I’ve complained before about the problems with package management and system updates on SuSE 10.1. Well, today, I have some good news to report:

I forced myself into a reboot this morning by pulling out my optical drive which froze my system completely. When I rebooted, the Zen Updated applet was complaining about updates so I clicked on it. Nothing happened really, so I thought I’d try (yet again) the System Update option in YaST. Well, I did, and it worked! It installed about five or six package updates and as soon as it was finished, I started Zen Updater and was able to start the real update option.

Ironically enough, this all happened five minutes after I started looking for my SuSE 10.1 installation discs to do a reinstall since the update process was fixed several weeks ago for new installs. If I hadn’t sent my CD’s to Brandon a few weeks ago for use on his M7, I’d be in the middle of format/install right now.

DesktopLinux.com has the scoop: Lenovo preloads SUSE on ThinkPad

On August 4th, we found out that Lenovo Group, the company that has taken over IBM’s Personal Computing Division, had made a deal with Novell Inc. to preload SLED 10 (SUSE Linux Enterprise Desktop) on its ThinkPad T60p mobile workstation.

Needless to say, this is HUGE both for the Linux community and Thinkpad users interested in Linux. Me? Yeah, if this were a Venn diagram, I’m in that center part – smiling like I just won the lottery. I can only imagine that better Linux support from the manufacturer for current owners and future Thinkpad models as well.

The timing on this is pretty funny, considering I just had a conversation with Brandon earlier in the week regarding just how far the end-user experience has come on Linux, even in just the past 18-24 months.

Lenovo – I had my doubts when you bought out IBM on the Thinkpad division, hence the reason I snatched my R51 up so fast after the aquisition (before you had time to ruin the best laptops on the market). However, the Thinkpad line continues to receive rave reviews from tech heads and folks are just loving the newer models too. Kudos to you, Lenovo. Kudos all around.

I contemplated titleing this post “How to Totally b0rk Your Server”, but decided instead to go with a more informative one instead of an editorial one. The following is a story about how to totally trash a cPanel/WHM server environment and then get it back to stable in no time. In our case, it took us about ten hours, but YOU my dear reader can bask in our silliness and be back in no time, assuming you’ve found this post that is. And oh how sorry I feel for those who don’t. Anyhow…

As many of you may know, I work for a hosting company – Mindstorm Hosting and my major responsibility is to keep the server running smoothly and address customer support issues directly related to technical problems. We run a dedicated “unmanaged” server, meaning that our hardware provider is responsible for just that – the hardware – and don’t lend very much support at all in relation to software and stability issues. So yesterday, I was attempting to move our final domain that hosts our billing/helpdesk package over to our dedicated server and stumbled upon the fact that the Zend Optimizer wasn’t installed. A quick search on some popular cPanel/WHM forums yielded me this results: run the ‘/scripts/installzendopt’ and follow the prompts. OK, piece of cake. A few strokes of the keyboard later, phpinfo(); is reporting that Zend Optimizer 3.0 is now installed.

As I started tweaking the billing software, I noticed the server getting extremely sluggish. A few apache restarts later, its not much better, so I moved on to a full system reboot. Great! Things are back to normal now. Fast forward 30 minutes and things are pretty much dead – I can’t get access to a remote shell and any page requiring PHP is a red light. Another hard reboot later (thank goodness for remote reboot capabilities), I get shell access and begin watching running processes and ‘top’ to see whats going on behind the scenes. It didn’t take but a few minutes to find the culprit – PHP! The longer I watched, the more PHP processes appeared. At first, it was five using 19.9% of the CPU each, then it was ten using 9.8% each – yikes! For those not in the know technically, 100% CPU utilization is always bad – but its especially not fun in the web hosting business. On your personal computer, you’re probably the only person using it, and you can sit and wait for a minute for your machine to catch up – every minute that passes on a dead web server is hundreds and hundreds of unserved web pages.

OK, so now we’ve identified the cause of the sluggishness, so how do we fix it? Well, my instant fix was to run the command ‘killall -9 php’ which worked, but only temporarily, so I setup a cron job to run that command every five minutes while I sorted it out. My first instinct was to recompile apache/PHP now that Zend Optimizer was installed. This took about two hours via the WHM interface thanks to those stuck PHP proceses before the cron job was fired off.

But still no joy. Finally, I smartened up and reverted my php.ini file to the pre-Zend Optimizer one. It was as simple on my server as going into /usr/local/lib, removing the symbolic link created by Zend and renaming the php.ini-zend_optimizer.bak to php.ini, and restarting apache.

Bingo! We have a fully working server now. But I still can’t run ModernBill because it’s complaining now that Zend Optimizer isn’t installed. So what’s a server admin to do? Sleep. Thats what.

After a good night’s rest – I set out this morning to make it happen – without bringing the server to it’s knees. It didn’t take long after my coffee to find the critical difference between the server it was currently being hosted on and our dedicated setup. Everything appeared to be identical at first glance – a few different options enabled/disabled on PHP, but nothing related. Both machines were running Apache 1.3.36 and PHP 4.4.2, but then it jumped off the page at me: Zend Optimizer v2.6.2.

