Hi. We recently upgraded from h2-1.3.169 to 176, and we experienced corruption issues (we have thousands of installed apps)
The solution was to downgrade to 169 (never gave us problems, used for more than a year). On Thursday, December 18, 2014 at 7:52:37 AM UTC, Marc Heinz wrote: > > > Hi everyone, > > For some years now, we are encountering recurring data corruption in one > of our product using H2. We hoped that the problem will eventually go away > after a clean-up of the database management code and an upgrade of our > libraries (migrating from HSQLDB to H2 , then from H2 1.1.x to version > 1.3.176, currently in use), but so far it didn't helped. > > Only a low percentage of users are affected (less than 0.2%), but they are > generating enough technical support to be a real problem for us, since we > need to manually recover every corrupted database. > > Here are some key facts: > - The database schema is really simple: a few tables with some basic > constraints, one of them containing a collection of BLOBs (PDF files). > - The database is encrypted using the built-in AES cipher. Our connection > string looks like this: jdbc:h2:/path/to/the/db/db_name;CIPHER=AES > secret_key > - Database management is performed through Hibernate (current version: > 4.1.7). > - Analysis of the corrupted DBs showed no obvious pattern: sometimes the > data itself is damaged (causing CRC failures which can be usually ignored > using the built-in recovery tool), sometimes the database indexes > themselves seem to be unusable (random exceptions at various point in > code), with sometime a mix of both. There is no meaningful exception in log > that may indicates that an error condition was reached in the past: we only > see the failed opening attempts. > - Most of the corrupted DBs, but not all of them, are rather big (> 2 Mo, > again, mostly PDF files). > - At least some of the user have stored the DB on a network drive where > they may have been concurrent access attempts. > > I cannot give an example of a corrupted DB file, since we cannot reproduce > the issue, and the kind of data we manage is really sensitive by nature... > > We are currently wondering what could be the most likely cause for such > symptoms: > - Some problem with our database management code? (Unclosed connection > and/or transactions ?) > - Compatibility problem between H2 and the JPA provider (Hibernate) ? > - Some encryption related problem ? > - Concurrent access attempts when stored on a network drive ? > - A bug in the H2 engine ? > > Any pointer or insight would be greatly appreciated, > Thanks > > Marc H. > -- You received this message because you are subscribed to the Google Groups "H2 Database" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/h2-database. For more options, visit https://groups.google.com/d/optout.
