The time has come to sort out our Apple Xserve G5s in an effort to make them useful. Given the experiences I’ve had with OS X Server, and given that Leopard Server doesn’t even seem to be usable in an office environment, I’m certainly not going to consider trying to use that in a hosting environment. Also, I want these machines to be set up as consistently as possible with our live servers, which are running CentOS Linux.
Unfortunately there are no recent or current versions of CentOS for PowerPC, so I’m using Yellow Dog Linux 6.0 instead. Fortuitously, Yellow Dog is also based on Red Hat, so I didn’t have to tweak my highly customised Kickstart build too much for it to work on Yellow Dog.
Unfortunately, the Yellow Dog kernel doesn’t seem to have been compiled with support for NIC bonding, so NIC redundancy isn’t possible without compiling a custom kernel, which is overkill just for two servers. Also, there’s no RAID controller in the servers, so I can’t use hardware RAID. And then I found that it doesn’t seem to be possible to set up software RAID on an Apple server in the way I’d like due to limitations in the boot loader.
Anyway, having installed Yellow Dog on one of them, I wanted to figure out how to monitor the hardware so I can get alerted in the event of fan failures etc. The only option for this seems to be the lm_sensors package, which, disappointingly, doesn’t appear to have support for the sensor hardware in the Xserve G5s. At least SMARTD works OK, so I will get notified if any of the disks start misbehaving.
So, all in all, this is a pretty frustrating situation. I have two servers that need to be used but which have poor redundancy and no way of telling me when some of their components have failed. Really not ideal.