Cosc 4750

1
Cosc 4750

• Chapter 23,25,26
• Maintenance Analysis

2
Chapter 23

• We covered most of this early in the semester and
  in lab
• There is some useful info in the chapter about
  confusing some pieces that we don't have the time
  to go over.

3
Maintenance Contracts

• Annual cost of 10-12 of components list price.
• On-site maintenance
• usually within 24-48 hours a vendors tech will
  come out and fix the problem.
• parts care
• Vendor will send you a replacement for the
  broken part, you put it in and send the old
  part back.

4

• Warranties
• usually good for 1 to 3 years
• normally a parts care contract
• Some company offering on-site service as well

5
Board Handling lore

• static electricity
• electronics are sensitive to static shock
• Should use a grounding strap on the wrist when
  working with electronics
• Reseating boards
• Sometimes, the a board works it way loose or
  loses contact, so you reseat a board.
• Also, wiggling a board will help as well.
• See demo in class

6

• Monitors
• Do NOT attempt to fix them yourself.
• Holds an electric charge that can seriously hurt
  or kill you.

7
Preventive Maintenance

• Computers should cleaned regularly.
• Normally, mid spring
• Either vacuumed or blown clean
• boards reseated or wiggled
• Dont lubricate fans or parts. Simply replace
  them.

8
Environment

• Temperature
• between 60 to 70oF
• Above 80 ambient temperatures means about 120
  inside the computer.
• Also kept computers out of direct sunlight
• Humidity
• about 45 is best
• A/C can help to kept both in check.
• Beware of your power drain by computers and UPS.
  You can cause local low power to your computers.
• one study estimated that 13 of electrical power
  in use is consumed by computers.

9
Sys Admin Tool Box
10
Performance and Analysis

• UNIX will give each process an illusion of
  infinite resources
• When UNIX runs out, it will work hard to maintain
  the illusion, but the system usually comes to a
  halt.

11

• Collect and review data on the system
• What is the average CPU usage and memory usage.
• When a problem happens you can find out why
  easier.
• When adding a new application, check with the
  vendor for average cpu and memory usage.
• Can you machine handle it?

12
Improving performance

• Turn off unused or un-necessary programs and
  daemons
• Adding more memory is a good idea.
• Attempt to correct usage problems
• Have huge processes run over night
• balance the system load with another machine if
  possible.

13

• Reorganize/organize hard drives for better use
• Move hard drives around, put them on separate
  controllers.
• Monitor network usage
• with netstat or other programs
• If you network traffic is above 80, you are in
  trouble.

14
Analyzing CPU usage

• vmstat and/or sar
• lists obtain summary information about cpu usage.
• sar u 5 5
• lists percentage of cpu time on user code, kernel
  code, and idle.
• vmstat shows similar information and page faults
  and memory information.

15

• uptime
• shows system up time and load averages for 1, 5,
  and 10 minutes (or 5,10, and 15 minutes)
• A lot average double the number of CPUs in the
  computer is when the system will begin to fail.

16
memory usage

• swap (sun, sgi), swapinfo (hp-UX), swapon s
  (redhat), pstat s (freeBSD)
• display swap space, free, total, and used.
• vmstat 5 5
• displays processes running, blocked I/O, and
  swapped out.

17
Disk I/O

• iostat 5 5
• displays tty activity, each harddrive, nfs, and
  cpu activity.
• iostat D 5 5
• display percentage of use

18
Redhat Performance data

• procinfo provides a summary of system performance
  information in a more readable form.
• procinfo f info and updates every 5 seconds.

19
quick fixes

• use top or ps to find runaway processes or power
  users, when the system is very slow.
• Can use a command called limit
• limit memoryuse 32m
• All subsequent commands can only use 32m of
  physical memory.

20
Q
A