Areca has released a new cli for Linux on 22 February 2005. This is a new release that allows non-interactive requests to the CLI, something that was not possible with the old version of the CLI.

Because I wrote significant nagios plugins for Areca that rely heavily on this cli tool I've tested all my plugins with the new version. At first sight it looks like all software works without a problem. However the software still uses the old way of getting information out of the CLI. I'll rewrite some parts of the software so it will use the new non-interactive way, which is cleaner code and is probably faster.

Today I discovered one of our secondary nameservers hadn't really vacuumed it's postgresql backend in quite some while. It resulted in a database of 7G which actually only holds about 180M of data. Ofcourse I started vacuuming the database, after quite some while (2, 3 hours) it finished. However I still noticed the database being 4G of data. So I wondered how to speed this up and make it work better. I've come up with the following fix:

The fix I use must only be used on non-mission-critical databases (read no need for 99.99% uptime):

 - Stop all daemons accessing the database (in my case PowerDNS)
 - Dump the database in question (I only dumped the records table)
 - Truncate all tables (or just drop database and recreate)
 - Insert the data again
 - Restart daemons

This resulted in a drop from 4G to 200M and was done in under 30 seconds. However this will create downtime (as also will happen when doing a vacuum full), but it will free a lot of space and speed up the db again, this might be worth doing for quite some people out there.

Yay, Unilogic Networks is now an official Areca FTP mirror (the only one). You can get all of Areca's drivers, software and firmwares from our FTP server now.

Yes, you are reading it correctly. We are currently busy casemodding a Cisco 7206 chassis. This casemod is sponsored and done by Unilogic Networks B.V. It started while we where examining how to replace the fan tray of this 7206. Which actually is pretty easy, here are the steps to do it yourself:

 - Remove the power supplies from your chassis
 - Remove NPE, IO and PA's
 - Unscrew the 4 thumb screws located in the power supply bays
 - Pull the entire backplane out of the chassis.

You now have the backplane seperated from the main chassis. As you can see the fan tray is screwed to the backplane with 2 screws. Unscrew them and there you go ;)

UPDATE: The Cisco 7200 casemod project has been halted temporarily as we are currently waiting for our 12cm fans to arrive! The box in question we are gonna case mod is a Cisco 7206 with a NPE-150 network processing engine and a IO controller. No PA's inside as these are used in production boxes. It currently has no power supply but for testing purposes we use a redundant PSU from one of the production 7200's. Our target is to add 12 cm fans with coloured lights and a TL tube inside the chassis. It will be a fully functional case mod that can do production routing!

UPDATE: At 23 February 2003 we received our 3 AC Ryan Fans from www.casemoddingstuff.nl and so we could continue working on our Cisco 7200 casemod. (more)

As we are using our Areca controller in a 1U Supermicro chassis we had to use a hack to make it work. Normally the fan on the RAID processor will block because of the underlying PCI slot. We made a little hack to the controller so that it will float a little above the PCI slot.

To make sure our 'hack' continued to work I had to implement a script that checks the fan status. Here you can find such a script for nagios. It checks if the fan speed is not 0. If it's 0 it generates a CRITICAL error within nagios.

Well to do something useful with my day off I wrote a checker script for nagios which checks the state of your Areca RAID.

It is currently limited to checking volume sets on 1 controller only and when your raid does break it will only report that it has broken, not which disk actually broke.

You can get it here. It's a shell script which parses output from the cli tool. Maybe in the future I'll write something that links again Areca's API.

UPDATE : I have updated the script to work correctly when the raid array degrades. I was checking output of 'vsf info' but should be checking 'rsf info' as 'vsf info' will keep the device normal when it's degraded and only change state on a rebuild. 'rsf info' does the correct and wanted thing. Version 2 can be found here.

My parents cat died yesterday. I've lived together for 15 years with her and she always was a really nice companion. Bye Bye kiki :( I'll miss you :(