SysMan Mall

Availability Reporting using SunMC Performance Report Manager

To get systems availability one can make use of uptime command which reports the time since the system was last restarted. The restart time is recorded in systems utmpx database and the uptime utility computes the difference between the two time stamps and reports the time for which the system has been up and running.

For performing anything fancier than simple uptime, that is figuring out the amount of time the system has been up in the last "n" years/months/days one can make use of UNIX "last -a" command and look for matching system boot and down messages during the specified time window. To say the least its tedious and error prone because no timestamps or recordings are made when for example power is lost or the system panics or some other unforseen situation occurs which results in utmpx not being updated.

I ran Simple Uptime Status reports using SunMC PRM and was able to drill into some of my old systems. It was no surprise that some systems which were used for testing experienced heavy rebooting and accordingly those have many dark lines (actually there is red in there if you were to zoom inside the graphs while using SunMC/ PRM). The big red on demo5 can be attributed due to a long  no operation period. The winner ofcourse appears to be demo2 with no outages and demo4 with just one downtime spanning few minutes in the last two years. demo6 system happens to be the SunMC server which has good availability numbers, staying up for most of the time while I played around, used it for internal/ external demos, upgraded SunMC versions from 3.0 through 3.6. Some of the outages (thin lines) captured on the graph are the power outages we experienced here in the lab.

PRM also provides uptime summary reports where percentage up/ downtime or the number of hours of up/down time information for each selected system and a total uptime report which will look similar to the output of last except that it is graphed or can be displayed inside a table with PRM. The reports detect and capture daylight savings time related changes as well, reflecting shorter and longer days as the case may be.

Posted at 04:30PM Jan 17, 2007 by adikhit in General  |  Comments[0]