Well, the first thing to do is figure out what the limit is. See: ' The
open-source software that Solr uses to decode the form post probably
adheres to the standards and rejects names that are out of spec.'
Karl
On Thu, Jul 25, 2019 at 5:25 AM wrote:
> Hi Karl,
>
> Concerning this thread, I tried to ask for explanations to the Solr
> mailing list and had no response. So I did a ticket few weeks ago
> https://issues.apache.org/jira/browse/SOLR-13612?filter=-2 but it did not
> have any response or comments either.
>
> I think we will need to handle this problem ourselves in the Solr output
> connector. What do you think ?
>
> Julien
>
> -Message d'origine-
> De : Karl Wright
> Envoyé : mercredi 19 juin 2019 22:45
> À : dev
> Objet : Re: Solr Output Connector - Too big metadata names
>
> Hi Julien,
>
> The Solr connector uses multipart form posting, and this is covered by
> specification pretty completely. I am therefore not surprised that a
> long-named chunk of metadata exceeds some limit. The open-source software
> that Solr uses to decode the form post probably adheres to the standards
> and rejects names that are out of spec.
>
> I think the best way to proceed would be to create a Solr ticket that
> precisely describes what you are posting, maybe with a Curl command, and
> see what the response is. There's nothing much that ManifoldCF can do
> other than truncate names at the limit, once we determine what that is.
>
> Karl
>
>
> On Wed, Jun 19, 2019 at 2:03 PM Julien
> wrote:
>
> > Hi Karl,
> >
> > I recently noticed a problem with the Solr Output connector that
> > triggered for some files an HTTP 500 error with the following
> > description « Header section has more than 10240 bytes (maybe it is not
> properly terminated) ».
> > After several retries the job is aborted.
> >
> > So I did some tests and I figured out that document metadata with a
> > long name are the issue here (in my tests I was able to evaluate the
> > threshold to 8k chars to be safe). And it only concerns metadata name,
> > I did tests on the metadata values and it seems not a problem that a
> > value is really big (more than 160k chars in my tests).
> >
> > As I am not really sure if it is a matter of configuration on Solr
> > side or not, I wonder if it makes sense to implement something to
> > handle this issue in the Solr Output Connector. What do you think ?
> >
> > Regards,
> > Julien
> >
> >
> >
> > ---
> > L'absence de virus dans ce courrier électronique a été vérifiée par le
> > logiciel antivirus Avast.
> > https://www.avast.com/antivirus
> >
>
>