geronimo-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From David Jencks <>
Subject Re: Geronimo disk footprint growth at startup
Date Fri, 19 Aug 2005 01:06:09 GMT
well, from the file locations I believe they are the activemq journal 
which starts with 2 20mb files and can grow them if necessary.  I'm not 
at all sure they are still using howl itself.  I believe there is a way 
to disable journaling and use plain database persistence, although that 
will be definitely slower.

The transaction manager can be configured to use no logging: there is a 
tm in the app client that does this.  My first question before getting 
to the question of how much disk space is available on a tiny device 
is, whether tx recovery is really necessary on such a device.  The tm 
logs are in var/txlog and not that big.

david jencks

On Aug 17, 2005, at 1:09 PM, Dain Sundstrom wrote:

> On Aug 17, 2005, at 12:24 PM, Dave Colasurdo wrote:
>> Dain Sundstrom wrote:
>>> On Aug 17, 2005, at 10:11 AM, Dave Colasurdo wrote:
>>>> Thanks Bruce.. A few followup questions..
>>>> Is there any bound on how large these files can grow or a way to  
>>>> configure a smaller value for the recovery log?
>>> I don't believe the files grow at all.  IIRC the logger preallocates 
>>>  the files so it is fast.
>>>> Why does a simple server startup require 41M of recovery data?  No  
>>>> user applications have been installed on this server yet.
>>> Most likely, but what kind of user machine doesn't have 40M of free  
>>> disk space?
>> Was thinking of the case where geronimo is an embedded technology.. 
>> ISVs generally like embedded technologies to be as small and fast as 
>> possible.  Just seems strange that starting the server nearly doubles 
>> the disk footprint when no user applications are installed. Any 
>> chance of shrinking the default size for the recovery log.  Is 40M 
>> really needed? It seems a bit excessive..
> Sure.  In an embedded scenario, the ISV would tweak the howl 
> configuration to reduce the foot print.  Now exact what you change is 
> something David Jencks will have to tell us.
> -dain

View raw message