3

Let me start off with a short question on which I will elaborate below: What setup would you use for a home file server that is highly unlikely to lose data, but keeps the cost down as much as possible?

I am currently looking to upgrade the home server (details of the current setup are below). I'm looking for something that mainly can act as a NAS, but if it's not too much more expensive it would be nice if it could also host a Plex server, a VPN server and maybe some other small things. It will store documents, photos, music and videos, in total approximately 3Tb. My main concern is data integrety and secondly speed and availability. E.g. I wouldn't mind a week of downtime when a drive fails.

It seems it is recommended to have three copies of all the data of which one is off-site. In my case, the second copy would be on a server at my parents place, sychronized via the internet. The third copy I would create on a local external hard drive that is only connected during the backup.

The problem now is: how do I guarantee that the data is consistent across all drives and that no files have been changed? Locally this could be done using ZFS and its checksumming, but this requires several drives at a single location dramatically increasing the cost of the setup.

Another problem is how to handle versioning of the data. E.g., in case of emergency I would like to be able to access last months data. Also this points seems nicely covered by ZFS, but again only seems like a viable option with more drives.

Currently I'm using two Raspberry Pi 2's as a file server, each with one external hard drive attached. Both are in different location, host their own files and backup to the other over the internet using rsync. Versioning is done using hard links. At the time I put this together I figured that in the event that one of the servers went down I could simply take the other offline, buy a new Pi and HDD and copy the good drive to the new server. However, both drives are formatted with ext3 and I have recently come to the understanding that errors in the stored data might not immediately be noticed. This did alarm me as the whole point is to keep the data safe. Hence this question.

Also, at the time I did not take into account that during the recovery the only good disk would be stressed a lot, of course making it more likely to break having me lose all of my data. Using this reasoning I keep adding redundant drives everywhere, of course increasing the price a lot. This bring me back to the original question: at what point do you have enough redundancy in a home storage solution?

10
  • Is backing up to the cloud an option? For example, Google Drive, Amazon, etc? Aug 17, 2017 at 11:47
  • You can't so checksum sync with Rsync and the --checksum option. Having the cloud backup at the offsite / cloud at your parents, and having the local backup to the external would seem like enough to me. Use the tools you are already using to verify data integrity, etc. You should be able to use fsck to check the status of the ext3 formatted partitions. It sounds like you're only interested in retaining a 100% accurate and non-corrupt copy of the data in the event of a disaster, etc. and performance and availability is not of such a concern, I'd try to test and use something like that. Aug 17, 2017 at 12:26
  • In other words, for your home system backup solution you mention, it would seem to be efficient. I'd look into a solution if you have money to spend to have the redundancy like RAID drive configuration, etc. at the main file server level so that's the resilient system at that level, and the other parts are the backup system to recover data from in the event there is a disaster, etc. (e.g. file servers catch on fire, file server room falls into a sinkhole, flood, etc.). The backup data can be both cloud and external HDs onsite that you take offsite manually on a rotating schedule, etc. Aug 17, 2017 at 12:31
  • So you can even take the external HD backups offsite every day, etc. if you have two (or three) you swap for example. Perhaps when the backup jobs are complete for the day on the attached external drive, you swap out with another one before you head to work, and take that one to work with you or wherever you go so worst case, you can recover data from this timely and then still have the cloud backup to get newer data it may have copied back over once recovery is done. Be sure you practice recovery too if you are smart, confirm you can restore what you back up since that's the whole point Aug 17, 2017 at 12:37
  • @JohnStoneman The cloud would be an option, but seems to be rather expensive for 3TB compared to setting up something myself. Also, when going this route I would still need a separate server for hosting other services.
    – Octaviour
    Aug 17, 2017 at 14:09

0

You must log in to answer this question.

Browse other questions tagged .