Recently I have suffered from unusual errors in wms requests sent in POST mode. In short, sometimes geoserver stops evaluating the request with an "IOException: Mark invalid" in ows Dispatcher.java, readOpPost method in the line doing an "input.reset()".
The exception is not predictable (I'm not able to reproduce it in a stable way; with the same request sometimes the request is evaluated, sometimes not).
I've seen that elsewhere in the code, an "input.mark(2048)" (in two different points in the code) sets the readAheadLimit to 2K. I tried to change it to 8192 and the problem seems disappeared.
Have you ever seen this kind of exception before? I'm using GeoServer 1.7.1a and 1.7.3a on a WindowsXP PC.
Thanks in advance,
Mauro Bartolomeoli
Hi Mauro,
I have never seen this before no. If you have some sample requests i am happy to try them out though.
I would also be fine to upping the read ahead limit to 8K. Andrea: thoughts?
-Justin
Mauro Bartolomeoli wrote:
Recently I have suffered from unusual errors in wms requests sent in POST mode. In short, sometimes geoserver stops evaluating the request with an "IOException: Mark invalid" in ows Dispatcher.java, readOpPost method in the line doing an "input.reset()".
The exception is not predictable (I'm not able to reproduce it in a stable way; with the same request sometimes the request is evaluated, sometimes not).
I've seen that elsewhere in the code, an "input.mark(2048)" (in two different points in the code) sets the readAheadLimit to 2K. I tried to change it to 8192 and the problem seems disappeared.
Have you ever seen this kind of exception before? I'm using GeoServer 1.7.1a and 1.7.3a on a WindowsXP PC.
Thanks in advance,
Mauro Bartolomeoli
------------------------------------------------------------------------------
_______________________________________________
Geoserver-devel mailing list
Geoserver-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/geoserver-devel
--
Justin Deoliveira
OpenGeo - http://opengeo.org
Enterprise support for open source geospatial.
Justin Deoliveira ha scritto:
Hi Mauro,
I have never seen this before no. If you have some sample requests i am happy to try them out though.
I would also be fine to upping the read ahead limit to 8K. Andrea: thoughts?
It seems for some reason we're not able to roll back the already
read portion of the xml and then keep on reading? Maybe Mauro
request exceed the 2k but not the 8k.
That is, is it possible that we can rollback the read only if
the request has been fully read?
Cheers
Andrea
--
Andrea Aime
OpenGeo - http://opengeo.org
Expert service straight from the developers.
The request I used for testing is about 5k, but I usually have bigger ones too. I'm going to verify if the 8k limit is enough for 10k or 15k queries too...
Mauro
Andrea Aime wrote:
Justin Deoliveira ha scritto:
Hi Mauro,
I have never seen this before no. If you have some sample requests i am happy to try them out though.
I would also be fine to upping the read ahead limit to 8K. Andrea: thoughts?
It seems for some reason we're not able to roll back the already
read portion of the xml and then keep on reading? Maybe Mauro
request exceed the 2k but not the 8k.
That is, is it possible that we can rollback the read only if
the request has been fully read?
Cheers
Andrea
10k and 15k queries seem good too. I would suggest to increase the readAheadLimit to 8k. Just an hint: could it be related to the BufferedReader default buffer size, which is 8k too?
Mauro
Andrea Aime wrote:
Justin Deoliveira ha scritto:
Hi Mauro,
I have never seen this before no. If you have some sample requests i am happy to try them out though.
I would also be fine to upping the read ahead limit to 8K. Andrea: thoughts?
It seems for some reason we're not able to roll back the already
read portion of the xml and then keep on reading? Maybe Mauro
request exceed the 2k but not the 8k.
That is, is it possible that we can rollback the read only if
the request has been fully read?
Cheers
Andrea