Applies to:
TimesTen Data Server - Version: 5.1.27 and later [Release: and later ]Linux x86-64
Goal
A flux is a form. of semaphore used to serialize certain operations on a TimesTen data store, such as connect, disconnect, recover, load and unload. When waiting for one a message containing 'Waiting for in-flux to clear' will be written to the log. If these waits are causing a problem, this article describes what information to collect in order to assist Oracle TimesTen support in diagnosing the problem.
Solution
This information needs to be collected whilst the problem is ocurring, though the message logs can be collected afterwards, as long as they still exist.
A UNIX or Linux platform. is assumed where relevant.
- Output of ttXactAdmin -latch
- If the above shows any PID holding a latch, pstack or gdb trace output of the process(es).
- Output of pstack or gdb run on the PID mentioned in the 'Waiting for in-flux' message.
- Output of pstack or gdb run on the main TimesTen daemon, whose PID is reported by ttStatus
- Output of ttStatus -v or ttStatus -debug
- ttmesg.log messages covering the period when the in-flux messages were seen.
- Use the OS utiity "top" to check for processes using lot of CPU on the system.
来自 “ ITPUB博客 ” ，链接：http://blog.itpub.net/17252115/viewspace-749037/，如需转载，请注明出处，否则将追究法律责任。
转载于:http://blog.itpub.net/17252115/viewspace-749037/
扫一扫
117
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