So I went over to http://downloads.zend.com/optimizer/ and grabbed the latest 2.x installer, extracted it, installed it, restarted apache, and ta-da!

If you’re reading this – it’s still running strong. If not – I’m fired.

IBM now has ready IBM Lotus Notes on Linux. I’m not sure that Notes is nearly as popular today as it was 10 years ago, but nevertheless, this is a bold move for IBM (lots of R&D dollars) and excellent news for the Linux community. As the author points out, Notes is “the industry’s first business-grade collaboration software to support Linux on the desktop.”

Let’s hope some other major software vendors will follow suit.

With a little help from Nicholas from the Linux-PowerEdge mailing list, I was able to get ipmitools working on my new Dell PowerEdge 830 thats running OpenSuSE 10.1. I was really making it harder than it needed to be, as most of the stuff I had found online was related to significantly older kernels. Nicholas was running some flavor of Ubuntu, so I have a feeling these are very general instructions that should work for any flavor of Linux running a 2.6.16 kernel. To check your kernel version, type

uname -r

at any command line.

Ok, it’s actually really simple:

1. Download/Install ipmitools. Go to the SourceForge site and download ipmitools and then install them. There’s an RPM that worked for me, but if you have trouble, grab the source and follow the included instructions.
2. Edit the boot loader configuration. If you’re running SuSE/OpenSuSE like I am, I suggest using YaST go to “System->Boot Loader.” Once there, make sure your default kernel is selected and choose the Edit option. Then in the box labeled Other Kernel Parameters, just add

   pnpacpi=off

   and then click OK and then Finish. For the non-SuSE users, you can edit /etc/lilo.conf or the corresonding grub file manually and then run whatever necessary tool to parse your new conf file (Example:

   /sbin/lilo

   for LILO users).

3. Load the kernel modules. You have two options here: a) load them manually using modprobe, or b) make the necessary modifications to have them loaded automaticaly at boot. I prefer to have the necessary modules load at boot time so that I don’t have to worry about it ever again. In SuSE, open YaST and go to “System->/etc/sysconfig Editor” and in the left-hand pane, choose “System->Kernel->MODULES_LOADED_ON_BOOT.” Now, in the box provided on the right side, add this:

   ipmi_si ipmi_devintf

   and then click Finish and OK. Exit YaST. If you want to do this the manual way, simply type

   modprobe ipmi_si

   and then

   modprobe ipmi_devintf

   on a command line.

And that’s it! You may need to reboot after changing the kernel boot parameters and before loading the ipmi modules. For SuSE and OpenSuSE users though, you can make both those changes within YaST and then reboot. Upon reboot, you should be able to use

ipmitools sel list

to view a list of errors from your sensors,

ipmitools sel clear

to clear the logs as well as all other standard ipmitools commands. To see the full list, just type

ipmitools --help

.

cnet is reporting today that four major mobile-phone makers are joining forces to develop a standard, Linux-based platform for their phones and other mobile devices. This is possibly the best tech news I’ve seen recently. With the two heavyweights involved (Motorola and Samsung), this is bound to just blow us all away in a few years. According to the article, Motorola has already shipped more than 5 million Linux-powered telephones and that number continues to grow.

Linux is extremely powerful and versatile, and hopefully, in the not too distant future, the reasons I use Linux as my primary operating system will be the same reason I’m buying a new cell phone – Freedom of Choice. Just imagine if there were hundreds (maybe even thousands) if applications available freely on the Internet for installing to your phone. Let’s just hope more carriers than NTT DoCoMo and Vodafone get on board with this.

I’ve about had all I can stomach of SuSE 10.1. The included version of Firefox crashes on me at least five times a day, normally when hitting my Backspace key in order to go back one page. It’s really freaking annoying whenever you have 8 important tabs open. I’m about to go uninstall it and see how many dependencies I can break and then install the binaries directly from Mozilla.

Also, I did finally manage to get some online package repositories added, but they’re useless. My package manager of choice, ‘apt-get’ has been removed in favor of ‘smart’ which is just totally screwed up. If I want to install one package, it decides that I need to install the X.org devel packages, Mesa, and about a dozen or so other ALREADY INSTALLED PROGRAMS in order to meet dependencies. Even better, it appears that I have zen-updater working, but in the two weeks or so since installing SuSE 10.1, there hasn’t been a single security update, let alone any bug fixes for the numerous known problems that have been documented. Which leads me to believe that zen-updated actually ISN’T working.

*Sigh* – I think I might have to rever to 10.0 or maybe even think of going with another distro – because this is borderline ridiculous.

I just ordered what will soon become our Asterisk server for work, and now we need to name it. Lately, I’ve been giving Star Trek’ish sort of nicknames to the Linux boxen around the workplace, and I think I’m gonna stick with that theme. After a brainstorming session with Matt, I have a short list of possibilities and I’ve created a poll to help decide on the name. The poll can be found at the top-right of any page on wantmoore.com (minus the gallery section), so be sure and vote! If you have a good suggestion that we missed, leave it in the comments here and I’ll get it added to the poll module ASAP.