Let me start out by saying I have a background in IT (more specifically, a Microsoft FTE as a Principal Software Engineer). I currently have an Unraid setup, and I want to love it-- but I am concerned with the current resiliency that is offered by your application. 

 

My use case:

- Unraid is used as a central backup system for all my machines at home

- Unraid is used to store critical files, that I concider important

 

I live in the Seattle area, and experienced power outages, that turned our electricity on, off, on, off. This inevitably corrupted my file system. Now.. I was able to get most of it up,,, except:

 

1. Docker -- I ended up having to rebuild my docker.iso file. Which is fine, but why can't the platform detect a file corruption and guide the user on fixing it?

2. I had corrupted data on one of my disks. I have double parity, but still, lost data. I read your warnings "do not do this for rebuilding or restoring data". I had zfs. there were no other easy options on getting my zfs store in a good state. Yes I lost data, but I think I can restore it from another location.

 

all in all, I think your team should spend a few sprints on identifying ways to automate resolving common issues for the user, and giving them more piece of mind on having a redundant configuiration-- and protecting against issues as I've described. 

 

Are you using a UPS?

 

 

11 hours ago, RoaringBark said:

I live in the Seattle area, and experienced power outages, that turned our electricity on, off, on, off.

As @foo_fighter already pointed out, a UPS is crucial if you have important data on your Server since it will protect you in such cases and gap the time where the power goes on, off, on, off.

 

I have a pretty beefy unit at home from Cyberpower and it will cover power outages for about an hour (up to two hours depending what services I disable in case the power is out).

 

You could look for example for something like this:

https://www.cyberpowersystems.com/product/ups/pfc-sinewave/cp1000pfclcd/

(Please note that this is a Pure Sine Wave model which in most cases is a bit pricier than a Simulated Sine Wave)

 

I always make sure the battery is user replaceable and that the runtime is high enough for my use case.

Most of manufacturers have runtime graphs on their websites.

 

In combination with the NUT plugin from the CA App you can define thresholds for runtime or battery percentage when the server is automatically shut down if the battery runs out of juice.