I discovered a few days OS Watcher, a tool of the "center of expertise" of Oracle, because I had to use it to investigate a problem of memory allocation. Actually I was requested by Oracle support, because I personally had no doubts about performance, but only about a potential memory leak . The Watcher doubt is that OS has become a bit 'that the tool is run to ask anyone to support, regardless of his specific problem. OS Watcher is a collection of shell scripts that use Unix tools to monitor and record what happens to the operating system level, in order to help the DBA to identify any problems that are beyond and between the database and 'hardware.
Installation is simple: just unzip the downloaded osw3b.tar MOS, enter the directory OSW and start collecting data (example):
$ nohup. / startOSW.sh 60 10
where the first parameter is the sampling time in seconds, and the second parameter is the retention policy of data in hours. The command line that I quoted and then sets the sampling once per minute and remove the files of the sampled data older than 10 hours. It does not need the trailing "&" because the script makes the prompt return immediately. nohup is to avoid that, out of the shell, the main process is halted.
To stop the monitor, run:
$. / StopOSW.sh
The script automatically detects the presence and accessibility of the Unix utility to control the system as vmstat and top, create a directory and archive will store the data.
Since the last version of 2009 OS Watcher is outsourced with OSWg, Java utility that graphically displays the data recorded.
OS Watcher is still useful to all, regardless of the use of the Oracle database.
See also Note 301137.1 MOS
0 comments:
Post a Comment