Hi Ming, It's been a long time since I looked at the region stuff in hedwig, but I think it could be that you don't seem to be setting the region identifier in hw_server.conf. You need to change "region" in hw_server to some identifier, like reg1 and reg2 for your example.
Hope this helps, Ivan On Thu, Jan 8, 2015 at 4:19 AM, Ming Chen <[email protected]> wrote: > Hi there, > > I am new to Hedwig, and have a question about setting up Hedwig with > two regions. I have successfully set up two separate single-region > instances of Hedwig. But I could not get them communicate with each > other. > > Here are my configuration: > > I have two physical machines (A and B), each hosting three VMs > mimicking a cluster in a region. In each region, the three VMs run > ZooKeeper servers, BookKeeper bookie servers, and Hedwig servers at > the same time. > > For "reg1" (machine A), the three VMs's IP are: > X.Y.Z.111 > X.Y.Z.112 > X.Y.Z.113 > > I successfully set them up as a single-region Hedwig instance. I could > run "bin/hedwig console", and successfully executed commands including > "show", "sub", "pub", "pubsub", "readtopic", "describe topic" etc. > > I did the same for "reg2" (machine B), where the three VMs are: > X.Y.Z.114 > X.Y.Z.115 > X.Y.Z.116 > > Again, "reg2" as a single-region Hedwig instance was running well. > > Then, I began to setup the across-region communication of the two > Hedwig instances. I shut down the two Hedwig instances, and set the > "regions" option in hw_server.conf of all 6 VMs to > > regions=X.Y.Z.111:4080 X.Y.Z.114:4080 > > All options, other than zk_host, region, and regions, in > hw_server.conf are left as default. After that, I restarted the two > Hedwig instances. I logged into the console of "reg1" and published > some test messages in a topic. But I could not observe the topic or > messages when I logged into console of "reg2". The same is true vice > versa. > > It seemed that the two regions can find each other because I observed > that reg1 contains hub(s) from reg2, and vice versa: > > [hedwig: (reg1) 24] show hubs > Available Hub Servers: > X.Y.Z.111:4080:9876 : info : [hostname: > "X.Y.Z.111:4080:9876", czxid: 12884902049], load : [numTopics: 0] > X.Y.Z.116:4080:9876 : info : [hostname: > "X.Y.Z.116:4080:9876", czxid: 12884902050], load : [numTopics: 0] > X.Y.Z.112:4080:9876 : info : [hostname: > "X.Y.Z.112:4080:9876", czxid: 12884902051], load : [numTopics: 1] > > [hedwig: (reg2) 8] show hubs > Available Hub Servers: > X.Y.Z.116:4080:9876 : info : [hostname: > "X.Y.Z.116:4080:9876", czxid: 34359738415], load : [numTopics: 0] > X.Y.Z.114:4080:9876 : info : [hostname: > "X.Y.Z.114:4080:9876", czxid: 34359738416], load : [numTopics: 0] > X.Y.Z.112:4080:9876 : info : [hostname: > "X.Y.Z.112:4080:9876", czxid: 34359738417], load : [numTopics: 0] > > Another thing I observed is that the "sub" command sometimes hung in > this case. But, I did not find any error message in the log file. > > Am I doing something wrong? Did I set the "regions" option correctly? > Any help is > deeply appreciated. > > Thanks, > Ming >
