[
https://issues.apache.org/jira/browse/CASSANDRA-5898?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13744216#comment-13744216
]
Yuki Morishita commented on CASSANDRA-5898:
-------------------------------------------
Correct. They are just sent back to L0.
As you can see in the attached log, SSTables sent back are all from clean up
during compaction.
LCS checks overlap at the end, and if SSTables from other operation like
cleanup are added during compaction, that ERROR shows up, but no harm.
> nodetool cleanup interacting badly with LCS
> -------------------------------------------
>
> Key: CASSANDRA-5898
> URL: https://issues.apache.org/jira/browse/CASSANDRA-5898
> Project: Cassandra
> Issue Type: Bug
> Reporter: Jeremiah Jordan
> Assignee: Yuki Morishita
> Fix For: 1.2.0
>
> Attachments: system.log
>
>
> Then run cassandra-stress read, cassandra-stress write and "nodetool cleanup"
> all at the same time, a few times, it can cause overlapping row errors.
> {noformat}
> ERROR [CompactionExecutor:3] 2013-08-16 19:20:26,174 LeveledManifest.java
> (line 223) At level 1,
> SSTableReader(path='/raid0/cassandra/data/Keyspace1/Standard1/Keyspace1-Standard1-hf-245-Data.db')
> [DecoratedKey(204853724659241194183955214890519, 30303132343830),
> DecoratedKey(69227335985728660912035125310473966323, 30393537373332)]
> overlaps
> SSTableReader(path='/raid0/cassandra/data/Keyspace1/Standard1/Keyspace1-Standard1-hf-243-Data.db')
> [DecoratedKey(217896711032704014921095870827202, 30333635363932),
> DecoratedKey(71430242198281555888954138354238066233, 30333035343132)]. This
> is caused by a bug in Cassandra 1.1.0 .. 1.1.3. Sending back to L0. If you
> have not yet run scrub, you should do so since you may also have rows
> out-of-order within an sstable
> ERROR [CompactionExecutor:3] 2013-08-16 19:20:26,174 LeveledManifest.java
> (line 223) At level 1,
> SSTableReader(path='/raid0/cassandra/data/Keyspace1/Standard1/Keyspace1-Standard1-hf-213-Data.db')
> [DecoratedKey(69231122848869715605691825585281602523, 30303236343834),
> DecoratedKey(138494249350583165433071928358975719316, 30393733303134)]
> overlaps
> SSTableReader(path='/raid0/cassandra/data/Keyspace1/Standard1/Keyspace1-Standard1-hf-247-Data.db')
> [DecoratedKey(71430378028108400200608251408083114019, 30383133313937),
> DecoratedKey(142788557025615975974639291059795386925, 30353437373934)]. This
> is caused by a bug in Cassandra 1.1.0 .. 1.1.3. Sending back to L0. If you
> have not yet run scrub, you should do so since you may also have rows
> out-of-order within an sstable
> ERROR [CompactionExecutor:3] 2013-08-16 19:20:26,175 LeveledManifest.java
> (line 223) At level 1,
> SSTableReader(path='/raid0/cassandra/data/Keyspace1/Standard1/Keyspace1-Standard1-hf-249-Data.db')
> [DecoratedKey(138495280291659088173025258556445559667, 30313039353935),
> DecoratedKey(170139918483433930784261689358244392849, 30313336303439)]
> overlaps
> SSTableReader(path='/raid0/cassandra/data/Keyspace1/Standard1/Keyspace1-Standard1-hf-252-Data.db')
> [DecoratedKey(142788613459989682219570222310727294191, 30383031333839),
> DecoratedKey(170140890916069420209456014750832266661, 30333731303535)]. This
> is caused by a bug in Cassandra 1.1.0 .. 1.1.3. Sending back to L0. If you
> have not yet run scrub, you should do so since you may also have rows
> out-of-order within an sstable
> ERROR [CompactionExecutor:3] 2013-08-16 19:22:28,912 LeveledManifest.java
> (line 223) At level 1,
> SSTableReader(path='/raid0/cassandra/data/Keyspace1/Standard1/Keyspace1-Standard1-hf-271-Data.db')
> [DecoratedKey(68545052253582928820015773344685671746, 30303739313938),
> DecoratedKey(119320297902762425433209003370359962009, 30373137323238)]
> overlaps
> SSTableReader(path='/raid0/cassandra/data/Keyspace1/Standard1/Keyspace1-Standard1-hf-270-Data.db')
> [DecoratedKey(69231122848869715605691825585281602523, 30303236343834),
> DecoratedKey(138494249350583165433071928358975719316, 30393733303134)]. This
> is caused by a bug in Cassandra 1.1.0 .. 1.1.3. Sending back to L0. If you
> have not yet run scrub, you should do so since you may also have rows
> out-of-order within an sstable
> {noformat}
> Steps to repro:
> # Run cassandra-stress -Z LeveledCompactionStrategy
> # Run nodetool cleanup
> # While cleanup is still running, run cassandra-stress -o write and
> cassandra-stress -o read at the same time.
--
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
