On 31 Mar 2011, at 12:53, Vidar Ramdal wrote: > On Wed, Mar 30, 2011 at 2:27 PM, John Langley <[email protected]> wrote: >> I'd be interested in hearing from people been using Amazon's RDS _and_ been >> using jackrabbit clustering with it. >> >> I ran an experiment a week or so ago where I could consistently cause a >> mysql deadlock under load. However, my current hypothesis is that this is >> due to a mysql configuration issue (it was an innodb lock, although I don't >> have the details of it to share right now although I can tell you it was on >> an index for one of the version tables). >> >> So, I'd be interested in finding out if other people have done this already. >> My code base was 2.2.5 and so has many of the recent lock/deadlock fixes in >> it. > > We are doing this, with Jackrabbit 1.6.3. It works mostly fine, but we > have seen some SQL Timeout exceptions, related to locking the > revisions table, when doing workspace.import of large dumps.
have a read of http://dev.mysql.com/doc/refman/5.1/en/innodb-record-level-locks.html the deadlocks may be due to gap locking. HTH Ian > > -- > Vidar S. Ramdal <[email protected]> - http://www.idium.no > Sommerrogata 13-15, N-0255 Oslo, Norway > + 47 22 00 84 00 > Quando omni flunkus moritatus!
