You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@cassandra.apache.org by "Sylvain Lebresne (JIRA)" <ji...@apache.org> on 2011/02/22 13:18:38 UTC

[jira] Created: (CASSANDRA-2216) Compaction can echo data which breaks upon sstable format changes

Compaction can echo data which breaks upon sstable format changes
-----------------------------------------------------------------

                 Key: CASSANDRA-2216
                 URL: https://issues.apache.org/jira/browse/CASSANDRA-2216
             Project: Cassandra
          Issue Type: Bug
          Components: Core
            Reporter: Sylvain Lebresne
            Assignee: Sylvain Lebresne
            Priority: Critical
             Fix For: 0.7.3


While compaction, if for a row we have only 1 sstable holding data, we echo this data. This breaks when we change the data format, creating mixed (corrupted) sstable.

(I suspect this is the cause of CASSANDRA-2195, but opening a new ticket until we can confirm that hunch)

-- 
This message is automatically generated by JIRA.
-
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

[jira] Updated: (CASSANDRA-2216) Compaction can echo data which breaks upon sstable format changes

Posted by "Jonathan Ellis (JIRA)" <ji...@apache.org>.
     [ https://issues.apache.org/jira/browse/CASSANDRA-2216?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Jonathan Ellis updated CASSANDRA-2216:
--------------------------------------

    Fix Version/s:     (was: 0.7.3)
                   0.7.1

> Compaction can echo data which breaks upon sstable format changes
> -----------------------------------------------------------------
>
>                 Key: CASSANDRA-2216
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-2216
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>            Reporter: Sylvain Lebresne
>            Assignee: Sylvain Lebresne
>            Priority: Critical
>              Labels: compaction
>             Fix For: 0.7.1
>
>         Attachments: 0001-Don-t-echo-data-during-compaction.patch, 2216_v2.patch
>
>   Original Estimate: 1h
>  Remaining Estimate: 1h
>
> While compaction, if for a row we have only 1 sstable holding data, we echo this data. This breaks when we change the data format, creating mixed (corrupted) sstable.
> (I suspect this is the cause of CASSANDRA-2195, but opening a new ticket until we can confirm that hunch)

-- 
This message is automatically generated by JIRA.
-
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

[jira] Updated: (CASSANDRA-2216) Compaction can echo data which breaks upon sstable format changes

Posted by "Sylvain Lebresne (JIRA)" <ji...@apache.org>.
     [ https://issues.apache.org/jira/browse/CASSANDRA-2216?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Sylvain Lebresne updated CASSANDRA-2216:
----------------------------------------

    Attachment: 2216_v2.patch

> Compaction can echo data which breaks upon sstable format changes
> -----------------------------------------------------------------
>
>                 Key: CASSANDRA-2216
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-2216
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>            Reporter: Sylvain Lebresne
>            Assignee: Sylvain Lebresne
>            Priority: Critical
>              Labels: compaction
>             Fix For: 0.7.3
>
>         Attachments: 0001-Don-t-echo-data-during-compaction.patch, 2216_v2.patch
>
>   Original Estimate: 1h
>  Remaining Estimate: 1h
>
> While compaction, if for a row we have only 1 sstable holding data, we echo this data. This breaks when we change the data format, creating mixed (corrupted) sstable.
> (I suspect this is the cause of CASSANDRA-2195, but opening a new ticket until we can confirm that hunch)

-- 
This message is automatically generated by JIRA.
-
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

[jira] Updated: (CASSANDRA-2216) Compaction can echo data which breaks upon sstable format changes

Posted by "Sylvain Lebresne (JIRA)" <ji...@apache.org>.
     [ https://issues.apache.org/jira/browse/CASSANDRA-2216?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Sylvain Lebresne updated CASSANDRA-2216:
----------------------------------------

    Attachment: 0001-Don-t-echo-data-during-compaction.patch

Attached patch completely remove the echoing of data when we have only one row. We could easily, as in CASSANDRA-2211, echo data if the sstable we are echoing from is at the last version.

