db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Mike Matrigali (JIRA)" <derby-...@db.apache.org>
Subject [jira] Updated: (DERBY-691) committed deleted row space reclamation may be missed if delete is actually an aborted insert.
Date Fri, 03 Mar 2006 23:46:41 GMT
     [ http://issues.apache.org/jira/browse/DERBY-691?page=all ]

Mike Matrigali updated DERBY-691:
---------------------------------

    Description: 
The system should queue post commit activity in the case of an aborted insert which internally
is implemented as a delete.  In this case post commit should be queued following the same
rules as straight line delete currently.  The result is that if the last  delete on a page
is actually an aborted insert then space on that page may never be automatically reclaimed
(It is always can be reclaimed by hand using the space reclamation system procedures).

Note that a duplicate key insert error internally generates this case.  It first inserts the
row into the base table.  Then it attempts to insert a row into the unique index and fails.
 As part of the failure it aborts the statement which turns the insert into a delete as part
of the undo.  

  was:The system should queue post commit activity in the case of an aborted insert which
internally is implemented as a delete.  In this case post commit should be queued following
the same rules as straight line delete currently.  The result is that if the last  delete
on a page is actually an aborted insert then space on that page may never be automatically
reclaimed (It is always can be reclaimed by hand using the space reclamation system procedures).


> committed deleted row space reclamation may be missed if delete is actually an aborted
insert.
> ----------------------------------------------------------------------------------------------
>
>          Key: DERBY-691
>          URL: http://issues.apache.org/jira/browse/DERBY-691
>      Project: Derby
>         Type: Improvement
>   Components: Store
>     Versions: 10.0.2.0
>     Reporter: Mike Matrigali
>     Priority: Minor

>
> The system should queue post commit activity in the case of an aborted insert which internally
is implemented as a delete.  In this case post commit should be queued following the same
rules as straight line delete currently.  The result is that if the last  delete on a page
is actually an aborted insert then space on that page may never be automatically reclaimed
(It is always can be reclaimed by hand using the space reclamation system procedures).
> Note that a duplicate key insert error internally generates this case.  It first inserts
the row into the base table.  Then it attempts to insert a row into the unique index and fails.
 As part of the failure it aborts the statement which turns the insert into a delete as part
of the undo.  

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
   http://issues.apache.org/jira/secure/Administrators.jspa
-
For more information on JIRA, see:
   http://www.atlassian.com/software/jira


Mime
View raw message