db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Knut Anders Hatlen <Knut.Hat...@Sun.COM>
Subject Re: regression errors on 10.6 branch
Date Tue, 08 Jun 2010 07:52:49 GMT
On 06/ 8/10 12:04 AM, Dag H. Wanvik wrote:
> Just saw these two fail on a regression run on the 10.6 branch.
> derbyall/derbynetclientmats/derbynetmats.fail:derbynet/testProtocol.java
> 148a148,149
>> FAILED - Expecting sqlCode 0 got ffffffff in line 1475
>> java.lang.Exception
> etc.
> derbyall/derbynetmats/derbynetmats.fail:lang/ShutdownDatabase.java
> Salient part of diff seems to be:
>> com.ibm.db2.jcc.c.DisconnectException: The application server
>> rejected establishment of the connection.  An attempt was made to
>> access a database, testOnlyTransactionWasCommitedDB;shutdown=true,
>> which was not found.
> Searched, but couldn't find likely open issues. Anybody else seen those?

FWIW, I ran derbyall on the 10.6 branch yesterday without seeing this error.

> I also saw one of the replication tests fail on a fork "not enough
> space", using the same setting I have used previously for suitesAll.
> -client -Xms128M -Xmx512M -XX:MaxPermSize=128m
> but maybe that is no longer sufficient?

I think the "not enough space" error when the process forks is rather
caused by the memory settings being too high, so that the original
process and the forked process combined take up too much physical
memory. I've been using the exact same memory settings myself, except
that I don't use -Xms, without problems so far. I have seen the "not
enough space" message once, though, but that was because the machine was
configured with little swap space. Increasing the swap space made the
tests run cleanly again.

Knut Anders

View raw message