I recently purchased a Unifi UVC-G3 network camera for the outside of my house. It works great in standalone mode, but has no recording capability without a Unifi NVR. The Ubiquiti hardware one is about $350, but a lot of people just repurpose a small linux box to do the job. That, to me, seemed silly since I have a Mac Mini that serves as my home media server, and runs my weather station (and a few other things). I also have a Raspberry Pi in the garage that runs Cacti (for network monitoring), and the Unifi wireless controller. This all got me thinking, “wait, I could just run the free version of VMWare ESXi (vSphere) and virtualize all of this on the Mac Mini”. The problem was, was that really possible? I know Apple allows virtualization of it’s OS on Apple hardware, and there are hacks to get it to work on non-Apple hardware, but I’d never actually managed a vSphere server before. Sure, I have tons of VMs at work, but the system is managed by another group, and all the VMs are Windows or Linux. So, I had to try.
Migrating Crashplan ProE
For about 5 years now, I have run a Crashplan ProE (at the time, just Crashplan Pro) server at my office. It ran on a server here in town, then a remote server as well (for offsite backup). This server largely stalled out at version 3.6, and has worked quite well, but had limited storage, so we basically only backed up the bare necessities (Documents and Desktop, no pictures, music, videos, etc). Which, while fine, isn’t exactly what Journalism faculty expect when they generate a lot of photos, videos, etc. =/
However, a couple years ago, central campus had enough pressure to start up its own CrashPlan offering. They chose to utilize CrashPlan’s cloud storage offering, which is great. It meant unlimited storage, at the cost of about $90/user (with up to 5 computers per user).
Repairing two Ubiquiti Toughswitch Pro 8’s
I recently purchased two listed-as-broken Ubiquiti Toughswitches off eBay for $65. They were listed as not working, unable to boot. They admitted very little troubleshooting had been done to the units, but I couldn’t pass up the price since I’d been looking for one of these switches, but they sell for more than I wanted to pay, and a friend also wanted one, so I figured I’d sell it to him if I got both working. Buying them, I figured there were two likely causes. One being a firmware corruption, and they just need to be re-flashed using their tftp mode, or two being the power supplies had failed (power surge, lightning strike, wear and tear, etc). I know from research that these switches have an internal AC/DC power supply that supplies the switch board with 48V, which I can replicate easily enough on my bench. It seems it’s not that uncommon for people to replace those internal PSUs with DC/DC converters for use in wireless installs that only have DC available, but that’s not what I’ll be dealing with here, hopefully.
The unit’s were delivered on Memorial day (oddly), and initial inspection showed no power up at all, and no ground continuity between the ports and the power ground. Huh. I opened both units, and found the failure modes to be completely different. On one unit, the power supply was dead. The fuse had blown (easy to replace), and the main SMPS chip, a TEA1755T was cracked. The failure was bad enough that it blew the trace off the ground connection. So I ran some 20AWG wire from the ground to the screw so there was a good ground again. I ordered and replaced the TEA1755T and the power supply started working again. Sadly, the switch board also had an issue, but more on that in a bit.
On the second unit, the PSU was good, but the LM5005 buck regulator that takes the 48V input from the PSU and converts to 24V was blown (complete with a hole in it). I ordered a replacement and replaced it, but man, TSSOP chips suck to replace. Once that was replaced…
One the first unit again, supplying power to it (even with a bench PSU), results in a hissing sound from the Buck Regulator section, but a clean 24V output from the regulator. But, none of the unit starts up, as I don’t see anything on the main oscillator, and the unit certainly doesn’t boot. Tracing down where the 5V/3.3V rail comes from was a bit of a challenge, but it originates from…