Mailing-List: contact derby-dev-help@db.apache.org; run by ezmlm
Precedence: bulk
Reply-To: <derby-dev@db.apache.org>
Received-SPF: pass (athena.apache.org: local policy)
Date: Tue, 11 Sep 2007 11:23:53 +0200
From: =?UTF-8?B?SsO4cmdlbiBMw7hsYW5k?= <Jorgen.Loland@Sun.COM>
Subject: Re: Debugging Replication
In-reply-to: <46E640D4.8030804@Sun.COM>
Sender: Jorgen.Loland@Sun.COM
To: derby-dev@db.apache.org
Message-id: <46E65EA9.5010100@sun.com>
MIME-version: 1.0
Content-type: text/plain; format=flowed; charset=UTF-8
Content-transfer-encoding: 8BIT
References: <34619.1189427849637.JavaMail.jira@brutus>
 <46E640D4.8030804@Sun.COM>
User-Agent: Thunderbird 2.0.0.0 (X11/20070419)

Mayuresh Nirhali wrote:
> I would like to know if there are any thoughts around debugging the 
> proposed Replication functionality ??
> I am not following these discussions closely enough yet! But, I would 
> like to know if for example the logShipper will print out some 
> information in derby.log when the X log is shipped OR there will be a 
> separate machinery like say replicationTrace OR both with different 
> levels of debugging information ?
> 
> I believe this will be helpful in tracking Replication failures.
> Please excuse if this question is too early in the development effort 
> planned!

Hi Mayuresh,

Thanks for a great suggestion. We will definitely take this into account :)

Planning has started for junit testing of the replication feature, but
as far as I know, no plans have been made for writing messages/tracing 
to e.g. derby.log. However, I think this is a very good idea. We should 
at least have messages like:

derby.log on master:
ReplicationMaster flushTo: 999
ReplicationMaster LogShipper: sent message 1234 to slave, highest log 
instant: 999

derby.log on slave:
ReplicationSlave: received message 1234, highest log instant: 999
ReplicationSlave: appended log instant 555
ReplicationSlave: appended log instant 672
ReplicationSlave: appended log instant 999

It is to early to make a guess for how much and detailed information 
would be required, but that can be decided upon as the work progresses. 
The derby.log messages should be turned on by a replication_debug 
variable being true since a lot of messages would otherwise be generated.

-- 
Jørgen Løland