Location of Not available problem
Some versions of C02 and C30 have the problem of monitoring and displaying not available. There are two common reasons for this problem: thread pool exhaustion and no 30s acquisition thread. Both are known problems and can be confirmed by the following means.
Thread pool exhaustion, related problem No. DTS2015011908793
Confirmation steps for thread pool exhaustion:
If there is a problem in the scene environment, it can be directly confirmed by the agent stack information:
A. Use PS command to get the PID of nodeagent, command details: ps-elf | grep NodeAgent
B. Export stack information using jstask command and PID of nodeagent, command details: su-omm; jstack PID > nodeagent_jstack
C. Search for "Collect" using grep Collect nodeagent_jstack. If the result has multiple rows and repeats, indicating that the threads are suspended, it can be identified as a known problem of monitoring thread suspension.
If there is no scene, it can be confirmed by log:
1. Find the node in question and determine the time period for the problem. View the log in the / var / log / Bigdata / nodeagent / monitor / directory as follows
2. Use the grep command to search for the keyword "Process: Host send period_30 resultList" in the monitor. log log of the corresponding time period to see if the search results appear every 30 seconds cycle.
3. If "Process: Host send period_30 resultList" does not occur once in 30 seconds, it may be a monitoring thread management problem that needs further confirmation.
4. Use the grep command to search for the keyword "Period 30 task size" in the log of the corresponding time period, and observe whether the search results appear once in a 30-second cycle.
5. If "task size is" occurs 30 seconds in a cycle, it can be confirmed that it is a known suspension problem of monitoring threads.
No 30 seconds acquisition thread, related issues single number DTS2015040203920
A. Check that the node agent thread stack is normal and no related acquisition tasks are suspended
B. Find the keyword "Process: Host send period_30r" in / var/log/Bigdata/nodeagent/monitor.log, that is, grep "Process: Host send period_30r" monitor.log. If no relevant results are found, the task for 30s is not started.
No 30s acquisition threads, and the environment has been upgraded or uninstalled cluster reinstallation operations, related question number
DTS2015081500934
A. All 30s threads are gone, there is no real-time data, only historical data.
B. Confirm Manager version and query through script
[omm@redhat51~]$/opt/huawei/Bigdata/om-0.0.1/sbin/queryVersion.sh
/opt/huawei/Bigdata/om-0.0.1
Fusion Insight_V100R002C50
Package Version BuildTime
Fusion Insight Fuison Insight Manager V100R002C50 2015-08-2423:35:52
Is this version available before August 14 (including August 14)
C. Has the cluster ever uninstalled the cluster and reinstalled it?
D. Printing in Controller Log
2015-08-27 04:00,766 ERROR [Real Time Task Dispatch Thread] invalid parameter. com. huawei. hadoop. om. controller. rest. resources. ClusterResource. synMaintenance (ClusterResource. java: 2330)
If all the above conditions are satisfied, then it is the problem. Please refer to the Questionnaire Processing.
