You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@forrest.apache.org by Thorsten Scherler <th...@apache.org> on 2008/02/01 20:12:33 UTC
Re: [jira] Commented: (FOR-1066) Dispatcher started failing on
seed sample site recently with out-of-memory error
On Fri, 2008-02-01 at 05:03 -0800, David Crossley (JIRA) wrote:
> [ https://issues.apache.org/jira/browse/FOR-1066?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12564744#action_12564744 ]
>
> David Crossley commented on FOR-1066:
> -------------------------------------
>
> Hmm, that will hide it and so people tend not to bother to investigate.
Actually I am using the dispatcher in a very big project and need to
enhance the response time over the next couple of weeks. You may have
seen some commits about clean up to the cocoon-2.2 block.
>
> Also, is the headless=true necessary?
No, copy and paste since our server does not have an xserver installed.
>
> Hey i saw that input modules are involved in the error trace. I wonder if FOR-591 and FOR-1060 are relevant.
>
Need to have a look into this issues and the stack. My project will be
online next week I will need to find time to have a look.
BTW the stack is not attached to the issue, right?
salu2
> > Dispatcher started failing on seed sample site recently with out-of-memory error
> > --------------------------------------------------------------------------------
> >
> > Key: FOR-1066
> > URL: https://issues.apache.org/jira/browse/FOR-1066
> > Project: Forrest
> > Issue Type: Bug
> > Components: Plugin: internal.dispatcher, seed (fresh-site), zones.a.o administration
> > Affects Versions: 0.9-dev
> > Reporter: David Crossley
> > Fix For: 0.9-dev
> >
> >
> > cd $FORREST_HOME/main
> > ./build.sh test
> >
> > The first stage (the skins-based site) builds fine.
> > The second stage (the dispatcher-based site) hangs.
> > See email:
> > http://mail-archives.apache.org/mod_mbox/forrest-dev/200801.mbox/%3c20080131070310.GE2051@igg.indexgeo.com.au%3e
> > Re: dispatcher hanging with out of memory errors
>
--
Thorsten Scherler thorsten.at.apache.org
Open Source Java consulting, training and solutions
Re: [jira] Commented: (FOR-1066) Dispatcher started failing on seed sample site recently with out-of-memory error
Posted by David Crossley <cr...@apache.org>.
Thorsten Scherler wrote:
> > David Crossley commented on FOR-1066:
> >
> > Hey i saw that input modules are involved in the error trace. I wonder if FOR-591 and FOR-1060 are relevant.
>
> Need to have a look into this issues and the stack. My project will be
> online next week I will need to find time to have a look.
Good. We shouldn't let such issues pass. As said in another
thread, we cannot hog memory on the ASF infrastructure.
> BTW the stack is not attached to the issue, right?
I didn't see any need. Everyone should be running
'build test' anyway and so can directly see it for themselves.
Though i suppose it might be good to have the stacktrace
in Jira for future search purposes.
-David
> > > Dispatcher started failing on seed sample site recently with out-of-memory error
> > > --------------------------------------------------------------------------------
> > >
> > > Key: FOR-1066
> > > URL: https://issues.apache.org/jira/browse/FOR-1066
> > > Project: Forrest
> > > Issue Type: Bug
> > > Components: Plugin: internal.dispatcher, seed (fresh-site), zones.a.o administration
> > > Affects Versions: 0.9-dev
> > > Reporter: David Crossley
> > > Fix For: 0.9-dev
> > >
> > >
> > > cd $FORREST_HOME/main
> > > ./build.sh test
> > >
> > > The first stage (the skins-based site) builds fine.
> > > The second stage (the dispatcher-based site) hangs.
Re: [jira] Commented: (FOR-1066) Dispatcher started failing on
seed sample site recently with out-of-memory error
Posted by Thorsten Scherler <th...@apache.org>.
On Fri, 2008-02-08 at 02:20 +1100, David Crossley wrote:
> Thorsten Scherler wrote:
> > David Crossley wrote:
> > > Thorsten Scherler wrote:
> > > > David Crossley wrote:
> > > > > Thorsten Scherler wrote:
> > > > > > > David Crossley commented on FOR-1066:
> > > > > > > -------------------------------------
> > > > > > >
> > > > > > > Hey i saw that input modules are involved in the error trace. I wonder if FOR-591 and FOR-1060 are relevant.
