lock acquisition failed - pessimistic locking - read_uncommi

[haribaasha]

Hi
i had posted a locking issue previously http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4117973#4117973

this was my mistake, there was a synchronization problem while cache creation and there were 2 instances running !

now i have fixed and upgraded to jboss cache 2.0.0. yesterday i got the following lock issue -

10:52:55,679 INFO [WorkerThread#3[192.168.20.157:54241]] Exception while trying to write into cache : failure acquiring lock: fqn=/Indicatives/SAIL04035, caller=GlobalTransaction:<192.168.20.134:51000>:838, lock=write owner=GlobalTransaction:<192.168.20.134:51000>:837 (org.jboss.cache.lock.LockStrategyReadUncommitted@edeb72)
10:52:55,679 INFO [WorkerThread#4[192.168.20.157:54246]] Exception while trying to write into cache : failure acquiring lock: fqn=/Indicatives/SAIL04034, caller=GlobalTransaction:<192.168.20.134:51000>:837, lock=write owner=GlobalTransaction:<192.168.20.134:51000>:838 (org.jboss.cache.lock.LockStrategyReadUncommitted@1011edf)


i can see that both threads have accessed at the same time. this is the drawback of pessimistic locking and i should move to optimistic locking ?

[mircea.markus]

looks like the classic deadlock scenario:

Tx838 tries to acquire WL on /Indicatives/SAIL04035 - WL owned by Tx837
Tx837 tries to acquire WL on /Indicatives/SAIL04034 - WL owned by Tx838

If this is a frequently used cache accessed pattern, it worths trying optimistic locking (in the prev scenario, with optimistic locking, at least one transaction would had succeeded)

Take a look here for some information on the drivers behind optimistic locking: http://wiki.jboss.org/wiki/Wiki.jsp?page=OptimisticNodeLocking