[jira] [Updated] (CASSANDRA-14945) During garbagecollect use index check for tombstone removals

2018-12-21 Thread Vitalii Ishchenko (JIRA) 


 [ 
https://issues.apache.org/jira/browse/CASSANDRA-14945?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Vitalii Ishchenko updated CASSANDRA-14945:
--
Priority: Minor  (was: Major)

> During garbagecollect use index check for tombstone removals
> 
>
> Key: CASSANDRA-14945
> URL: https://issues.apache.org/jira/browse/CASSANDRA-14945
> Project: Cassandra
>  Issue Type: Improvement
>  Components: Compaction
>Reporter: Vitalii Ishchenko
>Priority: Minor
>
> nodetool garbagecollect command may be further improved to actually remove 
> more tombstones. Right now bloom filter is used and may give false positives 
> preventing tombstone removal if min table timestamp is less than tombstone 
> timestamp.
> Disabling bloom filter should do the trick, but currently there is a bug in 
> compaction with bloom filter turned off CASSANDRA-14944
> As improvement index can be always checked if garbagecollect compaction is 
> done



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

-
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org

[jira] [Created] (CASSANDRA-14945) During garbagecollect use index check for tombstone removals

2018-12-21 Thread Vitalii Ishchenko (JIRA) 
Vitalii Ishchenko created CASSANDRA-14945:
-

 Summary: During garbagecollect use index check for tombstone 
removals
 Key: CASSANDRA-14945
 URL: https://issues.apache.org/jira/browse/CASSANDRA-14945
 Project: Cassandra
  Issue Type: Improvement
  Components: Compaction
Reporter: Vitalii Ishchenko


nodetool garbagecollect command may be further improved to actually remove more 
tombstones. Right now bloom filter is used and may give false positives 
preventing tombstone removal if min table timestamp is less than tombstone 
timestamp.

Disabling bloom filter should do the trick, but currently there is a bug in 
compaction with bloom filter turned off CASSANDRA-14944

As improvement index can be always checked if garbagecollect compaction is done



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

-
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org

[jira] [Created] (CASSANDRA-14944) Tombstones are not removed if bloom filter is turned off

2018-12-21 Thread Vitalii Ishchenko (JIRA) 
Vitalii Ishchenko created CASSANDRA-14944:
-

 Summary: Tombstones are not removed if bloom filter is turned off
 Key: CASSANDRA-14944
 URL: https://issues.apache.org/jira/browse/CASSANDRA-14944
 Project: Cassandra
  Issue Type: Bug
  Components: Compaction
Reporter: Vitalii Ishchenko


Well actually they are deleted, but not always even though they should, because 
check is done using Index of overlapped tables

When purge evaluator is constructed we are checking overlapping tables using 
bloom filter, but when it is disabled we are checking against index, but if 
condition is not properly constructed and we still check bloom filter which is 
AlwaysPresentFilter and every overlapping sstable is used to get minTimestamp 
and tombstones, that have their timestamp >= minTimestamp won't be deleted
{code:java}
if (sstable.getBloomFilter() instanceof AlwaysPresentFilter && 
sstable.getPosition(key, SSTableReader.Operator.EQ, false) != null
|| sstable.getBloomFilter().isPresent(key))
{
{code}
Should be something like this
{code:java}
boolean mightBePresentInTable = sstable.getBloomFilter() instanceof 
AlwaysPresentFilter ? sstable.getPosition(key, SSTableReader.Operator.EQ, 
false) != null : sstable.getBloomFilter().isPresent(key)
{code}
Code pointers in 3.11 
[https://github.com/apache/cassandra/blob/08363afa5354c00a7ecd62fe273c392a678db28a/src/java/org/apache/cassandra/db/compaction/CompactionController.java#L274]
 and 4.0 
[https://github.com/apache/cassandra/blob/11384c3279a66e6c0fb7861e2b188b25e963580f/src/java/org/apache/cassandra/db/compaction/CompactionController.java#L281]



