db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Kathey Marsden (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (DERBY-5994) Different SQL State is returned when executing "shutdown=true" - locally or remotely using ClientDriver
Date Fri, 16 Nov 2012 23:35:12 GMT

    [ https://issues.apache.org/jira/browse/DERBY-5994?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13499242#comment-13499242
] 

Kathey Marsden commented on DERBY-5994:
---------------------------------------

If you agree that this is an embedded/client difference then  it would be good to adjust the
title accordingly and then address in conjunction with DERBY-2068 in a feature release if
deemed worth the risk of changing SQLStates.

Perhaps it would be good to send a separate email to the user list about your general replication
best practices question:
"The reason why this is important is because when using Derby Replication, when replication
is started - the databases must be identical and in sync, so the database must be shutdown
cleanly, copied over, and replication will be started. Since the Salve Derby must start before
Master Derby, it is required to be able to remotely shutdown the Master Derby database from
the Slave Derby instance (host). "

 
I am not familiar with replication really at all but this seems a valid best practices question
and I wonder if  shutdown of the database from the client is the recommended way.  As discussed
in this thread http://old.nabble.com/How-to-reset-Derby-Database-in-JVM-td34541038.html, client
shutdown of the database is not something that is documented or necessarily a good idea.

 
                
> Different SQL State is returned when executing "shutdown=true" - locally or remotely
using ClientDriver
> -------------------------------------------------------------------------------------------------------
>
>                 Key: DERBY-5994
>                 URL: https://issues.apache.org/jira/browse/DERBY-5994
>             Project: Derby
>          Issue Type: Bug
>          Components: JDBC, Network Client, Network Server, Replication
>    Affects Versions: 10.8.2.2
>         Environment: Derby Embedded Network Server - Not specific to operating systems,
software platforms or hardware specifications.
>            Reporter: Oskar Zinger
>         Attachments: Derby5994Repro.java
>
>
> Different SQL State is returned when executing "shutdown=true" Derby URL attribute while
executing the shutdown Database connection locally or remotely. Note: in both cases the client
driver is used: "org.apache.derby.jdbc.ClientDriver"
> Here is some background information:
> - Using the embedded Derby Network Server deployment (derbynet.jar)
> - Using the JDBC Derby Client (derbyclient.jar)
> - Using the Derby Engine (derby.jar)
> Note: in both cases the client driver is used: "org.apache.derby.jdbc.ClientDriver".
The only difference is the URL, when host and port is not specified execution happens in the
same JVM (local), when host and port is specified execution happens into another JVM (remote).
> When executing the following - if database has not been booted yet (or already shutdown)
- the SQL State Code is "XJ004":
>   DriverManager.getConnection("jdbc:derby:MyDatabase;shutdown=true;deregister=false",
user, pass);
> When executing the following - if database has not been booted yet (or already shutdown)
- the SQL State Code is "08004":
>   DriverManager.getConnection("jdbc:derby://myhost.mycompany.net:1527/MyDatabase;shutdown=true;deregister=false",
user, pass);
> The SQL State Code of "08004" usually is returned with an SQL Exception if "User Authentication
is Invalid" for example username and / or password is incorrect.
> Here is the error message: 
> Error Code: '40000', SQL State Code: '08004' and Exception: 'java.sql.SQLNonTransientConnectionException:
The connection was refused because the database <database_name>;shutdown=true;deregister=false
was not found.'
> Separately, the error message "database was not found" is actually misleading because
the database actually does exist and it just has not been been booted yet, or it has already
been shutdown. The Derby Network Server has not been shutdown, only the database and it is
possible to check if the database exists on the file system.
> The reason why this is important is because when using Derby Replication, when replication
is started - the databases must be identical and in sync, so the database must be shutdown
cleanly, copied over, and replication will be started. Since the Salve Derby must start before
Master Derby, it is required to be able to remotely shutdown the Master Derby database from
the Slave Derby instance (host).

--
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