• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Building NAS Reliability

I

Iratus

[H]ard|Gawd
Joined
Jan 16, 2003
Messages
1,583
Greetings,

I need some advice but first the situation that drove me to look.

Basically, just spent the best part of three days doing some heavy lifting to recover data from a RAID 5 after having a double failure on my Synology DS916+ NAS after having 2*3TB WD Red drives fail. It's been in degraded and 'crashed' state so I've been able to get some stuff off, but it's driving me to distraction.

Screw Ups
  • One disk was out of warranty and I didn't realise.
  • I extended the array from RAID1 to 5 with my second copy disk that holds the critical stuff before the array had finished syncing offsite.
  • I tried to add the first failed disk back to the pool after testing it, it hadn't shown sector or I/O errors and Synology showed it as "not initialised", from reading people seem to think the Synology's are a bit over sensitive and often have zero problems when faced with the same situation. That was not true for me.
You guys are pretty good hence why I'm posting here, it's amazing how many storage boards just have people giving lectures to people who have messed up. Well done, you're very smart, why don't you lecture me on how smart you are and how you've never had a problem you basement dwelling, socially maladjusted neckbeard. Though embarrassingly I've spent a very long time working in the Infrastructure space so I do know better.

Unfortunately, my sync to glacier was still going (plus it'd cost a fair bit to pull what is there), the disk I added to the array was older than I thought (though the same type), that failed, and the rebuild showed up a whole bunch of bad sectors on another disk during the rebuild so the NAS marked it as failing and went into limp mode.

I've had a look at the Backblaze data, my disks had a 3.5% annual failure rate, which explains some things; shame I never have that luck with the lottery.

Anyway, I'm looking at replacements.
  • HGST have empirical data showing they are the most reliable.
  • HGST Deskstar drives only have a 3 year warranty. Ultrastars have a 5 year warranty but are twice as expensive
  • Seagate have obviously had recent problems and seem to have a 3 year warranty on most things so I'm leaning towards them staying in the naughty corner.
  • Western Digital can die in a ditch with Maxtor and Samsung as far as I'm concerned. It's not necessarily rational (and they own HGST now) but there it is.

So it seems to be that HGST wins, I was running 3 disk RAID 5. 1 disk is in the bin. Another is getting replaced under warranty. That leaves me with 2*3TB Western Digital Red's.

I'm thinking that I get two 4TB drives, use Synology Hybrid Raid with 2 disk redundancy (it died during a rebuild so I'm wary of hot spares), if I run out of space I can either use expansion units (bizarrely expensive) or hopefully just keep replacing drives with bigger ones as they get to the end of their warranty.

1 alternative is to have a 6 TB drive in the NAS and keep 1 disk redundancy in the main volume but replicate to the 6TB. More expensive (though only $110) and not sure it gives me anything.

It will all be replicated offsite, probably to google coldline rather than Glacier.

I'll have a replacement schedule to replace the drives when they are within 3 months of warranty, won't help reliability much but if I ebay the old drives it'll save some anxiety.

Does this sound sensible? Any ideas for how I can improve things without going crazy expensive.
 
Your first problem was running raid 5. Do raid 6. When dealing with single large arrays raid 5 leaves you vulnerable during a rebuild.
 
Vengance_01 said:
Your first problem was running raid 5. Do raid 6. When dealing with single large arrays raid 5 leaves you vulnerable during a rebuild.
Click to expand...

Constructive help, thanks very much :rolleyes:

Anyone who actually read to the end?
 
I know a lot of people that have used Drobo, Qnap, Western Digital, Synology and other consumer type boxen. Internally with my company I have banned all of those for any data storage. We do some storage on EMC and Netapp, mostly from inherited systems. Everything that we are building is based around either Solaris or FreeBSD(FreeNAS) type of systems since single drives are storing more than 1TB currently we always run Raid 6 or RaidZ-2, usually having multiple arrays and warm spares in each box. If this was my system, I would do something like six 2TB drives in a consumer chassis running RaidZ-2 giving you an effective 8TB of available storage. You could take up to two drive failures and have no problems, also with this solution when you needed more storage space all you would have to do is swap the 2TB drives out for larger drives, and regenerate/grow the array, you wouldn't have any other re-configuration or growing of the array to have more space.
 
Hi, we noticed you’re considering Seagate drives and wanted to say thank you. It’s always nice to be considered! Our engineers are working hard to create the best drives on the market. We hear your concern on the reliability of Seagate drives. In case anything does not meet our own high standards, we are trying to work on our process so that we can resolve any issue in a customer focused and timely manner. We want to share this resource as a start which may address some of your concerns.

If you encounter an issue with one of our drives, you can always contact our customer support or look into any warranty information here. Please feel free to reach out if you have any questions!
 
You must log in or register to reply here.
Back
Top