swamiforlife Posted September 29, 2021 Share Posted September 29, 2021 (edited) I have an array with total of 188TB of data 2 Parity Disks (10TB each) 24 Data Disks 2 - 8Tb disks 10 - 10Tb disks 12 - 6Tb disks I want to replace all 12 of the (6tb drives) with 16TB drives I will replace the Parity disks with 18TB drives. The array has used up 143TB and all the 6tb drives are over 80% utilized. What is the best method to make such a big change? Should i just backup all the data to external drives and start fresh? Or is there a better way? Edited October 15, 2021 by swamiforlife grammar Quote Link to comment
furian Posted September 29, 2021 Share Posted September 29, 2021 replace parity first. then do a 1 by 1 replacement of the 6tb disks. Quote Link to comment
ChatNoir Posted September 29, 2021 Share Posted September 29, 2021 1 hour ago, swamiforlife said: What is the best method to make such a big change? Should i just backup all the data to external drives and start fresh? Or is there a better way? There are several way to approach this. Each with its pros and cons. You would need to clarify your priorities if you want the best advices for you. For example, doing it one disk at a time would be very time consuming. You could also do a new config with the drive you want to keep and the new Parity drives ; then copy the data from the 6TB drives afterwards (from UD or the network). Could be faster but you will be unprotected for a time. Quote Link to comment
swamiforlife Posted September 29, 2021 Author Share Posted September 29, 2021 (edited) Well the number one priority for me is the safest way to do this. Meaning not losing or risking my data. The 2nd priority would be the quickest way to do this. Edited October 15, 2021 by swamiforlife Quote Link to comment
swamiforlife Posted October 15, 2021 Author Share Posted October 15, 2021 (edited) Anyone have any other ideas for this? Edited October 15, 2021 by swamiforlife Quote Link to comment
Frank1940 Posted October 15, 2021 Share Posted October 15, 2021 (edited) The thing I would do is: 1--- Replace the two parity disks with the 18TB drives. (If you have any tendency to paranoia, you might want to do one disk at a time...) 2--- At the juncture, pause and think a bit. You can't possibly need/use all of that additional storage instantly. Modern hard drives are (apparently) very reliable devices. The one thing to really be concerned about is buying into two possible scenarios: (A) a bad design change in a series of drives and (B) a bad manufacturing lot that has a manufacturing flaw that cause an abnormal high failure rate. (Both of these have happened in the past!!!) 3--- In light of the above, purchase and add each new 18TB data drive as you actually need them. (And always have a cold spare drive on hand!!!) Should you require more space instantly, you could do replace two disks at once. But if you do this, a single read failure on any of the 22 disks will cause the rebuilt to fail. (That is the reason why many folks use dual parity!) 4--- Another item to be considered is that one (or more) of your 10TB data drives may fail in the future before you have completed the conversion of the 6TB drives. (Doing this expansion, will require hundreds of hours of all the disks be spun up which will increase the possibility of a disk failure.) You would seem prudent to replace it with a 16TB drive rather then a 10TB unit. This would give you another 6TB of storage each time it happens. (Even if the 10TB drive were under warranty, you could always sell it.) I hope that you are running periodic parity checks on this server to (hopefully) check problems before they become major issues. (I often wonder how folks deal with the decrease in array performance during these parity checks when the checks times are over 24 hours.) On 9/29/2021 at 8:12 AM, swamiforlife said: Should i just backup all the data to external drives and start fresh? I am hoping this is a stupid observation on my part. You do realize that an Unraid server is not a fail-safe device and is not a replacement for a backup of irreplaceable data! That are a lot more ways to lose the data on a server besides hard disk failures. (Many of those ways involve physical damage/destruction of the server itself! With dual parity, the probability of data from these causes is higher than from disk failures on a well maintained server!) Edited October 15, 2021 by Frank1940 Quote Link to comment
trurl Posted October 15, 2021 Share Posted October 15, 2021 7 minutes ago, Frank1940 said: That are a lot more ways to lose the data on a server besides hard disk failures. (Many of those ways involve physical damage/destruction of the server itself! The most common way to lose data is probably user error. And, of course, parity won't help with that either. Quote Link to comment
itimpi Posted October 17, 2021 Share Posted October 17, 2021 On 10/15/2021 at 1:11 PM, Frank1940 said: often wonder how folks deal with the decrease in array performance during these parity checks when the checks times are over 24 hours.) A good reason to be using the Parity Check Tuning plugin so the check runs in increments outside prime time. The length of checks with modern large drives was one of the main reasons the plugin was developed in the first place. 1 Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.