I'm putting this info here for future reference, since it has taken me a couple of hours to solve.
My users of an Umbraco 4.0.4.1 installation got a YSOD on publishing (not saving) a page. Upon trying the same on the server itself I got a stacktrace starting with "IOException: read past EOF" for Lucene.
First, I republished the entire site, recreating the umbraco.config file under ~/data
Second, I removed the Lucene index (all files under ~/data/_systemUmbracoIndexDontDelete) and went to ~/umbraco/reindex.aspx (in the webbrowser of course). This recreated the Lucene index (I watched the aforementioned folder until activity stopped).
These actions did not solve the problem. However, Lucene did not throw an error reindexing the entire site, so I was a bit perplexed.
To experiment a bit more, I zipped up the entire website to copy it to a dev machine, when I noticed a huge number of "_1jg.f25" kind of files flying by in the zipping dialogue window. Wait a sec, I just reindexed Lucene and I know there were only a couple of index files left, so where do these other index files come from?
Turns out UmbSearch2 was installed as well (no, I'm not the only developer for this site). I figured it uses Lucene as well and it might hook into the publishing events just the same (as the built-in backend Lucene indexer).
Since it was installed under ~/usercontrols/UmbSearch2 I opened (from the server itself, in a webbrowser) the URL ~/usercontrols/UmbSearch2/ReindexSite.aspx
After clicking the "Start" button on the reindexing page, I saw the indexing files in ~/usercontrols/UmbSearch2/UmbSearch2IndexDontDelete were recreated.
I tried publishing a page again and boom! (yes I have watched one too many of Jobs' keynotes) the page published fine.
Since I've been searching the entire Internets and some more, I thought I'd share this with you, although the current version of Umbraco probably won't exhibit the same problem.
"IOException: read past EOF" Lucene - solved
I'm putting this info here for future reference, since it has taken me a couple of hours to solve.
My users of an Umbraco 4.0.4.1 installation got a YSOD on publishing (not saving) a page. Upon trying the same on the server itself I got a stacktrace starting with "IOException: read past EOF" for Lucene.
First, I republished the entire site, recreating the umbraco.config file under ~/data
Second, I removed the Lucene index (all files under ~/data/_systemUmbracoIndexDontDelete) and went to ~/umbraco/reindex.aspx (in the webbrowser of course). This recreated the Lucene index (I watched the aforementioned folder until activity stopped).
These actions did not solve the problem. However, Lucene did not throw an error reindexing the entire site, so I was a bit perplexed.
To experiment a bit more, I zipped up the entire website to copy it to a dev machine, when I noticed a huge number of "_1jg.f25" kind of files flying by in the zipping dialogue window. Wait a sec, I just reindexed Lucene and I know there were only a couple of index files left, so where do these other index files come from?
Turns out UmbSearch2 was installed as well (no, I'm not the only developer for this site). I figured it uses Lucene as well and it might hook into the publishing events just the same (as the built-in backend Lucene indexer).
Since it was installed under ~/usercontrols/UmbSearch2 I opened (from the server itself, in a webbrowser) the URL ~/usercontrols/UmbSearch2/ReindexSite.aspx
After clicking the "Start" button on the reindexing page, I saw the indexing files in ~/usercontrols/UmbSearch2/UmbSearch2IndexDontDelete were recreated.
I tried publishing a page again and boom! (yes I have watched one too many of Jobs' keynotes) the page published fine.
Since I've been searching the entire Internets and some more, I thought I'd share this with you, although the current version of Umbraco probably won't exhibit the same problem.
is working on a reply...