I agree it is out of date. I haven't worked on that page for over a year I think. I have done so much cool stuff in the industry since then and I keep meaning to document it and share what I've learned but I've just been lazy. I did however document how to do 801q packet tagging in there. I did a lot of that about a year ago at NetApp where I worked with a lot of VLANs and virtual machines.
Well out of date or not, for someone like me who's learning the sysadmin ropes and will be sitting for interviews very soon, this is extremely helpful.
You might expand the packaging section to include at least apt-get, and the source control section to include at least svn and preferably git. There's really no reason to still be using CVS at this point.
Don't forget darcs. I find it useful since they distribute statically linked executables. Installation then becomes a matter of just dropping darcs into a bin directory.
(Cheat sheet? Cheat Novella, more like! I think it's probably past the point where it needs to be split up a bit, even if only for your own benefit.)
"How do you check disk usage. How do you trouble shoot a high disk usage issue"
which is fine, until someone's done something similar to the next bit of advice - i.e. deleted some "old", large files that were still open by the application.
Better: compare filesystem usage (with df), against disk space used (with du), to deduce that there's an open file that has been unlinked. Then use lsof (whatever Sun say about pfiles or about dtrace, lsof is an essential tool) to look for processes with open files that have large offsets.
And, when deleting large files, check whether they're open, and truncate them before deleting them if you're not sure.
(note to another poster - ain't nothing wrong with Solaris! But I'm not really sure what interviewer might ask questions about the OSI 7 layer model, these days, and I agree that there's a real odd jumble of scripting snippets in there: best advice to anyone wanting to improve that side of their systems administration skills, is to read a copy of Unix Power tools.)
Much of this content is lifted from other sites and articles, without attribution, with little or no value added commentary or presentation. I'm not sure why I'm offended by this, but I am.
Much of it was, I created this in my spare time to help me pass interviews. I was being interviewed by people with less time than me in the industry who didn't know what they were talking about and they were reading from a script or just seeing if I knew as much about their own personalized environment as they did. I needed a one stop place to dump a bunch of stuff I'd need to be able to answer questions.
This seems to be geared towards the type of systems administrator that you're not really looking to hire, if you're the kind of startup that hangs out on hacker news.
Frankly, the Solaris & NIS+ questions are useless. You will not use either in a modern web startup, unless you're doing something wrong. Ditto for the AIX questions, LPARS, HP-UX, and Power-VM sections.
Ask a sysadmin questions relevant to your environment, or where you want your environment to go. Don't expect them to be a domain expert in every single region, and don't give weight to domain expertise irrelevant to your startup. I coached a friend through a phone interview with a generic LAMP-based startup recently whose hiring consultant spent an hour asking him questions about BGP, OSPF, Solaris, VMWare and EMC specific SAN devices.
The interviewer was a typical big company/academic systems administrator who had no startup experience. It was pretty ridiculous, these questions were asked by the interviewer because that's what his expertise was in, not because any of these were ever going to have any relevancy for this startup.
He passed, of course, and now their whole environment are AMIs.
I will say that I think any systems administrator today who isn't excited about learning Puppet or Chef is the wrong choice for a web startup.
I made this site quite a while ago when I was unemployed and the jobs available were few and far between and the questions being asked in technical interviews were lame questions that the interviewer was reading off of a script. I am the kind of person that will always look for the better way of doing something and I found many solutions for things that are not on this site.Solaris is much more stable and mature than Linux. I have had far less problems configuring RAC and other applications on Solaris that I have on any flavor of Linux. My favorite Linux dristro is Ubuntu which I wrote very little about on my site because nobody in the industry uses it. They all uses Centos, FC, and RH. And if you think that virtualization is not what a startup would be interested in then you are wrong. I have done an assignment lately where I worked with DS3400s, Brocate FC Swithes, Nortel BNT 7 Layer switches, Bladecenter HS22 with 7 x 7870 AC1 Blades. I setup ESXi 4 and Vsphere. I created a RAID 6 Array for the VMFS Store and several RAID 5s and O+1s for the databases. I created 2 Win2K8 RC2 Blades a MS 2K8 SQL Cluster between the two. They both shared the same LUN as per prerequisite by Microsoft.
Puppet and Chef are both Ruby based configuration management systems. One might say that CFEngine begat Puppet which begat Chef.
Puppet was written by Luke Kanies after years of frustration as a systems administrator working with CFEngine. It's roughly three years old. It's mailing list has over 2,000 subscribers, and when I attended the recent PuppetCamp there were roughly 200 active participants in attendance. My very anecdotal estimate is that there are at least 30,000+ nodes using Puppet (My old consulting company was responsible for 6,500 nodes). Puppet is now maintained by Luke's startup, http://reductivelabs.com/ which took a $2m Series A round from True Ventures this summer. Puppet is written in Ruby/Rails, and implements it's own DSL for describing the configuration of your systems and applications.
Chef was written by Adam Jacobs of HJK Solutions, now http://opscode.com/. Adam was a vocal Puppet consultant, and wrote Chef in response to dissatisfaction around the handling of Puppet Bug #1010 (http://projects.reductivelabs.com/issues/1010), related to problems with idempotency initiated from Puppet's directed graph. Chef is maintained by Opsocde, whose CEO is Jesse Robbins from Oreilly and whose CTO is Adam Jacobs, they received a Series A round of $2.5m from Draper Fisher this summer.
Both are open-source, and sit somewhat in the same genre as CFengine, ISconfv4 (mostly abandonware), and BCFG2.
* nslookup has been deprecated by its makers for years now.
* ip route's replaced 'route', which still works but won't show you all the things that ip route can.
* Nothing on dot1q VLAN tagging, which is pretty common these days
* No CDP from hosts, which is also pretty common
* No Postfix or any other non-Sendmail mailer.
* The somewhat bizarre 'bash script to validate RPM files'. Why not just teach the proper query syntax or, for scripting, RPM & yum APIs?
* Solaris.
If you're being asked these questions, or your interviewer is expecting these answers, you may wish to consider whether you want the role.