db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Igor Sereda (JIRA)" <j...@apache.org>
Subject [jira] [Created] (DERBY-5980) Extremely slow roll forward when booting database that wasn't properly closed
Date Sun, 04 Nov 2012 10:30:12 GMT
Igor Sereda created DERBY-5980:
----------------------------------

             Summary: Extremely slow roll forward when booting database that wasn't properly
closed
                 Key: DERBY-5980
                 URL: https://issues.apache.org/jira/browse/DERBY-5980
             Project: Derby
          Issue Type: Bug
    Affects Versions: 10.9.1.0
         Environment: Windows 7, Embedded mode
            Reporter: Igor Sereda


Reproduction sequence:
1. Open database in embedded mode
2. Run a considerably large transaction - for example, one that creates 10 - 20 MB of db logs
3. Commit
4. Do NOT close the Connection
5. Exit the application (through System.exit() for example)
6. Restart the application and open the database

Observed:
Opening connection is extremely slow. It takes up 30- 60 seconds on the 10-20 MB transaction,
and for larger transactions it takes forever.

Looking at the process with FileMon utility, I can see that redo log files are mostly read
*1 byte at a time*. :(
 I bet you can improve the performance 100-fold by getting the whole log into memory first.

--
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

Mime
View raw message