On Thu, Oct 21, 2010 at 18:15, [Ricardo Rodriguez] eBioTIC. <ricardo.rodrig...@ebiotic.net> wrote: > > > Thomas Mortagne wrote: >> On Thu, Oct 21, 2010 at 16:56, Thomas Mortagne >> <thomas.morta...@xwiki.com> wrote: >> >>> On Wed, Oct 20, 2010 at 15:10, [Ricardo Rodriguez] eBioTIC. >>> <ricardo.rodrig...@ebiotic.net> wrote: >>> >>>> An update, please, see below... >>>> >>>> [Ricardo Rodriguez] eBioTIC. wrote: >>>> >>>>> Thomas Mortagne wrote: >>>>> >>>>> >>>>>> On Tue, Oct 19, 2010 at 21:11, [Ricardo Rodriguez] eBioTIC. >>>>>> <ricardo.rodrig...@ebiotic.net> wrote: >>>>>> >>>>>> >>>>>> >>>>>>> Hi! >>>>>>> >>>>>>> In a XE/XEM 2.4.1 farm, our Lucene building index process stopped as ... >>>>>>> >>>>>>> *Lucene is currently building its index, 41 documents in queue. >>>>>>> >>>>>>> *And we can't access Search in XWiki Preferences of any of the subwikis >>>>>>> or even the controller.* >>>>>>> * >>>>>>> I see this in xwiki.log: >>>>>>> >>>>>>> com.xpn.xwiki.XWikiException: Error number 3301 in 3: Exception while >>>>>>> switching to database redpatrimonio >>>>>>> >>>>>>> >>>>>>> >>>>>> The only way to have this error I know is to have a >>>>>> XWiki.XWikiServerRedpatrimonio page with a wiki descriptor object and >>>>>> no corresponding database. >>>>>> >>>>>> >>>>>> >>>>>> >>>>>>> Wrapped Exception: Unknown database 'redpatrimonio' >>>>>>> at >>>>>>> com.xpn.xwiki.store.XWikiHibernateBaseStore.setDatabase(XWikiHibernateBaseStore.java:643) >>>>>>> at >>>>>>> com.xpn.xwiki.store.XWikiHibernateBaseStore.beginTransaction(XWikiHibernateBaseStore.java:742) >>>>>>> at >>>>>>> com.xpn.xwiki.store.XWikiHibernateStore.loadXWikiDoc(XWikiHibernateStore.java:683) >>>>>>> at >>>>>>> com.xpn.xwiki.store.XWikiCacheStore.loadXWikiDoc(XWikiCacheStore.java:289) >>>>>>> at com.xpn.xwiki.XWiki.getDocument(XWiki.java:1463) >>>>>>> at com.xpn.xwiki.XWiki.getDocument(XWiki.java:1506) >>>>>>> at com.xpn.xwiki.api.XWiki.getDocument(XWiki.java:166) >>>>>>> at com.xpn.xwiki.api.XWiki.getDocument(XWiki.java:153) >>>>>>> at >>>>>>> com.xpn.xwiki.plugin.lucene.SearchResult.<init>(SearchResult.java:98) >>>>>>> at >>>>>>> com.xpn.xwiki.plugin.lucene.SearchResults.getRelevantResults(SearchResults.java:72) >>>>>>> at >>>>>>> com.xpn.xwiki.plugin.lucene.SearchResults.getHitcount(SearchResults.java:22 >>>>>>> >>>>>>> >>>>>>> In fact, that database is called redepatrimonio, not redpatrimonio. >>>>>>> >>>>>>> The wiki descriptor page name in the main wiki is >>>>>>> XWiki.XWikiServerRedepatrimonio. >>>>>>> >>>>>>> >>>>>>> >>>>>> Are you sure you don't have XWiki.XWikiServerRedpatrimonio too. >>>>>> >>>>>> >>>>>> >>>>> I think I haven't it. Please, see... >>>>> >>>>> http://ebiotic.net/bin/ICT/Subwikis >>>>> >>>>> >>>>>> >>>>>>> Please, where the name of the database as listed in the exception is >>>>>>> stored? How could I correct it? >>>>>>> >>>>>>> I don't know how I've messed things and if this error is related with >>>>>>> problem, but it seems that need to be corrected before trying anything >>>>>>> else! >>>>>>> >>>>>>> >>>>>>> >>>>>> In any case, it's also a Lucene plugin bug since a ghost wiki should >>>>>> not break it, would be nice if you could report it on >>>>>> http://jira.xwiki.org. >>>>>> >>>>>> >>>>> I'm doing some more research. I'll create a JIRA issue with the results. >>>>> >>>>> Thanks! >>>>> >>>>> Ricardo >>>>> >>>>> >>>>> >>>> There were some corrupted attachments here that were throwing >>>> exceptions. I guess this has nothing to do with Lucene stopping >>>> >>> Yes corrupted attachments can't break Lucene indexing AFAIK. >>> >>> >>>> indexing, but I'm trying to "clean up" catalina.out to find the reason >>>> why Lucene is failing. >>>> >>>> Only one attachment not identified as "broken attachment" by Raluca's >>>> script >>>> (http://code.xwiki.org/xwiki/bin/view/Snippets/AllBrokenAttachments), in >>>> fact, it downloads without a glitch, keeps throwing an exception while >>>> indexing: >>>> >>>> ricardorodrig...