However, not doing so will allow potentially corrupted sstable to get repaired by compaction (in the case of corruption from the bloom filter change).

We could add back the echoing optimisation later on.

> Compaction can echo data which breaks upon sstable format changes
> -----------------------------------------------------------------
>
>                 Key: CASSANDRA-2216
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-2216
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>            Reporter: Sylvain Lebresne
>            Assignee: Sylvain Lebresne
>            Priority: Critical
>              Labels: compaction
>             Fix For: 0.7.3
>
>         Attachments: 0001-Don-t-echo-data-during-compaction.patch
>
>   Original Estimate: 1h
>  Remaining Estimate: 1h
>
> While compaction, if for a row we have only 1 sstable holding data, we echo this data. This breaks when we change the data format, creating mixed (corrupted) sstable.
> (I suspect this is the cause of CASSANDRA-2195, but opening a new ticket until we can confirm that hunch)

-- 
This message is automatically generated by JIRA.
-
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

[jira] Updated: (CASSANDRA-2216) Compaction can echo data which breaks upon sstable format changes

Posted by "Sylvain Lebresne (JIRA)" <ji...@apache.org>.
     [ https://issues.apache.org/jira/browse/CASSANDRA-2216?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Sylvain Lebresne updated CASSANDRA-2216:
----------------------------------------

    Attachment: 2216_v2.patch

Attaching v2 that does the optimisation of checking for last version. Implies we repair the inconsistencies introduced outside of compaction.

> Compaction can echo data which breaks upon sstable format changes
> -----------------------------------------------------------------
>
>                 Key: CASSANDRA-2216
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-2216
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>            Reporter: Sylvain Lebresne
>            Assignee: Sylvain Lebresne
>            Priority: Critical
>              Labels: compaction
>             Fix For: 0.7.3
>
>         Attachments: 0001-Don-t-echo-data-during-compaction.patch, 2216_v2.patch
>
>   Original Estimate: 1h
>  Remaining Estimate: 1h
>
> While compaction, if for a row we have only 1 sstable holding data, we echo this data. This breaks when we change the data format, creating mixed (corrupted) sstable.
> (I suspect this is the cause of CASSANDRA-2195, but opening a new ticket until we can confirm that hunch)

-- 
This message is automatically generated by JIRA.
-
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

[jira] Updated: (CASSANDRA-2216) Compaction can echo data which breaks upon sstable format changes

Posted by "Sylvain Lebresne (JIRA)" <ji...@apache.org>.
     [ https://issues.apache.org/jira/browse/CASSANDRA-2216?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Sylvain Lebresne updated CASSANDRA-2216:
----------------------------------------

    Attachment:     (was: 2216_v2.patch)

> Compaction can echo data which breaks upon sstable format changes
> -----------------------------------------------------------------
>
>                 Key: CASSANDRA-2216
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-2216
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>            Reporter: Sylvain Lebresne
>            Assignee: Sylvain Lebresne
>            Priority: Critical
>              Labels: compaction
>             Fix For: 0.7.3
>
>         Attachments: 0001-Don-t-echo-data-during-compaction.patch, 2216_v2.patch
>
>   Original Estimate: 1h
>  Remaining Estimate: 1h
>
> While compaction, if for a row we have only 1 sstable holding data, we echo this data. This breaks when we change the data format, creating mixed (corrupted) sstable.
> (I suspect this is the cause of CASSANDRA-2195, but opening a new ticket until we can confirm that hunch)

-- 
This message is automatically generated by JIRA.
-
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

[jira] Updated: (CASSANDRA-2216) Compaction can echo data which breaks upon sstable format changes

Posted by "Jonathan Ellis (JIRA)" <ji...@apache.org>.
     [ https://issues.apache.org/jira/browse/CASSANDRA-2216?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Jonathan Ellis updated CASSANDRA-2216:
--------------------------------------

