We have developed a problem with our 3.6.2 installation that seems to be cascading.
Initially, any subnetwork search would fail with no results -- ie, a known item within that subnetwork would not be found by any keyword search.
Shortly thereafter, creating a dynamic subcollection for any dataverse would fail. You can create the query, but the results for that query will be nil. Oddly, however, the previous subcollections which were extant still work.
This weekend we started getting more errors related to indexing:
I've been getting the Index update error notification email, which tells me one study out of our thousands is not being indexed and points me to server.log. Unfortunately, I'm not sure where in the 400+ lines of log I can find out what's going wrong.
To make matters worse, I just started getting the "IO problem" email from our DVN installation.
Again, "1 studies may not have been indexed142.103.160,105".
To make things much, much worse, now *any* search on our installation fails with
We are sorry. An application error has occurred.
Error Message:javax.faces.el.EvaluationException: java.lang.NullPointerException
I tried manually indexing (Utilities/Index/Index unindexed studies). That results in
Indexing update completed
which seems great until I get the IO error update again.
I've attached the server.log entry from the latest indexing attempt.
In addition to that, our server.log is now filled with errors like these:
Caught exception while executing combined colleciton query on VDC 10|#]
Caught exception while executing combined colleciton query on VDC 3|#]
. . .
At this point I haven't tried reindexing everything in case it makes things worse.
I would be eternally grateful for any help you can provide. If anything, how to find the unindexed study and expunge it would probably be a good start.
Paul Lesack
Data Services
Koerner Library
University of British Columbia