Mailing-List: contact derby-user-help@db.apache.org; run by ezmlm
Precedence: bulk
Reply-To: "Derby Discussion" <derby-user@db.apache.org>
Received-SPF: pass (nike.apache.org: local policy)
Message-ID: <49A72199.4000402@nuix.com>
Date: Fri, 27 Feb 2009 10:11:21 +1100
From: Daniel Noll <daniel@nuix.com>
Organization: Nuix Pty Ltd
User-Agent: Thunderbird 2.0.0.19 (Windows/20081209)
MIME-Version: 1.0
To: Derby Users Mailing List <derby-user@db.apache.org>
Subject: How many DRDAConnThreads is reasonable?
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

Hi all.

I'm seeing a situation where a Network Server is running out of memory 
creating new threads.  I say NetworkServer, but it's embedded into a 
little server application which does a few other things, though serving 
Derby databases is its primary purpose.

I had a look at the thread dump, and it looks kind of like this:

   428 x DRDAConnThread
   208 x derby.rawStoreDaemon
    50 x miscellaneous other threads (mostly our own)

208 corresponds to the number of databases which are open.  I imagine 
that 428 corresponds to the number of client connections.

Memory limits were set to 1G heap, 200M PermGen.  Heap is being lowered 
to 640M to see if that alleviates the problem a bit, but...

Is there any way we can reduce thread usage by setting some special 
Derby flag?  (e.g. some flag which makes it use selectors instead of 
having one thread per connection... or some way to reduce the number of 
raw store deamons.)

Daniel


-- 
Daniel Noll                            Forensic and eDiscovery Software
Senior Developer                              The world's most advanced
Nuix                                                email data analysis
http://nuix.com/                                and eDiscovery software