Yes, that's on for all members in the cluster (and it's the only way I set up sessions for every server I touch).
We think we've resolved the problem of members dropping out of the cluster. It appears that two of the boxes were set up for IPV6 and two weren't. Getting rid of the IPV6 config seems to have made them behave and they all respond to the discovery multicasts properly now. The problem of sessions giving a null pointer error on failover persists if we enable session replication. We're running the 1.5 JVM on these servers; at some stage I'll set up a test server with Java 6 to see if that's the cause. mxAjax / CFAjax docs and other useful articles: http://www.bifrost.com.au/blog/ 2008/10/9 Shannon Peevey <[EMAIL PROTECTED]> > On 1): have you checked "* Use J2EE session variables * "? You have to use > J2EE session variables in a clustered environment.. > > On Thu, Oct 2, 2008 at 12:56 AM, James Holmes <[EMAIL PROTECTED] > >wrote: > > > We're hoping to set up some monitoring to work that out. If there's a > > verbose log level I can turn on at the server end it might help. I'll dig > > around for that. > > mxAjax / CFAjax docs and other useful articles: > > http://www.bifrost.com.au/blog/ > > > > > > 2008/10/2 AJ Mercer <[EMAIL PROTECTED]> > > > > > for point 2there maybe some timeout setting on the network - if it cant > > see > > > it for a period of time it may think the cluster is dead. > > > > > > Is the newtowrk under load when it drops out? > > > > > > > > > On Thu, Oct 2, 2008 at 2:28 PM, James Holmes <[EMAIL PROTECTED]> > > > wrote: > > > > > > > We have two problems with clustering our CF 8.01 Multiserver (i.e. > > JRun) > > > > installs. > > > > > > > > 1) Session replication is broken. When we turn off an instance in a > > > cluster > > > > that has session replication and stickiness turned on, the next > server > > > that > > > > gets the request throws a null pointer error when trying to run > > > > Application.cfc to establish the session. We've confirmed this by > > > clearing > > > > the session cookies on the browser; this enables the server to > respond > > > > correctly as it can then set up its own session. Turning off session > > > > replication makes this go away. Any ideas what might be wrong with > our > > > > setup > > > > to cause session replication to fail? > > > > > > > > 2) Half way through yesterday, our clustering started to fail. Up > until > > > now > > > > all the servers have behaved properly, all joining the cluster > > correctly. > > > > Now, servers will drop out of the cluster for no reason, only to > > re-join > > > > later on. They intermittently drop in and out of the cluster. The > > > > most interesting part is that when they drop out, they do it in > pairs. > > > Each > > > > pair of servers in the same datacentre stay clustered; they seem to > > lose > > > > the > > > > connection with the servers in the other datacentre (we have four > > > servers, > > > > two in each datacentre, all on the same VLAN subnet). I've tried > adding > > > the > > > > servers as unicast peers rather than letting the multicast discovery > > find > > > > them, but it didn't help. If anyone has ideas about how best to > monitor > > > the > > > > connections between the servers and get an idea of why the servers > are > > > > dropping out of the cluster, please let me know. > > > > > > > > > > > > > > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~| Adobe® ColdFusion® 8 software 8 is the most important and dramatic release to date Get the Free Trial http://ad.doubleclick.net/clk;207172674;29440083;f Archive: http://www.houseoffusion.com/groups/cf-talk/message.cfm/messageid:313647 Subscription: http://www.houseoffusion.com/groups/cf-talk/subscribe.cfm Unsubscribe: http://www.houseoffusion.com/cf_lists/unsubscribe.cfm?user=11502.10531.4