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

-
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org

[jira] [Updated] (CASSANDRA-13808) Integer overflows with Amazon Elastic File System (EFS)

2017-09-05 Thread Vitalii Ishchenko (JIRA) 

 [ 
https://issues.apache.org/jira/browse/CASSANDRA-13808?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Vitalii Ishchenko updated CASSANDRA-13808:
--
Description: 
Integer overflow issue was fixed for cassandra 2.2, but in 3.8 new property was 
introduced in config that also derives from disk size  
{{cdc_total_space_in_mb}}, see CASSANDRA-8844

It should be updated too 
https://github.com/apache/cassandra/blob/6b7d73a49695c0ceb78bc7a003ace606a806c13a/src/java/org/apache/cassandra/config/DatabaseDescriptor.java#L484

  was:
Integer overflow issue was fixed for cassandra 2.2, but since then new property 
was introduced in config that also derives from disk size  
{{cdc_total_space_in_mb}}

It should be updated too 
https://github.com/apache/cassandra/blob/6b7d73a49695c0ceb78bc7a003ace606a806c13a/src/java/org/apache/cassandra/config/DatabaseDescriptor.java#L484


> Integer overflows with Amazon Elastic File System (EFS)
> ---
>
> Key: CASSANDRA-13808
> URL: https://issues.apache.org/jira/browse/CASSANDRA-13808
> Project: Cassandra
>  Issue Type: Bug
>  Components: Core
>Reporter: Vitalii Ishchenko
>Assignee: Benjamin Lerer
> Fix For: 3.11.1
>
>
> Integer overflow issue was fixed for cassandra 2.2, but in 3.8 new property 
> was introduced in config that also derives from disk size  
> {{cdc_total_space_in_mb}}, see CASSANDRA-8844
> It should be updated too 
> https://github.com/apache/cassandra/blob/6b7d73a49695c0ceb78bc7a003ace606a806c13a/src/java/org/apache/cassandra/config/DatabaseDescriptor.java#L484



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

-
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org

[jira] [Updated] (CASSANDRA-13808) Integer overflows with Amazon Elastic File System (EFS)

2017-09-05 Thread Vitalii Ishchenko (JIRA) 

 [ 
https://issues.apache.org/jira/browse/CASSANDRA-13808?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Vitalii Ishchenko updated CASSANDRA-13808:
--
Fix Version/s: 3.11.1

> Integer overflows with Amazon Elastic File System (EFS)
> ---
>
> Key: CASSANDRA-13808
> URL: https://issues.apache.org/jira/browse/CASSANDRA-13808
> Project: Cassandra
>  Issue Type: Bug
>  Components: Core
>Reporter: Vitalii Ishchenko
>Assignee: Benjamin Lerer
> Fix For: 3.11.1
>
>
> Integer overflow issue was fixed for cassandra 2.2, but since then new 
> property was introduced in config that also derives from disk size  
> {{cdc_total_space_in_mb}}
> It should be updated too 
> https://github.com/apache/cassandra/blob/6b7d73a49695c0ceb78bc7a003ace606a806c13a/src/java/org/apache/cassandra/config/DatabaseDescriptor.java#L484



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

-
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org

[jira] [Updated] (CASSANDRA-13808) Integer overflows with Amazon Elastic File System (EFS)

2017-08-25 Thread Vitalii Ishchenko (JIRA) 

 [ 
https://issues.apache.org/jira/browse/CASSANDRA-13808?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Vitalii Ishchenko updated CASSANDRA-13808:
--
Description: 
Integer overflow issue was fixed for cassandra 2.2, but since then new property 
was introduced in config that also derives from disk size  
{{cdc_total_space_in_mb}}

It should be updated too 
https://github.com/apache/cassandra/blob/6b7d73a49695c0ceb78bc7a003ace606a806c13a/src/java/org/apache/cassandra/config/DatabaseDescriptor.java#L484

  was:
Integer overflow issue was fixed for cassandra 2.2, but since then new property 
was introduced in config that also derives from disk size  
{{cdc_total_space_in_mb}}

