From: Patrick B. <pat...@jo...> - 2011-11-04 18:05:14
|
It seems to have made a difference to me. I was getting deadlocks during my tests very frequently, and since I made this change, and a few of the others I've submitted, I'm no longer getting them. On Fri, Nov 4, 2011 at 6:03 AM, Dmitriy Shabanov <sha...@gm...>wrote: > Did this change fix issue for you? release the lock that wasn't acquire is > just a warning message at the log. > > > On Fri, Nov 4, 2011 at 1:03 AM, Patrick Bosek <pat...@jo...>wrote: > >> >> In my stress testing I've found a deadlock issue that in the XQuery >> Update code, it's in Modification.selectAndLock(..) on line 147. The >> problem was that a READ_LOCK was being acquired on the BrokerPool inside of >> the try/finally that removed it. So, in the case the acquire failed and >> threw an error, the release would pop off the wrong READ_LOCK. >> > > > -- > Dmitriy Shabanov > -- Patrick Bosek Jorsek Software Cell (585) 820 9634 Office (877) 492 2960 Jorsek.com |