Indeed, system operations (a.k.a. a typeof(OPS)), seems to consume time and effort no matter how much one thinks one has it right. Don’t get me wrong, I’m far from being the absolute know-it-all, but I do try to get most things correct- well at least in the 1st 5 tries.

So, System OPS… what’s up, or sometimes down?

 I’ve been on a path to stop using my “heavy iron” (also known as the enterprise rack servers I have in my home lab) as a single server with a single OS installed. Instead I’m going down the path of doing VM servers (Linux or Windows) hosted on a Proxmox platform.

 For the most part I’ve moved away from VMWare and opted for Proxmox. This choice was based on the fact that I use Open Source or Freely available editions of these platforms, and I find that Proxmox offers a better usage of my “heavy iron”. The free license of VMWare will not scale the number of CPU’s beyond a certain point. I’m not complaining about that, it’s just how VMWare does their product. It’s a limitation based on them wanting one to buy an enterprise license. Again, not a complaint. VMWare is just not for me.

 Proxmox, yes I had to learn an entirely new platform. The journey has not been to bad as I already have a good foundation with VM’s based on my VMWare and Azure  experience. Not even the fact that Proxmox operates on a Linux base OS proves to be much of a challenge. After all, I do have extensive experience with Linux things (my office is littered with Raspberry PI’s, and I come from a modest background of being exposed HP Unix). 

 The main hurdle I’ve come across with Proxmox on my “heavy iron” is the disk setup. When one installs Proxmox, it is very clear that for file systems like ZFS, there is no support for Hardware based Raids. Well that poses a problem with my older servers that generally require that any and all disk drives be configured through the hardware raid controller. The older servers (HP DL380 Gen 8 and older) generally do not expose the hard drives to an OS unless a logic drive it setup first. This is not so much a problem with HP DL380 Gen 9 and above, because the typical raid controllers installed have an option call “HBA”- HBA is simply a pass-thru mode for the disk drives.

 With my older servers, without HBA, I’ve gone to a RAID 0 config (which is a simple stripped single drive). So far Proxmox has not had any issues with that. But I think for these, I’ll try to find a HBA controller that I can swap out for before push any major VM’s onto those older servers, or add them as a storage cluster. The main loss with this type of setup, that is not pass-thru, is that one loses S.M.A.R.T drive functions. That can be a thing if one wants to see failing drives stats in Proxmox, or wear stats for SSD’s. One will not see this for these types of drives.

 Now onto the path of pain… Before I stopped ignoring the warnings from Proxmox about using ZFS on a Hardware Raid, indeed YES my ZPOOLS on the ZFS storage went “belly up” and took my VM’s with them… The WOES of OPS.  I spend hours and hours getting those VM’s ready and enabled to host of all things… Word Press Web sites… oh like the one you are reading from!!! EPIC in the levels of failure… Computers make one humble at times. 😆 

 Lessons learnt… read extra carefully the warnings about Proxmox not supporting Hardware Raids for file systems like ZFS where one has one’s VM drives on that type of file system. Make sure one makes image backup of those VM drives at every major step in setting up a complex server VM.

 In the end, one learns by such things. I know far more about setting up a Word Press site on IIS from scratch than I ever wanted to… And on a positive note, I’m getting really good at Proxmox…

 Now back to our regularly scheduled program…

~SG