Drive replacements - how often? What circumstances?

[norbs]

Just trying to get an idea of what a good practice is with replacing your pool HDDs.

Does anyone ever replace drive before there is any signs of failure and not as part an upgrade?

I've got drives that are getting near 6 years of being powered. So far replaced 2 of the 4 disk in the pool (RAIDZ2). Should I let he other 2 start showing S.M.A.R.T. failures and let them live out their lives or proactively replace them?

 

[Ericloewe]

No. If they work, why bother?

 

[Spearfoot]

I run 'em until they either fail or start throwing SMART errors, particularly:

• 5 Reallocated Sector Count
• 197 Current Pending Sector
• 198 Offline Uncorrectable

I've got one WDC Green drive w/ over 42,000 hours on it; I intend to run it until it stops! ;)

 

[wblock]

In theory, proactive replacements could make for more predictable uptime. In reality... well, there might be statistics on that, but they would likely only apply to large groups of drives.

 

[norbs]

I run 'em until they either fail or start throwing SMART errors, particularly:

• 5 Reallocated Sector Count
• 197 Current Pending Sector
• 198 Offline Uncorrectable

I've got one WDC Green drive w/ over 42,000 hours on it; I intend to run it until it stops! ;)

That's useful.
I was planning on just waiting on either complete failure or SMART errors. I just wanted to make sure I wasn't being reckless or foolish.
Appreciate the details.

 

[Dice]

Have a look at the resource section - there is topic on trouble shooting hdds and SMART data.
It is a very useful read to equip yourself with an understanding of what to keep an eye out for.
Also, even if SMART doesn't always tell the full story, you could familiarize yourself with the solnet array test (also on the resource section) - which basically provides a test to check drive speeds, spotting problematic drives wrecking havoc on your performance.

 

[Stux]

Obviously you need to have scrubs enabled if you're planning to run your drives into the ground (which is what I do)

I go so far as to 'repair' some of the smart errors to keep drives in service. If they're not in warranty

 

[Alan W. Smtih]

Dice - Thanks for pointing to the guides. I hadn't run into those yet.

Since I just looked them up, here's quick links for anyone else hitting this thread:

- Hard Drive Troubleshooting Guide (All Versions of FreeNAS) - Basic information on how to troubleshoot hard drives using SMART data

- solnet-array-test - A shell tool for performance testing and burn-in of a disk array

 

[danb35]

I was planning on just waiting on either complete failure or SMART errors.

I'm trying to remember the last time I had a hard failure of a disk that was part of a live pool, and I'm coming up blank. Must be at least five years. I do run them until I see SMART errors, though. Between regular self-tests and email notification of errors, there seems to be enough warning of impending failure.

 

[Stux]

Used to happen a lot more with Seagate 1.5 and 3TB drives ;)

 

[danb35]

Used to happen a lot more with Seagate 1.5 and 3TB drives ;)

Wasn't it @DrKK who said that the ideal storage configuration for those drives was buried as deep as possible in the nearest landfill?

 

[Dice]

Wasn't it @DrKK who said that the ideal storage configuration for those drives was buried as deep as possible in the nearest landfill?

That is confirmed :P

 

[norbs]

Around the time when Seagate was racing to be the first 1TB drive most their stuff was terrible. I'd get brand new drive continuously clicking for no known reason. This was true from about 750GB all the way to 1.5TB and maybe early 2TB.

But to be fair the four ~6 year old drives that are just now starting to show some signs of age (mostly SMART errors) were 2TB Seagates. I replaced two because of SMART errors but they still worked.

 

[Stux]

Oh, now you're bring back baracuda bad memories that I'd managed to repress :(