@peat:/var/log/tomcat5/base> cat catalina.out |grep >>>> 'IML.Acreditacion' >>>> 2010-10-20 13:15:22,501 [Lucene Index Updater] WARN >>>> lucene.AttachmentData - error getting content of attachment >>>> [IsoIec17025-1999IRam301-00NormaArgentina.pdf] for document >>>> [gmx:IML.Acreditacion] >>>> ricardorodrig...@peat:/var/log/tomcat5/base> >>>> >>>> >>> That weird because this error is because storage returned an error to >>> Lucene plugin. >>> >>> >>>> I'll delete it, change its name and try again. >>>> >>>> I'll keep cleaning errors and keep this thread posted. >>>> >>>> BTW: I've not seen the error related with the Redpatrimonio ghost >>>> subwiki any more... :-( >>>> >> >> Maybe you recreated the Lucene index as i suggested in another before >> seeing this one. >> >> > > I've recreated the index several times (completely removed > /tmp/xwiki/lucene folder and restarted the servlet container) before and > after deleting broken attachments. I've also removed those documents > reported by Lucene as having problems with them. Three PDF files and > three Excel xlsb files. I can download them from the wiki and open them > without problems. Now, no more "error getting content of attachment" > errors, but I got many warns. I've started a different thread when I've > posted a link to a short catalina.out file. > > It is available here (3,6Mb): > > http://ftp.ebiotic.net/fromPeat/catalina.out.201010211227.txt
Things like 2010-10-20 22:23:51,538 [Lucene Index Updater] WARN font.PDFontFactory - Failed to create Type1C font. Falling back to Type1 font (or anytime you see "org.apache.tika.parser" in the stack trace of the error) are not directly related to Lucence, it means that apache tika (tika.apache.org) failed to read you file either because it's wrong or because there is a bug in tika. But that can't break Lucene. We use tika to find content of attachments and i guess it's not perfect and there is files it does not support. One thing you could do if you are sure your file is correct and you can read it with you usual viewer is sending report it to them with the file (if it's not a security issue of course). Or at least creat a jira issue with each file (when they are very different) on jira.xwiki.org some that someone from XWiki team can take care of this with apache tika. Java taking 100% of CPU is certainly not normal unless you just started to reindex Lucene and your wiki is very big (not really because of lucene but because loading all the document from the database can take a lot of resources). If it's not that it could probably had incidence on Lucene plugin like on anything else of course. > > The new thread is this one... http://markmail.org/message/gqnyxq7ivevka2l5 > > I don't know if both things are related: Lucene not working and Java > using 100% processor, but there are a lot of Lucene related warnings in > catalina.out. >>>> Any idea will be welcome! Thanks! >>>> >>>> Ricardo >>>> >>>> -- >>>> Ricardo Rodríguez >>>> CTO >>>> eBioTIC. >>>> Life Sciences, Data Modeling and Information Management Systems >>>> >>>> _______________________________________________ >>>> users mailing list >>>> users@xwiki.org >>>> http://lists.xwiki.org/mailman/listinfo/users >>>> >>>> >>> >>> -- >>> Thomas Mortagne >>> >>> >> >> >> >> > Thanks! > > -- > Ricardo Rodríguez > CTO > eBioTIC. > Life Sciences, Data Modeling and Information Management Systems > > _______________________________________________ > users mailing list > users@xwiki.org > http://lists.xwiki.org/mailman/listinfo/users > -- Thomas Mortagne _______________________________________________ users mailing list users@xwiki.org http://lists.xwiki.org/mailman/listinfo/users