Re: How to deregister Lateral Cache listeners in cluster?
Aaron Smuts <[email protected]> Thu, 9 Jul 2009 06:33:41 -0700 (PDT)
| Newsgroups | gmane.comp.jakarta.turbine.jcs.devel |
|---|---|
| Message-ID | <[email protected]> |
The discovery system needs to deregister after a period of inactivity. =0A= =0AThe remote cache client queues items during the reconnect period. It wi= ll send them once reconnected. This isn't as easy to do in the lateral. = =0A=0AAaron=0A=0A--- On Mon, 7/6/09, Andrei Povodyrev <povodyrev@nodalexcha= nge.com> wrote:=0A=0A> From: Andrei Povodyrev <[email protected]>= =0A> Subject: How to deregister Lateral Cache listeners in cluster?=0A> To:= [email protected]=0A> Date: Monday, July 6, 2009, 11:10 PM=0A> = =0A> =0A> =0A> I am using Lateral Cache with Lateral Cache with UDP=0A> dis= covery. When a node=0A> in cluster goes down the other nodes do not discard= their=0A> senders=0A> (LateralTCPSender) for the node. Is there way to rem= ove=0A> them?=0A> Calling CompositeCacheManager.shutDown() on the node does= =0A> not help=0A> =0A> I am wondering why this behavior was programmed this= way.=0A> Would it be better=0A> to recreate the senders/listeners when the= node again joins=0A> the cluster via=0A> UDP discovery?=0A> =0A> The curre= nt implementation presents me with a risk of=0A> losing cache events as=0A>= the object stream oos in LateralTCPSender is closed anyway=0A> and first a= ttempt=0A> to send fails (EVEN when the node is back).=0A> After failure is= detected the sender is recreated by=0A> LateralCacheMonitor=0A> which=0A> = =A0 *operates in a failure driven mode. That is,=0A> * it goes into a wait= state until there is an error. Upon=0A> the notification=0A> of a=0A> * c= onnection error, the monitor changes to operate in a=0A> time driven mode.= =0A> That=0A> * is, it attempts to recover the connections on a periodic= =0A> basis. When all=0A> * failed connections are restored, it changes bac= k to the=0A> failure driven=0A> mode=0A> =0A> All consecutive sends will be= fine but the first one will=0A> not get through. I=0A> consider this to be= a SEVERE case.=0A> =0A> Is there any trick I missed?=0A> =0A> Thank you ve= ry much for your help=0A> -- =0A> View this message in context: http://www.= nabble.com/How-to-deregister-Lateral-Cache-listeners-in-cluster--tp24367699= p24367699.html=0A> Sent from the JCS - Dev mailing list archive at=0A> Nabb= le.com.=0A> =0A> =0A> -----------------------------------------------------= ----------------=0A> To unsubscribe, e-mail: [email protected]= ache.org=0A> For additional commands, e-mail: [email protected]= rg=0A> =0A>