Next Up – Cache drive upgrades

      2 Comments on Next Up – Cache drive upgrades

Turns out, I’m outgrowing my cache drives.

It was going to happen eventually, so I figured I would begin to address the issue now.

Seems that – after several years of data accumulation – 500GB is no longer cutting it for cache.  There are several reasons for this, which require knowledge of how “cache” is typically utilized within Unraid:

  • “Appdata” – which is the configuration data for docker containers.  Depending upon the container, the amount per container can be minuscule (currently 8K for my smallest container) or massive (currently 100G (and growing) for my largest container).
  • Virtual Machines – By default, the system drive for each VM resides on the cache drive (and face it, for performance reasons we wouldn’t want those file systems on the array).
  • “System Stuff” – This includes the docker image (which is currently sitting at 16GB) as well as “libvirt.img”, which is sitting at 1GB.
  • Cache – space where files land when they are copied to Unraid shares.  The theory here is that files can be copied to the system array faster by adding them to cache first, and then later on a mover process executes to move these files to the array.  In my case, the mover runs once a day, in the middle of the night sometime.  For the most part, however, I don’t have caching enabled on my shares because I don’t really need it that much.

As you can see in the accompanying screen shot, I am currently consuming 375GB of the 500GB available.  We never want the free space to be anywhere close to zero, due to the primary caching function noted above.  I have therefore found myself in the unenviable position of not having enough space available to comfortably deploy more VMs, which is kind of a drag because I need to upgrade some of them soon and I generally do that by creating new VMs with updated code on them and then copy my configs over to them.  Just for the record, there are currently 4 VMs provisioned on the server, with a combined consumed space on the cache of 107GB.

With corporate America gobbling up all semblance of technical resources (storage, RAM, you name it), I figured if I was gonna jump I better do it now.  With that in mind, I went drive hunting online, ensuring that I limited results to reputable sellers to reduce the likelihood that I would receive a counterfeit drive instead of what I actually ordered.  During this process, I came across many results that limited me to “one drive per household”, which was not at all ideal since my cache pool consists of two drives in a RAID1 mirroring configuration.

I finally came across this 2TB drive on Newegg.  There was a 5-drive limit, so I bought 3 (I have an unrelated use case for the 3rd drive which I’ll get into another time).  They have since shown up on my doorstep and I’m now planning to do some tinkering on my server at some point in the near future.  Of course, I will detail the pitfalls and victories here as they happen!

2 thoughts on “Next Up – Cache drive upgrades

  1. phil

    Well this is great info! Is this specific to unraid, or cache drives in general? My NAS don’t currently support them anyway, but my next upgrade will for sure. Had no idea they needed so much space.

    Reply
    1. Mike Post author

      Can’t speak to anything other than Unraid, well I guess I could speak for Proxmox… I’m not sure if PM supports cache but I can tell you I am not using it. Then again, I’m also not using any form of RAID on my PM server so it is likely a moot point. My understanding that cache only really comes into play when you have a need for a storage medium that is not slowed down by having to write every change to parity. That’s huge overhead when there are consistent writes, such as swap, databases, and such.

      Reply

Leave a Reply to Mike Cancel reply

Your email address will not be published. Required fields are marked *