August 9, 201015 yr Since hard drive RMAs can often take weeks to be processed, it is best to replace the failed drive ASAP. That is why I like Seagate and WD so much- the have the fastest (and most fair) RMA around.
August 9, 201015 yr Since hard drive RMAs can often take weeks to be processed, it is best to replace the failed drive ASAP. That is why I like Seagate and WD so much- the have the fastest (and most fair) RMA around. My beef with Seagate is you have to pay $20 into the RMA for an advanced replacement. It's not just a hold on your CC, it's a $20 charge.
August 9, 201015 yr Plus, worst case if one dies I send it back for a new one and the data is not lost since it is a Unraid server. I thought that was part of the reason we have unraid, so when a drive fails that data can be rebuilt. True, but the longer you run in emulated mode (where a failed disk is being emulated by the parity disk and all the data disks) the greater your chances of losing data if another drive were to fail. This is why many people keep a spare drive handy so that they can minimize the chance of data loss. Since hard drive RMAs can often take weeks to be processed, it is best to replace the failed drive ASAP. Very good point. My current plan (because I actually don't have one) is if a drive fails I would just shut down my array until I have a replacement. It's a low tech solution that anyone can default to. I could always go out and buy a drive at a store if I just couldn't wait for the RMA to return. After that I'd have a cold spare. A thought occurs to me that 4 7200 rpm drives produce less heat than 6 5400 rpm drives. So if power and heat are really a concern then you must always maintain the largest sized disks possible. In reality most of us have a mixture of preferences that involve heat, power, coolness (as in bling), cost, reliability, past experience... I pick a good solid company and get to know it's products. Then I select those products based on some amount of research. I used to buy what was cheapest. That caused problems like too many failed drives. Yes, I could RMA them but I had to deal with multiple companies each with different procedures. My time is worth something afterall. Seagate is the largest company in the hard drive market and has the highest reliability ratings - it's just a fact. But even they have some bad products sometimes. I avoided buying the 7200.11 until they were reliable. I bought 7200.12 even though they only went up to 1TB. Now that 7200.11 are fixed I'm buying them. Do your research on the product, not just the company.
August 9, 201015 yr My current plan (because I actually don't have one) is if a drive fails I would just shut down my array until I have a replacement. It's a low tech solution that anyone can default to. I could always go out and buy a drive at a store if I just couldn't wait for the RMA to return. After that I'd have a cold spare. Yep, that's my plan too. I'm too cheap to keep a spare drive around, so I just figure I can live without the data on my server for a few days until a new drive arrives in the mail. Or if I were really desperate I would buy a new drive in town, but that would be throwing away at least $30. Interesting thought about the four fast drives versus six green drives, that makes sense to me. However, since the green drives tend to be cheaper anyway, I really see no point in using fast drives in a stock unRAID server. If you are doing extra fancy stuff with your server then sure, but for the general media-server type use that most of us do the fast drives just don't have much benefit, but the drawbacks (heat, power) are significant. I think the labels on WD green drives are pretty blingin' Edit: A thought I just had. Let's say you are taking queeg's and my approach and just buying a new drive when a drive fails. So you keep your server powered off until the new drive arrives. When the new drive does arrive, do you bother to preclear it? Does preclearing it put your other drives at any risk? Preclearing a replacement drive is definitely a good idea, since you really don't want your replacement drive to fail during the rebuild from parity. However, unless you have a test server, preclearing a drive means running your server for a long time. Would the solution be to just preclear the new drive while keeping the array stopped (but not powered down)?
August 9, 201015 yr Would the solution be to just preclear the new drive while keeping the array stopped (but not powered down)? That's what I would do.
August 9, 201015 yr I think the time it takes to preclear a drive is so small in terms of a hdd's life span that it would be fine in my book to do that before throwing it into the array. 2tb takes ~31 hours or so for one cycle, which isn't bad at all.
August 9, 201015 yr A thought I just had. Let's say you are taking queeg's and my approach and just buying a new drive when a drive fails. So you keep your server powered off until the new drive arrives. When the new drive does arrive, do you bother to preclear it? Does preclearing it put your other drives at any risk? Preclearing a replacement drive is definitely a good idea, since you really don't want your replacement drive to fail during the rebuild from parity. However, unless you have a test server, preclearing a drive means running your server for a long time. Would the solution be to just preclear the new drive while keeping the array stopped (but not powered down)? Before I deploy a drive, I always preclear it. Usually in another machine. Yet, I always preclear the spare before it is even considered a spare anyway. It is a good way to grab smart stats and re-assign and questionable sectors before handing them over valuable data. I'm too cheap to keep a spare drive around... My time, peace of mind and effort into possible reconstruction/acquisition is to valuable to not have a "precleared" spare around.
Archived
This topic is now archived and is closed to further replies.