Hi Diego, First of all, I'm sorry for the delay in replying.
I went through you configs. Do not use host names in your configurations. Please use IP addresses for now. There is a known issue with Hazelcast. Here is a related JIRA [1]. Use IPs in <parameter name="localMemberHost" /> and <members> sections. Please try this and let us know. Also in your loadbalancer.conf you can specify one host for worker. When you point that hostname to ELB IP, ELB will send the requests using round robin algorithm Thanks! [1] https://wso2.org/jira/browse/CARBON-14573 On Wed, Dec 11, 2013 at 8:35 AM, dseoanel . <dseoan...@gmail.com> wrote: > Hi everybody, and good evening. > > I am working with WSO2 AS versions 5.2 with ELB 2.1 on a fresh > installation in cluster mode, in which we have 1 machine with a ELB, > Manager and Worker1 node, and another different machine with only the > second worker (worker2). > > I followed the instructions at > docs.wso2.org/display/CLUSTER420/Clustering+Application+Server to get it > started and troubleshooted a couple of misconfiguration problems on the > way, but, at last, we have it working and balacend through ELB (thanks > Isuru ;-)). > > Our problem comes when adding the second node. I can't found any > instructions of how to set up the second node on a different machine, and > have read several forums in which (with previous versions of the clustering > agent, tribes, and not Hazelcast) people say that will be enough to copy > the configuration on worker1 to the new machine and have it connected to > the ELB. > > Messing around with the configurations, I get the second worker node > connected to the ELB, but, when I consume a service through the ELB > hostname (or through the hostname I introduce in the Carbon.xml under > <hostname> in the manager node), the second worker node starts to unjoin, > and rejoin constantly, every few seconds, leaving all in a unconsistent > state. > > No network problems detected at all. > > My scenario now is as follows: > - Starting ELB, Manager Node and Worker1 --> Everything's OK > - Starting ELB, Manager Node, Worker1 and Worker2 (in that order) --> > Everythings OK, I can see how the two workers connect to the cluster in the > ELB console, but, in a few minutes, Worker 2 starts to unjoin, and rejoin > automatically the cluster > - Starting ELB, Manager Node and Worker2-->Everything's OK. Like when we > start only the Worker1. > > > The problem seems to be on starting the 2 nodes, because if I start the > two worker nodes, I can consume services right. But, if I stop the Worker1 > (the first worker node I connect to the cluster), the requests stops > working. It seems that the second node I connect doesn't really connect or > be balanced through the ELB (not minding the order)... > > I attach the configuration files (the ones we have touched to have our > cluster working) in the .rar file. The axis2.xml, Carbon, loadbalancer and > catalina_server files on every node of our cluster. > > The worker node (worker2) is the only node that is in a different machine > and IP address. > > I would appreciate any help on this matter...We have been stuck at this > point for almost two weeks and we need to make a proper presentation of the > cluster working to our customer, in order for him to valorate wso2 as his > future standard bus. > > Thank you very much and best regards. > > Diego > > > > _______________________________________________ > Dev mailing list > Dev@wso2.org > http://wso2.org/cgi-bin/mailman/listinfo/dev > > -- Isuru Perera Senior Software Engineer | WSO2, Inc. | http://wso2.com/ Lean . Enterprise . Middleware about.me/chrishantha
_______________________________________________ Dev mailing list Dev@wso2.org http://wso2.org/cgi-bin/mailman/listinfo/dev
