsistent select count > - > > Key: CASSANDRA-9484 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9484 > Project: Cassandra > Issue Type: Bug >Reporter: Philip Thompson >Assignee: Benjamin Lerer > Fix For: 3.x, 2.2.x > > > I am running the dtest simultaneous_bootstrap_test located at > https://github.com/riptano/cassandra-dtest/compare/cassandra-7069 and finding > that at the final data verification step, the query {{SELECT COUNT (*) FROM > keyspace1.standard1}} alternated between correctly returning 500,000 rows and > returning 500,001 rows. Running cleanup or compaction does not affect the > behavior. I have verified with sstable2json that there are exactly 500k rows > on disk between the two nodes in the cluster. > I am reproducing this on trunk currently. It is not happening on 2.1-head. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (CASSANDRA-9484) Inconsistent select count
[ https://issues.apache.org/jira/browse/CASSANDRA-9484?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14644598#comment-14644598 ] Philip Thompson commented on CASSANDRA-9484: I'm reproducing this locally on trunk with the validation harness. I have not seen it yet on linux, but it happens pretty consistently on OSX. > Inconsistent select count > - > > Key: CASSANDRA-9484 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9484 > Project: Cassandra > Issue Type: Bug >Reporter: Philip Thompson >Assignee: Benjamin Lerer > Fix For: 3.x, 2.2.x > > > I am running the dtest simultaneous_bootstrap_test located at > https://github.com/riptano/cassandra-dtest/compare/cassandra-7069 and finding > that at the final data verification step, the query {{SELECT COUNT (*) FROM > keyspace1.standard1}} alternated between correctly returning 500,000 rows and > returning 500,001 rows. Running cleanup or compaction does not affect the > behavior. I have verified with sstable2json that there are exactly 500k rows > on disk between the two nodes in the cluster. > I am reproducing this on trunk currently. It is not happening on 2.1-head. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (CASSANDRA-9484) Inconsistent select count
[ https://issues.apache.org/jira/browse/CASSANDRA-9484?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14561114#comment-14561114 ] Philip Thompson commented on CASSANDRA-9484: [~blerer], this passes on 2.1 before 8940 was committed, so it seems entirely unrelated. I'm unable to get a good bisect on trunk. > Inconsistent select count > - > > Key: CASSANDRA-9484 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9484 > Project: Cassandra > Issue Type: Bug >Reporter: Philip Thompson >Assignee: Benjamin Lerer > Fix For: 3.x, 2.2.x > > > I am running the dtest simultaneous_bootstrap_test located at > https://github.com/riptano/cassandra-dtest/compare/cassandra-7069 and finding > that at the final data verification step, the query {{SELECT COUNT (*) FROM > keyspace1.standard1}} alternated between correctly returning 500,000 rows and > returning 500,001 rows. Running cleanup or compaction does not affect the > behavior. I have verified with sstable2json that there are exactly 500k rows > on disk between the two nodes in the cluster. > I am reproducing this on trunk currently. It is not happening on 2.1-head. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (CASSANDRA-9484) Inconsistent select count
[ https://issues.apache.org/jira/browse/CASSANDRA-9484?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14560525#comment-14560525 ] Benjamin Lerer commented on CASSANDRA-9484: --- In CASSANDRA-8940 we only add a random number of missing rows. Never an extra one. > Inconsistent select count > - > > Key: CASSANDRA-9484 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9484 > Project: Cassandra > Issue Type: Bug >Reporter: Philip Thompson >Assignee: Benjamin Lerer > Fix For: 3.x, 2.2.x > > > I am running the dtest simultaneous_bootstrap_test located at > https://github.com/riptano/cassandra-dtest/compare/cassandra-7069 and finding > that at the final data verification step, the query {{SELECT COUNT (*) FROM > keyspace1.standard1}} alternated between correctly returning 500,000 rows and > returning 500,001 rows. Running cleanup or compaction does not affect the > behavior. I have verified with sstable2json that there are exactly 500k rows > on disk between the two nodes in the cluster. > I am reproducing this on trunk currently. It is not happening on 2.1-head. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (CASSANDRA-9484) Inconsistent select count
[ https://issues.apache.org/jira/browse/CASSANDRA-9484?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14560520#comment-14560520 ] Benjamin Lerer commented on CASSANDRA-9484: --- I might be wrong but I do not think that this issue is the same that CASSANDRA-8940. The delta seems to always be of +1 according to the description which make me think more about a paging issue. > Inconsistent select count > - > > Key: CASSANDRA-9484 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9484 > Project: Cassandra > Issue Type: Bug >Reporter: Philip Thompson >Assignee: Benjamin Lerer > Fix For: 3.x, 2.2.x > > > I am running the dtest simultaneous_bootstrap_test located at > https://github.com/riptano/cassandra-dtest/compare/cassandra-7069 and finding > that at the final data verification step, the query {{SELECT COUNT (*) FROM > keyspace1.standard1}} alternated between correctly returning 500,000 rows and > returning 500,001 rows. Running cleanup or compaction does not affect the > behavior. I have verified with sstable2json that there are exactly 500k rows > on disk between the two nodes in the cluster. > I am reproducing this on trunk currently. It is not happening on 2.1-head. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (CASSANDRA-9484) Inconsistent select count
[ https://issues.apache.org/jira/browse/CASSANDRA-9484?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14559793#comment-14559793 ] Philip Thompson commented on CASSANDRA-9484: This is with vnodes enabled, which from reading CASSANDRA-8940, was an issue more easily reproduced with vnodes. I'll try out a bisect. > Inconsistent select count > - > > Key: CASSANDRA-9484 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9484 > Project: Cassandra > Issue Type: Bug >Reporter: Philip Thompson >Assignee: Benjamin Lerer > Fix For: 3.x, 2.2.x > > > I am running the dtest simultaneous_bootstrap_test located at > https://github.com/riptano/cassandra-dtest/compare/cassandra-7069 and finding > that at the final data verification step, the query {{SELECT COUNT (*) FROM > keyspace1.standard1}} alternated between correctly returning 500,000 rows and > returning 500,001 rows. Running cleanup or compaction does not affect the > behavior. I have verified with sstable2json that there are exactly 500k rows > on disk between the two nodes in the cluster. > I am reproducing this on trunk currently. It is not happening on 2.1-head. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (CASSANDRA-9484) Inconsistent select count
[ https://issues.apache.org/jira/browse/CASSANDRA-9484?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14559630#comment-14559630 ] Jonathan Ellis commented on CASSANDRA-9484: --- Is this a vnode-enabled test? If it is separate from CASSANDRA-8940 then it could be worth bisecting to see where the trunk regression is from, but if it wasn't working on 2.1 either before 8940 that's probably not very useful. > Inconsistent select count > - > > Key: CASSANDRA-9484 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9484 > Project: Cassandra > Issue Type: Bug >Reporter: Philip Thompson >Assignee: Benjamin Lerer > Fix For: 3.x, 2.2.x > > > I am running the dtest simultaneous_bootstrap_test located at > https://github.com/riptano/cassandra-dtest/compare/cassandra-7069 and finding > that at the final data verification step, the query {{SELECT COUNT (*) FROM > keyspace1.standard1}} alternated between correctly returning 500,000 rows and > returning 500,001 rows. Running cleanup or compaction does not affect the > behavior. I have verified with sstable2json that there are exactly 500k rows > on disk between the two nodes in the cluster. > I am reproducing this on trunk currently. It is not happening on 2.1-head. -- This message was sent by Atlassian JIRA (v6.3.4#6332)