1. Zookeeper+ metaq-1.4.6 and storm-0.9.4 programs appear in eclipse running
Will not attempt to authenticate using SASL (java.lang.SecurityException: Unable to locate login configuration) 
Use local mode:
LocalCluster cluster = new LocalCluster();
cluster.submitTopology("test", config, tbuilder.createTopology());
Running doesn't matter.
2.worker-xxxx.log 显示 2015-04-30 08:31:53 worker [WARN] Received invalid
messages for unknown tasks. Dropping... cause data loss problems
The problem is the host configuration problem, there is a problem parsing the host
solution:
Execute hostname pc-name on each cluster machine (indicates that pc-name is the defined cluster name)
vi /etc/sysconfig/network
Modify HOSTNAME=pc-name
The effect is as follows:
vi /etc/hosts
Add the following information:
3. Both nimbus and supervisor on StormUI show 127.0.0.1 or localhost
The solution is the same as 2
4. After topologykill, kill supervisor or nimbus, start immediately, sometimes shutdown automatically
After the supervisor and nimbus are started and shut down, the information on the zookeeper will be synchronized. Every time it starts, it tries to resume and stop the last task. For the supervisor, you can wait until the topology on StormUI is completely stopped before restarting to avoid this problem.
5. How to find the running log of a spout and bolt?
Find the topology on the Storm UI, click in and find the target (spout or bolt, etc.), click in, find the Executor column, which contains Host and Port, which is actually the worker that executes the component's task. Each port will correspond to a log file of worker-<port>.log in the logs directory, and you can directly go to the host to find the file and view it.
Note:
Storm has a logviewer function, that is, after starting ./storm logviewer on each superviosr, you can directly view the static log after clicking the port in stormUI.
If it cannot be opened, check logs/logviewer.log on the supervisor to see if there is any error. Then check whether the host name in the opened URL can be correctly resolved to the target IP locally. If not, change the hostname to IP to view (only suitable for nimbus in the cluster environment).
The effect of modifying nimbus to the IP of the corresponding PC is as follows:
