|
|
|
| Menu | Resume |
UNIX Performance Monitoring System
Year: 2001
Sponsor Company: AT&T Broadband
Technology: Oracle, SQL*Plus, SQL, SQL*Loader, Seagate Crystal Reports, MRTG, SNMP, FTP, Crontab, Windows clients, Sun Solaris, Tru64 Unix, HP-UX
Detail: Designed and programmed this system to gather key statistics from various UNIX servers. Data was put into an Oracle database and then reports were generated using Crystal Reports Server.
It gathered the statistics using a k shell script by redirecting the output of iostat, mpstat, vmstat, ps, netstat, df and kmstat to files with the date as a file name. The script was started by a cron job. Various versions of the commands would be run depending on if the server was Solaris, Tru64 or HP-UX.
(If vmstat does not work you can try vmstat -p. If netstat does not work try netstat -i.)Each command would gather ten samples of data with a 30 second delay between samples. (vmstat 30 10 >> file)
Once an hour files were sent to a central server using ftp. Then the central server ran a perl script that would format the data into a format that could be loaded into an Oracle database using SQL*Loader. After it was loaded, Crystal Reports was used to display the data as charts showing the trends of usage.
In addition I set up MRTG for the systems that had a SNMP agent running. This was also used to monitor a few NT servers. I configured the MS SNMP agent and MIBs on these servers.
Features:
- Web based.
- Graphs included in reports.