OSwatcher作为Oracle官方推荐的OS层面运行状态检测的脚本工具。在Exadata是默认已经安装。
但是Exadata是如何在系统启动后，自动启动OSwatcher呢？我们如何去修改OSwatcher的参数，来调整监控和日志保存的策略呢？
本文正是介绍，从系统启动到OSwatcher运行，中间经历过的脚本调用，以及如何修改OSwatcher参数的。
1. 首先检查rc.local文件，可以发现/etc/rc.d/rc.Oracle.Exadata
# vi /etc/rc.d/rc.local ----------------------------------------- ########### BEGIN DO NOT REMOVE Added by Oracle Exadata ########### if [ -x /etc/rc.d/rc.Oracle.Exadata ]; then . /etc/rc.d/rc.Oracle.Exadata <<<<<<<<<<<<<<<<<<<<<<This script will be run automaticlly, when the OS starts fi ########### END DO NOT REMOVE Added by Oracle Exadata ########### -----------------------------------------
2. 查看rc.Oracle.Exadata，找到/opt/oracle.cellos/vldrun -all
# vi /etc/rc.d/rc.Oracle.Exadata ----------------------------------------- # Perform validations step /opt/oracle.cellos/vldrun -all <<<<<<<<<<<<<<<<<<<<<<This script will be run automaticlly, when the OS starts -----------------------------------------
3. 检查当前OSwatcher的设定，每15秒收集一次，生成的日志保存168小时（7天），bzip2的压缩模式，最大日志尺寸是3G
# ps -ef | grep OSW root 15962 1 0 04:00 pts/1 00:00:00 /bin/ksh ./OSWatcher.sh 15 168 bzip2 3 root 15994 15962 0 04:00 pts/1 00:00:00 /bin/ksh ./OSWatcherFM.sh 168 3 root 16272 9529 0 04:00 pts/1 00:00:00 grep OSW
4. 脚本/opt/oracle.cellos/vldrun会调用oswatcher脚本来启动oswatcher
# ls -al /opt/oracle.cellos/validations/init.d/oswatcher -r-xr-x--- 1 root root 5128 Aug 19 03:39 oswatcher # chmod 750 oswatcher <<<<<<<<<<<<<<<<<<<<<<<<Change the right, then we can edit it as per our expected. # ls -al oswatcher -rwxr-x--- 1 root root 5128 Aug 19 03:39 oswatcher
5. 检查当前oswatcher脚本中的设定，并修改（本次修改将原有的最大保存3G的日志，修改为最大日志尺寸为4G）
# vi oswatcher ---------------------------------------- fi (umask 0037; nohup ./startOSW.sh 15 168 bzip2 4 >/var/log/cellos/start_oswatcher.log 2>&1 &)& <<<<<<<<<<<<change this part of the script, will let the script run as per our expected. # Dont direct logs to startosw.log. It grows too large and fast # (nohup ./startOSW.sh 15 168 bzip2 3 >/dev/null 2>&1 &)& popd >/dev/null ----------------------------------------
6. 停止oswatcher
#/opt/oracle.oswatcher/osw/stopOSW.sh
# ps -ef | grep OSW
root 10528 9529 0 03:59 pts/1 00:00:00 grep OSW
7. 手动启动oswatcher
# /opt/oracle.cellos/vldrun -script oswatcher
Logging started to /var/log/cellos/validations.log
Command line is ./validations/bin/vldrun.pl -quiet -script oswatcher
Run validation oswatcher - PASSED
The each boot completed with SUCCESS
8.再次检查OSwatcher的设定，最大日志尺寸已经改为4G
# ps -ef|grep OSW root 109041 1 0 03:37 pts/1 00:00:00 /bin/ksh ./OSWatcher.sh 15 168 bzip2 4 root 109073 109041 0 03:37 pts/1 00:00:00 /bin/ksh ./OSWatcherFM.sh 168 4 root 109348 59019 0 03:37 pts/1 00:00:00 grep OSW
更对oswatcher细节，可以参考以下官方文档(注：目前文档1585389.1中已经记录如何永久修改exadata上osw的设定， 但是本人测试的时候，该文档并没有相关记录)
OSWatcher does not Retain Data for the Full Retention Period Specified (Doc ID 1585389.1)
OSWatcher Black Box (Includes: [Video]) (Doc ID 301137.1)
