On 26 Apr 2012, at 14:42, Erik Salter wrote: > FWIW, I implement a TryLock like this: > > boolean success = cache.getAdvancedCache() >            .withFlags(Flag.FAIL_SILENTLY).lock(key); +1, this seems like solving the issue. > > Regards, > > Erik > > -----Original Message----- > From: infinispan-dev-bounces(a)lists.jboss.org > [mailto:infinispan-dev-bounces@lists.jboss.org] On Behalf Of Galder > Zamarreño > Sent: Thursday, April 26, 2012 6:59 AM > To: Mircea Markus > Cc: infinispan -Dev List > Subject: [infinispan-dev] Time for a tryLock() ? > > Looks like rolling back the transaction when a lock timeout is encountered > can be problematic: https://community.jboss.org/message/731307#731307 > > Maybe time to implement a tryLock() that attempts to acquire the lock but > does not rollback the transaction if it cannot acquire it? > > Thoughts? > -- > Galder Zamarreño > Sr. Software Engineer > Infinispan, JBoss Cache > > > _______________________________________________ > infinispan-dev mailing list > infinispan-dev(a)lists.jboss.org > https://lists.jboss.org/mailman/listinfo/infinispan-dev > _______________________________________________ infinispan-dev mailing list infinispan-dev(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/infinispan-dev

On 26 Apr 2012, at 13:05, Dan Berindei wrote: > What seems to be missing is an overload with a custom timeout, it may > be useful to try locking with a shorter timeout in the first attempt > (maybe even 0?) like we do for deadlock detection. There is a flag for that.

> > There is a flag for that. > > https://github.com/infinispan/infinispan/blob/master/core/src/main/java/o... > Are we sure ZERO_LOCK_ACQUISITION_TIMEOUT is the best way to name this flag ? Maybe that is why people can not find it :-(

On 26 Apr 2012, at 13:05, Dan Berindei wrote: What seems to be missing is an overload with a custom timeout, it may be useful to try locking with a shorter timeout in the first attempt (maybe even 0?) like we do for deadlock detection. There is a flag for that. https://github.com/infinispan/infinispan/blob/master/core/src/main/java/o...

But that is a bit orthogonal, as it would be useful for write operations as well, so maybe a 6.0 idea... On Thu, Apr 26, 2012 at 2:55 PM, Mircea Markus <mircea.markus(a)jboss.com&gt; wrote: On 26 Apr 2012, at 14:42, Erik Salter wrote: FWIW, I implement a TryLock like this: boolean success = cache.getAdvancedCache()            .withFlags(Flag.FAIL_SILENTLY).lock(key); +1, this seems like solving the issue. Regards, Erik -----Original Message----- From: infinispan-dev-bounces(a)lists.jboss.org [mailto:infinispan-dev-bounces@lists.jboss.org] On Behalf Of Galder Zamarreño Sent: Thursday, April 26, 2012 6:59 AM To: Mircea Markus Cc: infinispan -Dev List Subject: [infinispan-dev] Time for a tryLock() ? Looks like rolling back the transaction when a lock timeout is encountered can be problematic: https://community.jboss.org/message/731307#731307 Maybe time to implement a tryLock() that attempts to acquire the lock but does not rollback the transaction if it cannot acquire it? Thoughts? -- Galder Zamarreño Sr. Software Engineer Infinispan, JBoss Cache _______________________________________________ infinispan-dev mailing list infinispan-dev(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/infinispan-dev _______________________________________________ infinispan-dev mailing list infinispan-dev(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/infinispan-dev _______________________________________________ infinispan-dev mailing list infinispan-dev(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/infinispan-dev -- Manik Surtani manik(a)jboss.org twitter.com/maniksurtani Lead, Infinispan http://www.infinispan.org _______________________________________________ infinispan-dev mailing list infinispan-dev(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/infinispan-dev

On 26 Apr 2012, at 18:29, Dan Berindei wrote: > On Thu, Apr 26, 2012 at 3:09 PM, Manik Surtani <manik(a)jboss.org&gt; wrote: >> >> On 26 Apr 2012, at 13:05, Dan Berindei wrote: >> >> What seems to be missing is an overload with a custom timeout, it may >> be useful to try locking with a shorter timeout in the first attempt >> (maybe even 0?) like we do for deadlock detection. >> >> >> There is a flag for that. >> >> https://github.com/infinispan/infinispan/blob/master/core/src/main/java/o... >> > > Forgot about that one :) > But still, it only allows a 0 timeout, so it's not exactly what I had > in mind - which would be a custom timeout per operation. Flags are Enum and not parametrizable, I guess we'd need to enhance the API to allow associating a param with a Flag as well.

I wasn't thinking of a flag, I was thinking of a withTimeout(timeout) method on AdvancedCache. There may be other "flags" that could be customized with a value as well, but then you'd need to keep them in a map and it could be more trouble than it's worth. The complication with withTimeout/tryLock OTOH is that as a user I'd expect it to limit everything, not just the lock acquisition - so it would have to override sync.replTimeout as well.