> > > > > >
> > > > > > Need to have a look into this issues and the stack. My project will be
> > > > > > online next week I will need to find time to have a look.
> > > > >
> > > > > Ralph Goers recently did a lot of work at Cocoon to fix the input modules
> > > > > memory issues and instead created a more versatile system. See FOR-1060.
> > > > >
> > > > > Could Forrest use that independently
> > > >
> > > > In theory yes, not had looked deep enough into the diff of
> > > > http://issues.apache.org/jira/browse/COCOON-1574
> > >
> > > I actually meant his new XPathXMLFileModule mentioned at the
> > > end of COCOON-1574.
> > >
> > > However, looking at the Stack trace that you attached
> > > to this FOR-1066, i don't see input modules mentioned.
> > > So perhaps i was mistaken. Sorry for the noise.
> >
> > See my other mail, the attached stack may be different from the one you
> > saw. One would need to attach a stack without the profiler activated.
>
> You are correct, i should have gone to the extra effort
> to show the actual errors. Input modules are mentioned.
> Added output to FOR-1066.
>
> -David
Thanks.
salu2
--
Thorsten Scherler thorsten.at.apache.org
Open Source Java consulting, training and solutions
Re: [jira] Commented: (FOR-1066) Dispatcher started failing on seed sample site recently with out-of-memory error
Posted by David Crossley <cr...@apache.org>.
Thorsten Scherler wrote:
> David Crossley wrote:
> > Thorsten Scherler wrote:
> > > David Crossley wrote:
> > > > Thorsten Scherler wrote:
> > > > > > David Crossley commented on FOR-1066:
> > > > > > -------------------------------------
> > > > > >
> > > > > > Hey i saw that input modules are involved in the error trace. I wonder if FOR-591 and FOR-1060 are relevant.
> > > > >
> > > > > Need to have a look into this issues and the stack. My project will be
> > > > > online next week I will need to find time to have a look.
> > > >
> > > > Ralph Goers recently did a lot of work at Cocoon to fix the input modules
> > > > memory issues and instead created a more versatile system. See FOR-1060.
> > > >
> > > > Could Forrest use that independently
> > >
> > > In theory yes, not had looked deep enough into the diff of
> > > http://issues.apache.org/jira/browse/COCOON-1574
> >
> > I actually meant his new XPathXMLFileModule mentioned at the
> > end of COCOON-1574.
> >
> > However, looking at the Stack trace that you attached
> > to this FOR-1066, i don't see input modules mentioned.
> > So perhaps i was mistaken. Sorry for the noise.
>
> See my other mail, the attached stack may be different from the one you
> saw. One would need to attach a stack without the profiler activated.
You are correct, i should have gone to the extra effort
to show the actual errors. Input modules are mentioned.
Added output to FOR-1066.
-David
Re: [jira] Commented: (FOR-1066) Dispatcher started failing on
seed sample site recently with out-of-memory error
Posted by Thorsten Scherler <th...@juntadeandalucia.es>.
On Thu, 2008-02-07 at 12:22 +1100, David Crossley wrote:
> Thorsten Scherler wrote:
> > David Crossley wrote:
> > > Thorsten Scherler wrote:
> > > > > David Crossley commented on FOR-1066:
> > > > > -------------------------------------
> > > > >
> > > > > Hey i saw that input modules are involved in the error trace. I wonder if FOR-591 and FOR-1060 are relevant.
> > > >
> > > > Need to have a look into this issues and the stack. My project will be
> > > > online next week I will need to find time to have a look.
> > >
> > > Ralph Goers recently did a lot of work at Cocoon to fix the input modules
> > > memory issues and instead created a more versatile system. See FOR-1060.
> > >
> > > Could Forrest use that independently
> >
> > In theory yes, not had looked deep enough into the diff of
> > http://issues.apache.org/jira/browse/COCOON-1574
>
> I actually meant his new XPathXMLFileModule mentioned at the
> end of COCOON-1574.
>
> However, looking at the Stack trace that you attached
> to this FOR-1066, i don't see input modules mentioned.
> So perhaps i was mistaken. Sorry for the noise.
