[jira] [Commented] (CASSANDRA-5921) Don't return empty list when the L0 compaction candidates could cause overlap in L1

2014-02-06 Thread Ravi Prasad (JIRA)

[ 
https://issues.apache.org/jira/browse/CASSANDRA-5921?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13894154#comment-13894154
 ] 

Ravi Prasad commented on CASSANDRA-5921:


we're seeing overlaps in L1 in cassandra-2.0.4, multithreaded_compaction:false, 
concurrent_compactors:default(num_cores). Also reported by other user in ilist:
http://qnalist.com/questions/4702288/exception-during-add-node-due-to-test-beforeappend-on-sstablewriter

reverting the changes here to pre-2.0/CASSANDRA-5907, don't see any overlaps.
 

 Don't return empty list when the L0 compaction candidates could cause overlap 
 in L1
 ---

 Key: CASSANDRA-5921
 URL: https://issues.apache.org/jira/browse/CASSANDRA-5921
 Project: Cassandra
  Issue Type: Improvement
Reporter: Marcus Eriksson
Assignee: Marcus Eriksson
Priority: Minor
 Fix For: 2.0.1

 Attachments: 
 0001-instead-of-doing-no-compaction-if-we-have-sstables-t.patch, 
 0001-instead-of-doing-no-compaction-if-we-have-sstables-t.patch, 5921-v3.txt


 Followup to CASSANDRA-5907 - instead of returning empty list when the 
 compaction candidates could cause overlap in L1, remove sstables that would 
 cause the overlap from the candidates.



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)


[jira] [Commented] (CASSANDRA-5921) Don't return empty list when the L0 compaction candidates could cause overlap in L1

2013-08-29 Thread Yuki Morishita (JIRA)

[ 
https://issues.apache.org/jira/browse/CASSANDRA-5921?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13753729#comment-13753729
 ] 

Yuki Morishita commented on CASSANDRA-5921:
---

+1

 Don't return empty list when the L0 compaction candidates could cause overlap 
 in L1
 ---

 Key: CASSANDRA-5921
 URL: https://issues.apache.org/jira/browse/CASSANDRA-5921
 Project: Cassandra
  Issue Type: Improvement
Reporter: Marcus Eriksson
Assignee: Marcus Eriksson
Priority: Minor
 Attachments: 
 0001-instead-of-doing-no-compaction-if-we-have-sstables-t.patch, 
 0001-instead-of-doing-no-compaction-if-we-have-sstables-t.patch, 5921-v3.txt


 Followup to CASSANDRA-5907 - instead of returning empty list when the 
 compaction candidates could cause overlap in L1, remove sstables that would 
 cause the overlap from the candidates.

--
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-5921) Don't return empty list when the L0 compaction candidates could cause overlap in L1

2013-08-29 Thread Jonathan Ellis (JIRA)

[ 
https://issues.apache.org/jira/browse/CASSANDRA-5921?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13753870#comment-13753870
 ] 

Jonathan Ellis commented on CASSANDRA-5921:
---

committed to 2.0 and trunk

 Don't return empty list when the L0 compaction candidates could cause overlap 
 in L1
 ---

 Key: CASSANDRA-5921
 URL: https://issues.apache.org/jira/browse/CASSANDRA-5921
 Project: Cassandra
  Issue Type: Improvement
Reporter: Marcus Eriksson
Assignee: Marcus Eriksson
Priority: Minor
 Fix For: 2.0.1

 Attachments: 
 0001-instead-of-doing-no-compaction-if-we-have-sstables-t.patch, 
 0001-instead-of-doing-no-compaction-if-we-have-sstables-t.patch, 5921-v3.txt


 Followup to CASSANDRA-5907 - instead of returning empty list when the 
 compaction candidates could cause overlap in L1, remove sstables that would 
 cause the overlap from the candidates.

--
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-5921) Don't return empty list when the L0 compaction candidates could cause overlap in L1

2013-08-28 Thread Yuki Morishita (JIRA)

[ 
https://issues.apache.org/jira/browse/CASSANDRA-5921?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13752544#comment-13752544
 ] 

Yuki Morishita commented on CASSANDRA-5921:
---

v3 works fine with the correction above.
I went a little bit further not to iterate over already compacting L0, but no 
success(LCSTest didn't succeed).
I think we can commit the patch now and revisit in the future.

 Don't return empty list when the L0 compaction candidates could cause overlap 
 in L1
 ---

 Key: CASSANDRA-5921
 URL: https://issues.apache.org/jira/browse/CASSANDRA-5921
 Project: Cassandra
  Issue Type: Improvement
Reporter: Marcus Eriksson
Assignee: Marcus Eriksson
Priority: Minor
 Attachments: 
 0001-instead-of-doing-no-compaction-if-we-have-sstables-t.patch, 
 0001-instead-of-doing-no-compaction-if-we-have-sstables-t.patch, 5921-v3.txt


 Followup to CASSANDRA-5907 - instead of returning empty list when the 
 compaction candidates could cause overlap in L1, remove sstables that would 
 cause the overlap from the candidates.

