Ways not to install Proliant ML330 G2

This is such a hack on how to get things going I don’t even want to admit it but I also don’t want to forget in case I need to do it again.

(Way back in the beginning, I figured out that getting the LSI Logic ATA raid controller — megaide was going to be a pain.  So I formated the drives under the <ctrl-a> scsi bios to be independent.  I figured, I can always do a software raid later, if I need to.)

I installed some form of Debian on this box 1.5 years ago and I can’t quite remember what I did.  I think I started with Debian 3.01 (sarge?) installer and ran it with the expert boot option.  Then when it tried to autodiscover the devices I would deselect the ide-cdrom device the first time through (it is the ide- that is listed last)  The next time through I would select the last module and go from there.

This time I installed Ubuntu’s Breezy like a regular system but on reboot the system would tell me that it couldn’t find /dev/hda1.  If you run ‘cat /proc/partitions’ you should see what partitions are available, mine was obviously installed on /dev/hde1.  Now ‘mount /dev/hde1 /mnt’ to temporarily get access to the root partition.  If you push on, you can edit the /etc/fstab file from the busybox prompt using ’sed -i.bak s/hda1/hde1/ /mnt/etc/fstab’ and ’sed -i.bak s/hda1/hde1/ /mnt/boot/grub/menu.lst’

I wasn’t able to boot correctly without manually editing the grub on the fly using ESC and ‘e’ on the line with the /dev/hda1.  After pressing enter to commit the grub change, you will need to use the ‘b’ option to boot.  It took a long while for grub to hand off from initrd to hde but it eventually worked (I thought that I should have modified the hd(0,0) to hd(5,0) but grub failed.  Tab completion in grub is great)

I ran ’sed -i.bak s/breezy/dapper/ /etc/apt/sources.list’ after getting a decent boot.  ’sudo apt-get update && sudo apt-get dist-upgrade’ and sat back.  While waiting, I found from /var/log/dmesg that my cdrom was /dev/hda (go figure) and the extra 6 Gb drive I stuck on a IDE channel for “ease of booting” was /dev/hdc.  I ended up not using the /dev/hdc drive.

Now I’m going to look for an easy way to call the LAMP server installation system.  I thought it would be available as a task but I’m not finding it.

I’ll post my lshw here to give others an idea of the hardware I’m using.

How To Test Rails Code

How to Use ZenTest with Ruby | Linux Journal

Lesson on Scaling Ruby on Rails

Here is a blog highlighting some issues when migrating eins.de from a php application to a ruby on rails application.  The old php site was able to handle 900,000 page impressions before having problems.  Their first day on Ruby on Rails was failing after 150,000 page impressions, however after tweaking (Can you call changing applications and server rols tweaking?) they eventually were left with an application that could easily handle 1,200,000 page impressions without crashing.

The old codebase roughly consisted of around 50.000 lines of PHP code (plus a closed-source CMS that’s not included in this calculation). We’ve rewritten most of it (some features were left out on purpose) in about 5.000 lines of Rails code.

The adventures of scaling, Stage 1

Update March 18: A follow-up article addressing reader comments is now available here.

Update March 20: Stage 2 is online.

Update March 27: Stage 3 is online.

Update April 03: Stage 4 is online

technorati tags:Rails, Optimization

FastCGI verses Mongrel

So far this solution has proven much more stable (at least on FreeBSD) and was able to handle our peak traffic of 350.000 page request per day. In practice we use up to 8 Mongrel processes on each cluster node and it seems that Apache is the bottleneck and not our application servers as before. The next step for us is to introduce another web server that handles the incoming HTTP requests and has it’s own Mongrel cluster.

Scaling Rails with Apache 2.2, mod_proxy_balancer and Mongrel

technorati tags:FastCGI, Mongrel, Ruby, Rails, Cluster