IT Certifications Aren’t Worth the Paper They’re Not Printed On!

At work we are trying to attract a new client. They toured the facility, they looked around and asked questions, all while my peers and I demonstrated our knowledge and abilities. They were impressed as we took actual phone calls for actual users. They oohed and awed as we remotely administered dozens of servers, of varying types and platforms. It all seemed to be going smoothly until they wanted to see our certifications.
 
We were like see our what… You are kidding right? We are doing the work for many other customers and we have real working relationships with fortune 100 companies, but they want us to produce papers to prove that we can do it. We all agree we should go get the papers; how tough could it be? So we all start working on our Server+ certifications and other IT(Information Technology) Certifications.

Server-like Systems

This is how related Server+ is

• Created in 1986.

In an industry that you have to read the news daily or get left behind, what from 23 years ago could possibly still be relevant. I skipped the 32bit/64bit transition because I knew I didn’t have time to keep up to date with all the news, when I stepped out of IT for a while. I picked it back up just earlier this year, and I hit the books (The Web really) relearning what was needed. Only occasionally do I need knowledge generated during my 18 month IT Hiatus a year ago. How could I ever need knowledge about something 23 years ago rather than more knowledge from now, or at least the historical relevance of archaic tools. I could understand studying why one technology beat out another. But no, they go into technical details like this:

• Standard: SCSI-1
• Data Bus Size (bits):8
• Clock Speed (MHz): 5
• Throughput (MB/s): 5
• Maximum Devices (including SCSI controller): 8
• Cable(s): 50-pin Centronics

When the information wasn’t old or unrelated it was inaccurate. I could see how some of these could be judgment calls, like when they said Itanic Itanium as a good 64bit chip. I was was there for the beginnings of the Itanic, and I know better, and the Itanium Sales Forecasts agree with me. But, someone still could still disagree because on some level it is still an opinion. However some of the items that used in this Certification are just plain wrong by the math.

Incorrect Question

How I Fixed 2 PCs Today

Big surprise, I didn’t work for Dell of Tek-Systems today, I am still waiting for paperwork. I did have some computers from a customer though. I fixed 2 desktop PCs today. I frequently get computers with little or no explanation from my customers. The explanations I got with these was “this one is slow” and “I put this one in the closet when it stopped working”, this is better some explanations I get. One time I got a work order with only the customer phone number and the phrase “don’t work”.

Digital Multimeters are more reliable than power supply testers.

1. Power Supply
2. RAM
3. CPU & Motherboard
4. Hard disks and other storage
5. Software
6. Thermal stress

There are some very specific reasons why I check in this order. First let me explain why I do not run short an simple tests first. Many technicians, look and the machine and try to get it to make an error, which is good for some things. However, what I go for is a guarantee that is is fixed. Too often I have had to clean up after other technicians or even clean up after I have taken shortcuts, but anything short of a guarantee when fixing computers is nothing in my opinion.
 
When I say guarantee I mean as close to 100% as possible. Of course there is 100% accurate test. Memtest86+ is a great example. I have seen it report incorrectly exactly once, and the cpu had bad cache. CPU cache is kind of like RAM, so I can’t fault it entirely. I wish every test had this kind of accuracy, but in practice anything better than 1 in 50(98%) is pretty usable. Any less accurate, and I tend to second guess the results when things start to getting complicated. Second guessing and uncertainty are the last things a technician need.
 
The reason I check the power supply first is that it can damage other components and duplicate any other issue. I find power supply testers unreliable, so I use a multimeter instead. To test RAM I use memtest86+. To test CPUs I use PC-Check, but I would like an open source alternative. To test hard drives I use Gwscan or badblocks from a Linux rescue CD. For scanning software I like spybot s&d (only for windows), housecall.trendmicro.com, clam antivirus, and if you have access to it, the geek squads MRI. Anything that uses all the CPU and steadily uses RAM can be a stress test, for windows prime95 is very popular.
 
I didn’t realize until I was halfway through it, this would make a great How-To. I think I will formalize this a bit and release it as a that way. I glad you read about my thoughts on this.
 
EDIT, I almost forgot, here is check list I write down info as I check the computer: Diag form.odt or Diag form.pdf

I am Tech Support.

I don’t know which is worse, that a random Air Force guy… Airman First Class Matt calls me about his Linux machine running ZFS-Fuse an obscure technical issue or that I am able to identify his issue within minutes and tell him how to get a solution.
 
First, some background info, I built Matt’s computer almost 2 years ago. It has Ubuntu Linux It was to be the Anime storage machine to end (Not safe for work) Anime storage. He has three 1 TB Hard Drives to help with this task.
 
Normally he logs in and his giant ZFS volume is his home folder (1.5 TB because it makes 2 copies of everything). Now he could not log in, and got some generic error message. He kept telling me different things, and I didn’t want to argue the point. Over the phone people are rarely capable of providing exact error messages. He did mention something about his home folder not existing each time he told me he tried to login. I suspected that something was going on with with his ZFS setup.
 
First, I had to get him logged in to at least a command line. If your Ubuntu machine gets to the login screen, you can select “Failsafe Terminal” from the “Sessions menu”. If you do, then login you will be presented with a simple desktop with just a terminal. Now that he could log in I ask him to run this:
#sudo zpool status
On a healthy system it would look something like this:
#sudo zpool status
pool: storage
state: ONLINE
scrub: none requested
config:
 
NAME STATE READ WRITE CKSUM
storage ONLINE 0 0 0
sda3 ONLINE 0 0 0
sdb3 ONLINE 0 0 0

Matt was unclear, but he did keep saying “bad intent log” so a quick google search later and I found the ZFS Log Device Recovery in the ZFS administration guide. This tells me his is not healthy, so it probably looks more like what they mention in the guide.
 
The name of his storage pool is “storage” and the documentation says we should simply bring the device back on-line. So we ran:
#sudo zpool online storage sda3
We try this and several variations on it, and it still stated various messages all still mentioning a “bad intent log”. This told me I had to try something else. So next we tried:
#sudo zpool clear storage
The lead to a great deal of error messages that sounded like “could not bring online; zpool is faulted”. As far as I can tell this means that I have multiple failing hard drives. Does anyone else have input on this? I will be doing more reading on ZFS, but from what I understand that the only way to fault a zpool that has ditto blocks set to 2 (make 2 copies of all the data) is for 2 of the devices to fail. I think I did alright for 10 minutes of phone troubleshooting though, Sucks to be Matt’s computer though.