Configuring APC Rack PDUs

A brand new APC should be ready to have its IP set via using an ‘arp and ping’ technique. Its in the manual, but worth noting, since I normally google search before RTFM! And so does most of the world..

  1. On any running host on the same network, set the ARP for the MAC address of the device. Eg: arp -s 00:C0:B7:76:08:CC will tell the host you are on to send to hat mac address when you use the IP
  2. On the same host, before the arp expires (normally 5 mins), send a ping to this IP with a special size of 113 bytes. Eg: ping -l 113 from Win, or ping -s 113 if you have a better operating system (Linux)!

If this fails, then you need to get the serial cable out and wire it up (9600 8N1 is default I think, turn off ALL flow control; no hardware, no software). Password recovery: find the small hole and use a paper clip or similar to reset the device. You have to reset it twice for it to take effect, and then you can use the default username and password of apc/apc.

Modprobe, kernel 2.4 and 2.6

Random note: getting the bonding module to work, edit /etc/modprobe.d/bonding (a new file) and add:

alias bond0 bonding
options bond0 miimon=100 mode=active-backup

If you are using a 2.6 kernel, then the bonding module will be loaded with the correct options when you try to bring up the interface. Referencing the interface by name triggers the loading of the module, and the reading of this config file. However, and I talk to myself here since I forgot today, this doesnt work with the old 2.4 kernel modutils package. Only the new module-init-tools package when using a 2.6 kernel!

To complete the rant here, populate your /etc/network/interfaces file with something like the following, adjusting the address as required of course:

auto iface bond0

iface bond0 inet static

pre-up ifconfig bond0 up

pre-up ifconfig eth0 up

pre-up ifconfig eth1 up

pre-up ifenslave bond0 eth0 eth1





down ifconfig eth0 down

down ifconfig eth1 down

Now, if you don’t know which intrefaces you want to bond together into a bond (what Windows calls a ‘team’, or ‘teaming’), then you can take a look at Nick Ferrier‘s solution involving doing a quick DHCP request on each interface and working out which interfaces are on common networks; a very clever piece.

So, if you’re still reading this about bonding, you should probably be monitoring your bonded connections. Otherwise, one link will fail, and you’ll never notice! Check out my SVN repository for the Nagios plugin to check the /proc/net/bonding/* file(s).