I have a server that I’ve owned since 2010. It’s a quarter-depth 1U server that I owned before I even owned a rack (which I do now, but that’s a different story). It’s been running my home network for over 15 years without issue.
A couple of weeks ago, I received a notification that the hard drive was getting “pre-failure” warnings on the daily SMART tests. Fair enough; everything in this tiny thing is 15 years old and a good ol’ fashioned spinning-disk drive is bound to fail eventually.
I sourced a reasonably-priced replacement harddrive with the intention of installing a fresh Ubuntu system. 15 years ago, this server had started life as a Ubuntu-based distribution called ISPConfig because I wanted something that could offer a web-based dashboard that I could use instead of relying solely on SSH. I scrapped most of the “ISPConfig” stuff almost immediately after realizing just how much I actually didn’t need this particular flavor of dashboard, but there are some remnants still lurking deep in the bowels of this tiny system.
As part of the process for installing a new system, I wanted to get a bootable USB drive attached with the ubuntu-server installer. Unfortunately, inserting a USB drive into either of the rear USB ports resulted in no device detection. I opted to try the internal USB port, thinking that maybe 15 years ago, I disabled the rear-panel USB ports on purpose (I honestly don’t remember).
No good deed goes unpunished. What should have been a quick 3 or 4 minutes to unmount from rack, remove 4 screws to remove the top of the case and insert the USB drive turned into an hour of troubleshooting because, as I came to find out, the power supply appears to have failed.
So on top of the $25 new HDD, I now have a $25 replacement PSU on its way. At least the parts are cheap. I also picked up another 2GB of RAM, to bring it to the max of 4G, for about $15.
In the meantime, I built a VM on my proxmox cluster that is temporarily handling all of the services the dead server normally provides- and in the process remapped those services onto floating IPs that are dynamically activated by my L3 switch (ip sla-based static routes, which I never had the option to use before until I got a new switch a few months back)