    Affects Version/s: 0.7.1
        Fix Version/s:     (was: 0.7.1)
                       0.7.3

> Compaction can echo data which breaks upon sstable format changes
> -----------------------------------------------------------------
>
>                 Key: CASSANDRA-2216
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-2216
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>    Affects Versions: 0.7.1
>            Reporter: Sylvain Lebresne
>            Assignee: Sylvain Lebresne
>            Priority: Critical
>              Labels: compaction
>             Fix For: 0.7.3
>
>         Attachments: 0001-Don-t-echo-data-during-compaction.patch, 2216_v2.patch
>
>   Original Estimate: 1h
>  Remaining Estimate: 1h
>
> While compaction, if for a row we have only 1 sstable holding data, we echo this data. This breaks when we change the data format, creating mixed (corrupted) sstable.
> (I suspect this is the cause of CASSANDRA-2195, but opening a new ticket until we can confirm that hunch)

-- 
This message is automatically generated by JIRA.
-
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

[jira] Commented: (CASSANDRA-2216) Compaction can echo data which breaks upon sstable format changes

Posted by "Hudson (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/CASSANDRA-2216?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12997971#comment-12997971 ] 

Hudson commented on CASSANDRA-2216:
-----------------------------------

Integrated in Cassandra-0.7 #304 (See [https://hudson.apache.org/hudson/job/Cassandra-0.7/304/])
    fix compaction echoing old-style data into new sstable version
patch by slebresne; reviewed by jbellis for CASSANDRA-2216


> Compaction can echo data which breaks upon sstable format changes
> -----------------------------------------------------------------
>
>                 Key: CASSANDRA-2216
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-2216
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>            Reporter: Sylvain Lebresne
>            Assignee: Sylvain Lebresne
>            Priority: Critical
>              Labels: compaction
>             Fix For: 0.7.1
>
>         Attachments: 0001-Don-t-echo-data-during-compaction.patch, 2216_v2.patch
>
>   Original Estimate: 1h
>  Remaining Estimate: 1h
>
> While compaction, if for a row we have only 1 sstable holding data, we echo this data. This breaks when we change the data format, creating mixed (corrupted) sstable.
> (I suspect this is the cause of CASSANDRA-2195, but opening a new ticket until we can confirm that hunch)

-- 
This message is automatically generated by JIRA.
-
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

[jira] Commented: (CASSANDRA-2216) Compaction can echo data which breaks upon sstable format changes

Posted by "Jonathan Ellis (JIRA)" <ji...@apache.org>.
    [ https://issues.apache.org/jira/browse/CASSANDRA-2216?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12997892#comment-12997892 ] 

Jonathan Ellis commented on CASSANDRA-2216:
-------------------------------------------

bq. We could easily echo data if the sstable we are echoing from is at the last version

Let's do that, and introduce a separate command to force deserialization.  (Telling people "compact to fix it" is not something we want to do since that leaves you with One Big SSTable and all the problems associated w/ that.)

> Compaction can echo data which breaks upon sstable format changes
> -----------------------------------------------------------------
>
>                 Key: CASSANDRA-2216
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-2216
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Core
>            Reporter: Sylvain Lebresne
>            Assignee: Sylvain Lebresne
>            Priority: Critical
>              Labels: compaction
>             Fix For: 0.7.3
>
>         Attachments: 0001-Don-t-echo-data-during-compaction.patch, 2216_v2.patch
>
>   Original Estimate: 1h
>  Remaining Estimate: 1h
>
> While compaction, if for a row we have only 1 sstable holding data, we echo this data. This breaks when we change the data format, creating mixed (corrupted) sstable.
> (I suspect this is the cause of CASSANDRA-2195, but opening a new ticket until we can confirm that hunch)

-- 
This message is automatically generated by JIRA.
-
For more information on JIRA, see: http://www.atlassian.com/software/jira