--
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-5921) Don't return empty list when the L0 compaction candidates could cause overlap in L1

2013-08-28 Thread Marcus Eriksson (JIRA)

[ 
https://issues.apache.org/jira/browse/CASSANDRA-5921?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13752744#comment-13752744
 ] 

Marcus Eriksson commented on CASSANDRA-5921:


yeah +1

 Don't return empty list when the L0 compaction candidates could cause overlap 
 in L1
 ---

 Key: CASSANDRA-5921
 URL: https://issues.apache.org/jira/browse/CASSANDRA-5921
 Project: Cassandra
  Issue Type: Improvement
Reporter: Marcus Eriksson
Assignee: Marcus Eriksson
Priority: Minor
 Attachments: 
 0001-instead-of-doing-no-compaction-if-we-have-sstables-t.patch, 
 0001-instead-of-doing-no-compaction-if-we-have-sstables-t.patch, 5921-v3.txt


 Followup to CASSANDRA-5907 - instead of returning empty list when the 
 compaction candidates could cause overlap in L1, remove sstables that would 
 cause the overlap from the candidates.

--
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-5921) Don't return empty list when the L0 compaction candidates could cause overlap in L1

2013-08-28 Thread Jonathan Ellis (JIRA)

[ 
https://issues.apache.org/jira/browse/CASSANDRA-5921?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13752779#comment-13752779
 ] 

Jonathan Ellis commented on CASSANDRA-5921:
---

Should we sneak this into 2.0.1?

 Don't return empty list when the L0 compaction candidates could cause overlap 
 in L1
 ---

 Key: CASSANDRA-5921
 URL: https://issues.apache.org/jira/browse/CASSANDRA-5921
 Project: Cassandra
  Issue Type: Improvement
Reporter: Marcus Eriksson
Assignee: Marcus Eriksson
Priority: Minor
 Attachments: 
 0001-instead-of-doing-no-compaction-if-we-have-sstables-t.patch, 
 0001-instead-of-doing-no-compaction-if-we-have-sstables-t.patch, 5921-v3.txt


 Followup to CASSANDRA-5907 - instead of returning empty list when the 
 compaction candidates could cause overlap in L1, remove sstables that would 
 cause the overlap from the candidates.

--
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-5921) Don't return empty list when the L0 compaction candidates could cause overlap in L1

2013-08-26 Thread Yuki Morishita (JIRA)

[ 
https://issues.apache.org/jira/browse/CASSANDRA-5921?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13750296#comment-13750296
 ] 

Yuki Morishita commented on CASSANDRA-5921:
---

[~jbellis] I suppose you mean:

{code}
if (!Sets.intersection(overlappedL0, compactingL0).isEmpty())
continue;
{code}

instead of:

{code}
if (!Sets.intersection(overlappedL0, overlappedL0).isEmpty())
continue;
{code}

?

 Don't return empty list when the L0 compaction candidates could cause overlap 
 in L1
 ---

 Key: CASSANDRA-5921
 URL: https://issues.apache.org/jira/browse/CASSANDRA-5921
 Project: Cassandra
  Issue Type: Improvement
Reporter: Marcus Eriksson
Assignee: Marcus Eriksson
Priority: Minor
 Attachments: 
 0001-instead-of-doing-no-compaction-if-we-have-sstables-t.patch, 
 0001-instead-of-doing-no-compaction-if-we-have-sstables-t.patch, 5921-v3.txt


 Followup to CASSANDRA-5907 - instead of returning empty list when the 
 compaction candidates could cause overlap in L1, remove sstables that would 
 cause the overlap from the candidates.

--
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-5921) Don't return empty list when the L0 compaction candidates could cause overlap in L1

2013-08-26 Thread Jonathan Ellis (JIRA)

[ 
https://issues.apache.org/jira/browse/CASSANDRA-5921?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13750307#comment-13750307
 ] 

Jonathan Ellis commented on CASSANDRA-5921:
---

Yes. :)

 Don't return empty list when the L0 compaction candidates could cause overlap 
 in L1
 ---

 Key: CASSANDRA-5921
 URL: https://issues.apache.org/jira/browse/CASSANDRA-5921
 Project: Cassandra
  Issue Type: Improvement
Reporter: Marcus Eriksson
Assignee: Marcus Eriksson
Priority: Minor
 Attachments: 
 0001-instead-of-doing-no-compaction-if-we-have-sstables-t.patch, 
 0001-instead-of-doing-no-compaction-if-we-have-sstables-t.patch, 5921-v3.txt


 Followup to CASSANDRA-5907 - instead of returning empty list when the 
 compaction candidates could cause overlap in L1, remove sstables that would 
 cause the overlap from the candidates.

--
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-5921) Don't return empty list when the L0 compaction candidates could cause overlap in L1

2013-08-22 Thread Jonathan Ellis (JIRA)

[ 
https://issues.apache.org/jira/browse/CASSANDRA-5921?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13747570#comment-13747570
 ] 

Jonathan Ellis commented on CASSANDRA-5921:
---

- You changed {{compactingL0}} to just {{compacting}} which would unfix 5907
- You'll also need to do the L0 difference before the union with L1 overlaps; 
otherwise the difference could remove L1 sstables that need to be included for 
correctness
- Time to address {{TODO try to find a set of L0 sstables that only overlaps 
with non-busy L1 sstables}} ?

 Don't return empty list when the L0 compaction candidates could cause overlap 
 in L1
 ---

 Key: CASSANDRA-5921
 URL: https://issues.apache.org/jira/browse/CASSANDRA-5921
 Project: Cassandra
  Issue Type: Improvement
Reporter: Marcus Eriksson
Assignee: Marcus Eriksson
Priority: Minor
 Attachments: 
 0001-instead-of-doing-no-compaction-if-we-have-sstables-t.patch


 Followup to CASSANDRA-5907 - instead of returning empty list when the 
 compaction candidates could cause overlap in L1, remove sstables that would 
 cause the overlap from the candidates.

--
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