This was somewhat of a shocker and I fully acknowledge I may be missing some info here during my limited 2 day research debugging why an app was binding to localhost instead of waiting for DHCP to do it’s thing even though it had After=network.target defined in it’s systemd service file.

Worth noting:

Proxmox v4 Beta as of right now uses linux kernel v4. I updated my homelab to it recently & PCI passthrough (per my previous post) stopped working. Asked on the Proxmox forums & apparently an additional option is required due to the newer v4 kernel:

I’ve come to the realization that I may be a virtualization junkie. I find myself more interested in trying out different virtualization platforms than I am the end result of using the platforms for the VMs themselves. My last all-in-one iteration of Proxmox and FreeNAS worked great, but before long I got the itch to try something new. A fellow IRC’er was telling me about SmartOS, so I decided to give it a look.

Why?

After using my VMware/NexentaStor All-In-One for a while, I grew tired of VMware’s bloat & limitations. Doing “cool stuff” in VMware requires a license, & vSphere Client only runs on Windows. I got tired of starting up a Windows VM just to manage my hypervisor. That’s the only thing I started Windows up for, and it got old. I wanted something I could manage directly from my primary OS, OS X, as well as lightweight & preferably open source.

Intro

There are a million “Migrating from Wordpress to Octopress” blogs out there. I didn’t want to be yet another one, but I had a few thoughts I wanted to get out there. Most Octopress blogs I see out there are owned by developers, and I’m a sysadmin. I had been using Wordpress for a long time, & Blogger before that. Octopress is… different.

Following up on my last post, Netatalk on NexentaStor, I mentioned there that I only use Time Machine for “bare metal” type backups.  As in if I need to restore something specific to the OS itself, or reload the OS in event of a hardware failure.  I use CrashPlan for all my actual data backups, because I think it’s superior to Time Machine.  For why I think that, check out the details.  I’ve also had Time Machine crap out on me more than once when the disk gets full & tries to expire too much data, the only solution being to wipe the drive & start backups over.  Not cool.  Never had that problem with CrashPlan.

My 3 year old son has been using my wife’s iPad since he was 1. He’s pretty good at it & we think it adds some valuable child development skills. Of course, we use a heavy duty case, but we haven’t really had an instance where he tried to break it. It’s great for things like Netflix for Kids & playing Angry Birds, but also educational apps that makes learning how to count & spell more fun. He’s doing pretty well in those areas so we think it’s been worth it. However, we have to be careful not to let him get too addicted to it. We kinda fell into this rut a little, especially since our second son was born in December & it’s easier to let the 3 year old keep himself occupied with the iPad. We’re scaling back, but in the process I looked for ways to make limiting usage a little easier.

NexentaStor is NCP (Nexenta Core Platform) underneath, so you really do have all the power & flexibility of an open unix system.  That’s one of the reasons I love the project.  If you want to add on additional functionality, it’s not so hard to do so.  Everything I’ve done to “extend” NexentaStor to my liking has not interfered with the core NMV/NMC functionality.  That said, one of the great things about apt in Solaris is apt-clone; something gets messed up, you can revert your system back leveraging the power of ZFS snapshots.  So, when installing a lot of stuff using apt, use apt-clone instead of apt-get.

One of the things I wanted to use all that disk space I have in my ZFS/ESXi All-In-One for is Time Machine Backups for the 3 Macs in my house.  I use a combination of Time Machine & CrashPlan for my backups.  Yes, I’m using CrashPlan on NexentaStor as well; that’s a future post.

Prelude

To (finally) follow up on my original post, ZFS/ESXi All-In-One, Part 1, this post will go over how to configure ESXi & NexentaStor to work with each other, all from within one physical server. Typically, ESXi will connect to a physically separate server or appliance that provides storage. For production environments, this is preferred. For testing/development, it can be prohibitively expensive. The All-In-One solution provides a good alternative.

First off, this post will help alleviate some of the guilt I’ve been having for neglecting my blog.  Second, I just wanted to write it down for my reference.  There’s probably a better way to do this but this worked for me.