It should be updated too 
https://github.com/apache/cassandra/blob/6b7d73a49695c0ceb78bc7a003ace606a806c13a/src/java/org/apache/cassandra/config/DatabaseDescriptor.java#L484

Related to CASSANDRA-13067


> Integer overflows with Amazon Elastic File System (EFS)
> ---
>
> Key: CASSANDRA-13808
> URL: https://issues.apache.org/jira/browse/CASSANDRA-13808
> Project: Cassandra
>  Issue Type: Bug
>  Components: Core
>Reporter: Vitalii Ishchenko
>
> Integer overflow issue was fixed for cassandra 2.2, but since then new 
> property was introduced in config that also derives from disk size  
> {{cdc_total_space_in_mb}}
> It should be updated too 
> https://github.com/apache/cassandra/blob/6b7d73a49695c0ceb78bc7a003ace606a806c13a/src/java/org/apache/cassandra/config/DatabaseDescriptor.java#L484



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

-
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org

[jira] [Updated] (CASSANDRA-13808) Integer overflows with Amazon Elastic File System (EFS)

2017-08-25 Thread Vitalii Ishchenko (JIRA) 

 [ 
https://issues.apache.org/jira/browse/CASSANDRA-13808?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Vitalii Ishchenko updated CASSANDRA-13808:
--
Description: 
Integer overflow issue was fixed for cassandra 2.2, but since then new property 
was introduced in config that also derives from disk size  
{{cdc_total_space_in_mb}}

It should be updated too 
https://github.com/apache/cassandra/blob/6b7d73a49695c0ceb78bc7a003ace606a806c13a/src/java/org/apache/cassandra/config/DatabaseDescriptor.java#L484

Related to CASSANDRA-13067

  was:
Integer overflow issue was fixed for cassandra 2.2, but since then new property 
was introduced in config that also derives from disk size  
{{cdc_total_space_in_mb}}

It should be updated too

https://github.com/apache/cassandra/blob/6b7d73a49695c0ceb78bc7a003ace606a806c13a/src/java/org/apache/cassandra/config/DatabaseDescriptor.java#L484


> Integer overflows with Amazon Elastic File System (EFS)
> ---
>
> Key: CASSANDRA-13808
> URL: https://issues.apache.org/jira/browse/CASSANDRA-13808
> Project: Cassandra
>  Issue Type: Bug
>  Components: Core
>Reporter: Vitalii Ishchenko
>
> Integer overflow issue was fixed for cassandra 2.2, but since then new 
> property was introduced in config that also derives from disk size  
> {{cdc_total_space_in_mb}}
> It should be updated too 
> https://github.com/apache/cassandra/blob/6b7d73a49695c0ceb78bc7a003ace606a806c13a/src/java/org/apache/cassandra/config/DatabaseDescriptor.java#L484
> Related to CASSANDRA-13067



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

-
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org

[jira] [Created] (CASSANDRA-13808) Integer overflows with Amazon Elastic File System (EFS)

2017-08-25 Thread Vitalii Ishchenko (JIRA) 
Vitalii Ishchenko created CASSANDRA-13808:
-

 Summary: Integer overflows with Amazon Elastic File System (EFS)
 Key: CASSANDRA-13808
 URL: https://issues.apache.org/jira/browse/CASSANDRA-13808
 Project: Cassandra
  Issue Type: Bug
  Components: Core
Reporter: Vitalii Ishchenko


Integer overflow issue was fixed for cassandra 2.2, but since then new property 
was introduced in config that also derives from disk size  
{{cdc_total_space_in_mb}}

It should be updated too

https://github.com/apache/cassandra/blob/6b7d73a49695c0ceb78bc7a003ace606a806c13a/src/java/org/apache/cassandra/config/DatabaseDescriptor.java#L484



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

-
To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org
For additional commands, e-mail: commits-h...@cassandra.apache.org