How are you loading jetty when you run it independently of paster?
 GeoNetwork runs afoul of a pretty nasty JVM bug which causes (as far as I
can tell) unbounded CPU and memory consumption.  'paster host' and the
startup.sh script avoid this by passing some extra options to the JVM
hosting Jetty, but if you just use "mvn jetty:run" then you'll have the
issue.

Other than that, I'll admit we haven't pushed GeoNode very hard in terms of
massive datasets; the biggest updatelayers job I've personally run has been
on demo.geonode.org which only has ~120 layers.  We are definitely
interested in having such imports run smoothly, so I'd be interested in
seeing those log files. When running via "paver host" the jetty and paster
output end up in .log files in the root directory of the geonode build.
 When running otherwise, you're on your own.  A useful trick for dumping
these files is the 'tee' command which dumps text to a file while also
displaying it to the console so you aren't flying blind.  If you are running
via startup.sh then using tee would look like:

$ sh startup.sh 2>&1 | tee jetty.log


Ideally (for me :) you would find a quick way to reproduce the memory
problem and reproduce the issue immediately after starting the server to
reduce unrelated messages in the log file.

In regards to layers with spaces in the names... I guess you are getting
these issues in gsconfig.py?  The geoserver catalog and the django backend
should both handle spaces fine. However, layer names with spaces can cause
problems in several places; for example, WFS responses will come back with
invalid XML because the layer name is used as an XML prefix.  Please provide
a stack trace of this as well, but in general layer names should not contain
spaces or other characters that are not valid as  part of an XML tag name.

--
David Winslow
OpenGeo - http://opengeo.org/

On Wed, Sep 22, 2010 at 11:06 AM, Joanne Cook <[email protected]>wrote:

> Hi David,
>
> Many thanks. I think I have this working, but only if I load jetty and
> paster separately. I'm still trying to track down why running paver host
> doesn't work. However, I have hit a big gotcha in that a lot of my data
> stores in my main geoserver installation have spaces in their names, and
> this is causing geonode to fail when I update the layers.
>
> On a slightly separate issue, when I run jetty and paster separately, I
> encounter some out of memory errors when loading data into geonode (this is
> from before I configured it to work with a remote geoserver instance). I
> have changed the settings for jetty, but these still occur. However, when I
> call them together using paver host, I don't get the out of memory errors.
> Unfortunately, if the jetty load times out because it has a lot of updates
> to do, then this causes errors- and also I have the problem noted above when
> running it in this way... I'm happy to provide more information on these
> things if you can tell me which logs you need and where to find them.
>
> Thanks
>
> Jo
>
> ----- Original Message -----
> From: "David Winslow" <[email protected]>
> To: [email protected]
> Sent: Tuesday, 21 September, 2010 2:07:02 PM
> Subject: Re: [geonode] Questions
>
> The paster/Django portion of GeoNode communicates with GeoServer
> exclusively over HTTP. So jetty must be running when you run the
> django-admin.py command. Additionally, the Django application must be
> running in order to provide GeoServer without authentication
> information. (That is, both paster and jetty must be running when you
> use the admin command.)
>
>
> The updatelayers command only imports layers, it does not remove
> existing ones, so your existing test layers will not be overwritten.
>
>
> -- David Winslow
> OpenGeo - http://opengeo.org/
>
>
> On Tue, Sep 21, 2010 at 6:15 AM, Joanne Cook < [email protected] >
> wrote:
>
>
> Thanks David,
>
> So to confirm I have understood this correctly- I would change
> GEOSERVER_BASE_URL in src/geonode/settings.py, run the django-admin.py
> command as below, then start geonode without jetty, using paster serve
> --reload shared/dev-paste.ini?
>
> Would this overwrite existing test layers that I have loaded into
> geonode (I am quite happy for it to do that)?
>
> Jo
>
>
>
>
> ----- Original Message -----
> From: "David Winslow" < [email protected] >
> To: [email protected]
> Sent: Monday, 20 September, 2010 5:29:22 PM
> Subject: Re: [geonode] Questions
>
> Once GeoNode is installed, the update script is integrated into the
> Django administration tool. You can invoke it like so:
>
>
>
>
> $ django-admin.py updatelayers --settings=geonode.settings
>
>
> Currently GeoNode only supports syncing with one GeoServer instance this
> way.
>
>
> Hope this helps.
>
>
> -- David Winslow
> OpenGeo - http://opengeo.org/
>
> On Mon, Sep 20, 2010 at 12:13 PM, Joanne Cook < [email protected]
> > wrote:
>
>
> Hi All,
>
> I met some of you at FOSS4G a few weeks ago, and also attended the
> tutorial on SDI best practices with Geonode. In that, someone
> (Sebastian?) mentioned being able to connect Geonode to a remote
> Geoserver instance, and run a script to get Geonode up to date with the
> Geoserver layers. Can someone give me some more information on where I
> might find this script?
>
> Thanks
>
> Jo
>
> --
> ----------------------------------------------------- Joanne Cook
> Senior IT Support and Development
> Oxford Archaeology (North)
> 01524 880212
> http://thehumanjourney.net
>
>
> ------ Files attached to this email may be in ISO 26300 format (OASIS
> Open Document Format). If you have difficulty opening them, please visit
> http://iso26300.info for more information.
>
> --
>
>
>
> ----------------------------------------------------- Joanne Cook
> Senior IT Support and Development
> Oxford Archaeology (North)
> 01524 880212
> http://thehumanjourney.net
>
>
> ------ Files attached to this email may be in ISO 26300 format (OASIS
> Open Document Format). If you have difficulty opening them, please visit
> http://iso26300.info for more information.
>
> --
> -----------------------------------------------------
> Joanne Cook
> Senior IT Support and Development
> Oxford Archaeology (North)
> 01524 880212
> http://thehumanjourney.net
>
>
> ------
> Files attached to this email may be in ISO 26300 format (OASIS Open
> Document Format). If you have difficulty opening them, please visit
> http://iso26300.info for more information.
>
>

Reply via email to