As I mentioned in my last entry I was scheduled to take my LPI certification exam at a nearby VUE affiliated testing center this morning. It took about an hour; though I had an hour and a half to devote to it.
Note: my last diary entry seems to have somehow overwritten and munched my previous two. I have no idea what happened to them. I do have a copy of the long (8K) second one if anyone really cares
I'll just give a bit of an overview and offer some comments.
First, I am on the LPI advisory council. However, my participation has been severely limited by my travels and travails. So I've mostly just been lurking on the mailing list, and I haven't been able to make it to any of the meetings. I signed up through the normal process, and I have no "inside" information that could affect my performance on these tests.
I'll also note that Linuxcare (my employer) is an official sponsor of the LPI.
Make what you will of my presumed biases.
Now, on to my comments.
The Basic Process
My test consisted of 72 questions. These were presented by a simple Windoze program on a plain PC. (Actually I noticed it was an e-Machines). Most of the questions were multiple choice. A few (less than five) were short answer (fill-in-the-blank). The requisite fill-in-the-blank answers were either full paths or simple command names (no arguments or options).
There were no diagrams or graphics. There were no "interative simulation" questions. This whole program could have been written as a */Tk script ( TCL/Tk, Perl/Tk, Python/Tk, whatever) or even as a shell script, without too much trouble.
The font used by this program was not good for displaying finer points of UNIX commands. (The 'l' ("ell") and 'I' ("capital I") were indistginguishable, which could have been an issue on the one vi question that I got).
There were no problems with the hardware, sofware. Windoze didn't decide to die on me. Yay!
This testing software allows you to page back and forth through all of the questions. On each question's dialog screen you can also check a small box to mark it for later review. There is also a "Comments" button on each question. I used that extensively (on about a dozen questions).
Overall I'd say that the content of the questions was pretty close to the target. It seemed relevant to the intended purpose (gauging whether someone has the basic knowlege necessary to be a competent Linux sysadmin), without being overly esoteric or trivial.
Note: I did not say that these exams are intended to test for competence. That might not be possible within this examination format. I said that it's intended to gauge whether or not someone has knowlege that is comparable to a competent system administrator. I only highlight this point because the matter is so controversial among the SAGE and other UNIX/Linux communities.
I did not know all of the answers to all of the questions presented. However I was able to make pretty good guesses, and there is only one that I'm sure I missed (since I looked up the man page when I got back here).
I would not characterize any of the questions as "tricky." None of them were about programming.
About half of the questions were not Linux specific. Those were about common UNIX utilities and TCP/IP networking fundamentals. Only one was about PC architecture (about the conventional IRQ assignments of a common port/device interface). Very few (maybe three or four) of the questions were specific to Linux on x86 platform. Linux users with a predominantly SPARC, PowerPC, or other hardware background would have had little trouble with this test.
Only two of the questions involved defining some term or acronym.
About eight questions (a bit over 10% of the test) was about package management using the rpm and dpkg commands. These two were about equally represented. There was nothing that was overtly biased to any distribution. (There are now at least three Debian based distributions --- which isn't nearly as many as the plethora of RPM based distributions, but seems appropriate to me).
There was a half dozen questions on the X Windows System and Xfree86. There were a few (three or four) questions on printing (all assumed a BSD lpd/lpr subsystem). There was only one question on vi.
I only saw one typo (using "its" when they meant "it's").
There was one question that was out-and-out blatantly WRONG. I noted this in my "Comments" and the testing center staff have assured me that the LPI should should get a copy of all my comments. (The question had to do with the mapping of SCSI disk device/partitions to /dev/ names).
Other than that there were only a couple of questions that might be ambiguous or might have controversial answers. One, in particular, had to do with the recommended partitioning and swap space allocation for a given system configuration. I'd never partition any system in any of the proposed ways. (Well ... maybe at gunpoint).
As I said, I think the test did about as well as I could expect. I'd like to see some interactive simulation (like "enter a series of commands to ...", or "using vi edit this text to look like ...").
Ideally I'd like to see a practicuum test. This would involve setting the subject in front of a trio of systems, and a couple of CDs and with instructions to "make this into a web server server, that into mail/POP host and that other one into a router/firewall for it."
Unfortunately I don't think that this is currently economically feasible.
Of course I don't know how well I did on this exam. Since the exam is still in the beta stage, I won't know my score for several weeks. Thus you should take everything I've said here with the same large lump of proverbial salt as anything else you read on the 'net.
I'm sorry to hear about the flammage from the Stampede camp.
Ironically when I was at the Sunday meeting of the South Bay Community Network group we bumped into one of the crew that was hanging out in the Stampede booth at the LinuxWorld Expo in New York. She was working at the pizza place and apparently only recently moved to the area. (I can hear the headhunters swarming now!).
Ian said only in the Silicon Valley...