See my other mail, the attached stack may be different from the one you
saw. One would need to attach a stack without the profiler activated.
salu2
--
Thorsten Scherler thorsten.at.apache.org
Open Source Java consulting, training and solutions
Re: [jira] Commented: (FOR-1066) Dispatcher started failing on seed sample site recently with out-of-memory error
Posted by David Crossley <cr...@apache.org>.
Thorsten Scherler wrote:
> David Crossley wrote:
> > Thorsten Scherler wrote:
> > > > David Crossley commented on FOR-1066:
> > > > -------------------------------------
> > > >
> > > > Hey i saw that input modules are involved in the error trace. I wonder if FOR-591 and FOR-1060 are relevant.
> > >
> > > Need to have a look into this issues and the stack. My project will be
> > > online next week I will need to find time to have a look.
> >
> > Ralph Goers recently did a lot of work at Cocoon to fix the input modules
> > memory issues and instead created a more versatile system. See FOR-1060.
> >
> > Could Forrest use that independently
>
> In theory yes, not had looked deep enough into the diff of
> http://issues.apache.org/jira/browse/COCOON-1574
I actually meant his new XPathXMLFileModule mentioned at the
end of COCOON-1574.
However, looking at the Stack trace that you attached
to this FOR-1066, i don't see input modules mentioned.
So perhaps i was mistaken. Sorry for the noise.
-David
> > or do we need the
> > rest of Forrest core to be upgraded to use the current
> > Cocoon 2.2 ?
Re: [jira] Commented: (FOR-1066) Dispatcher started failing on
seed sample site recently with out-of-memory error
Posted by Thorsten Scherler <th...@apache.org>.
On Wed, 2008-02-06 at 12:21 +1100, David Crossley wrote:
> Thorsten Scherler wrote:
> > > David Crossley commented on FOR-1066:
> > > -------------------------------------
> > >
> > > Hmm, that will hide it and so people tend not to bother to investigate.
> >
> > Actually I am using the dispatcher in a very big project and need to
> > enhance the response time over the next couple of weeks. You may have
> > seen some commits about clean up to the cocoon-2.2 block.
>
> However, i gather that this is not yet used in the dispatcher-enabled
> forrest seed site, where we are seeing the memory issues.
I just committed the changes as well into the plugins. However this
changes do not enhance memory issues while generating the page but (like
stated in the commit message) "Will enhance response time for cached
pages.".
Actually as Ross pointed out the memory problem started when I activated
caching of the structurer file (jx). This enabled caching for the whole
pipe and it seems that the cached objects are quite big.
>
> > > Hey i saw that input modules are involved in the error trace. I wonder if FOR-591 and FOR-1060 are relevant.
> >
> > Need to have a look into this issues and the stack. My project will be
> > online next week I will need to find time to have a look.
>
> Ralph Goers recently did a lot of work at Cocoon to fix the input modules
> memory issues and instead created a more versatile system. See FOR-1060.
>
> Could Forrest use that independently
In theory yes, not had looked deep enough into the diff of
http://issues.apache.org/jira/browse/COCOON-1574
salu2
> or do we need the
> rest of Forrest core to be upgraded to use the current
> Cocoon 2.2 ?
>
> -David
--
Thorsten Scherler thorsten.at.apache.org
Open Source Java consulting, training and solutions
Re: [jira] Commented: (FOR-1066) Dispatcher started failing on seed sample site recently with out-of-memory error
Posted by David Crossley <cr...@apache.org>.
Thorsten Scherler wrote:
> > David Crossley commented on FOR-1066:
> > -------------------------------------
> >
> > Hmm, that will hide it and so people tend not to bother to investigate.
>
> Actually I am using the dispatcher in a very big project and need to
> enhance the response time over the next couple of weeks. You may have
> seen some commits about clean up to the cocoon-2.2 block.
However, i gather that this is not yet used in the dispatcher-enabled
forrest seed site, where we are seeing the memory issues.
> > Hey i saw that input modules are involved in the error trace. I wonder if FOR-591 and FOR-1060 are relevant.
>
> Need to have a look into this issues and the stack. My project will be
> online next week I will need to find time to have a look.
Ralph Goers recently did a lot of work at Cocoon to fix the input modules
memory issues and instead created a more versatile system. See FOR-1060.
Could Forrest use that independently or do we need the
rest of Forrest core to be upgraded to use the current
Cocoon 2.2 ?
-David