Currently unreadable (pending) sectors

[Avena]

Anyone knows what's the meaning of this error on log?

Apr 20 20:10:51 freenas smartd[2598]: Device: /dev/ada4, 17 Currently unreadable (pending) sectors


Apr 20 20:10:51 freenas smartd[2598]: Device: /dev/ada4, 17 Offline uncorrectable sectors


Apr 20 20:40:52 freenas smartd[2598]: Device: /dev/ada4, 17 Currently unreadable (pending) sectors


Apr 20 20:40:52 freenas smartd[2598]: Device: /dev/ada4, 17 Offline uncorrectable sectors


Apr 20 21:10:51 freenas smartd[2598]: Device: /dev/ada4, 17 Currently unreadable (pending) sectors


Apr 20 21:10:51 freenas smartd[2598]: Device: /dev/ada4, 17 Offline uncorrectable sectors


Apr 20 21:40:52 freenas smartd[2598]: Device: /dev/ada4, 17 Currently unreadable (pending) sectors


Apr 20 21:40:52 freenas smartd[2598]: Device: /dev/ada4, 17 Offline uncorrectable sectors

 

[Jailer]

Just as is says, ada4 has 17 unreadable sectors. Are you running periodic smart test? I'd take a look at ada4 and see what's going on with it.

 

[Chris Moore]

The error message is entirely self explanatory. What is your question?

This typically indicates that the drive is going bad and needs to be replaced.

Hard Drive Troubleshooting Guide (All Versions of FreeNAS)
https://forums.freenas.org/index.ph...bleshooting-guide-all-versions-of-freenas.17/

 

[danb35]

The error message is entirely self explanatory. What is your question?

This. OP, not to pile on (though it probably sounds that way), but what part of the message were you having trouble understanding? There are a lot (seriously, a LOT) of threads here asking exactly the same question, and I've never understood how it isn't perfectly clear from the message itself. Perhaps you can help us understand what's less than clear about the message.

 

[Mike amrijo]

So, I'm going to answer danb35's question.
Let me go though some of my steps so its clear.
I'm currently on TrueNAS-12.0-U4 [have not updated yet]

I started receiving "Currently unreadable (pending) sectors" errors. " Humm. what does "pending" mean? Are these real errors?
I check the dashboard. I see an alert about the "pending" error but the pool is showing heathy.
All drives are showing heathy.
I look to the TrueNAS Documentation page hoping for a troubleshooting section or a description of the error message.
There does not seem to be one.

I start googling for answers.
I find a few pages that have a process of using dd to overwrite bad blocks on a disk.
[site - http://bytesandbolts.com/fixing-freenas-error-currently-unreadable-pending-sectors/ ]
[site - https://deadc0de.re/articles/unreadable-sectors.html]
- info I have not see on the TrueNAS site
Now I can [and did ] follow these directions but I still have questions.

I believe at this point I should replace the drive.. That is my plan and have one on order..

But TrueNAS still says everything is heathy. So what's the level of urgency here?
Am I looking at complete drive failure here or am I seeing something I can patch and live with a while?

So the error message has alerted me to a (pending) problem but I still don't know if (pending) means (urgent).

Personally I would love to see something in the Documentation side that lists error messages like this and recommended actions.

 

[jgreco]

I started receiving "Currently unreadable (pending) sectors" errors. " Humm. what does "pending" mean? Are these real errors?

Yes, unreadable sectors are real errors. They may be something that can be corrected. An overwrite is the classic way to find out.

I look to the TrueNAS Documentation page hoping for a troubleshooting section or a description of the error message.
There does not seem to be one.

That's because it isn't a TrueNAS message. It's coming from the SMART on your hard drive. It's one of many potential errors an intelligent drive can throw.

I believe at this point I should replace the drive.. That is my plan and have one on order..

Depends on your level of ZFS protection and your risk level acceptability. On a RAIDZ3 pool, for example, a pending sector developing is probably no big deal. On a pool with no redundancy, it represents potential data loss.

But TrueNAS still says everything is heathy. So what's the level of urgency here?
Am I looking at complete drive failure here or am I seeing something I can patch and live with a while?

A single pending sector is like your car not starting the first time you turn the key. It's unusual, it could very well be something amiss, but it could also just be a one-off nonissue. It's worth keeping an eye on and making a note of.

TrueNAS is mostly focused on how ZFS reports the health of the pool. The problem with that is, if you have ten 20TB HDD's but only 1GB of data on your pool, ZFS is only seeing the state of a fraction of a percent of your raw disk space, and can only detect problems with that fraction. There is no proactive scanning of unused disk space by ZFS. SMART is the classic way that this shortcoming is addressed. Two different mechanisms addressing two different-but-related disk health issues.

Personally I would love to see something in the Documentation side that lists error messages like this and recommended actions.

The developers have not been particularly interested in delving into this level of complexity. It's a bit of an art form, and they tend to avoid the ambiguous stuff, or paint over it a bit. The remediations for pending sectors are all "risky" in a direction that the developers don't seem to care to support. The classic "overwrite the sector and see if it comes back to life" is a bit like handing people a can of gas, a pack of matches, and saying "go burn down those weeds over there," because if not done properly, you can ruin a ZFS pool when you forget to put "count=1" on your dd options.

The official recommendation has typically been to replace the drive, though this may be a bit overly paranoid. It's certainly a great point at which to acquire a spare drive though, just in case things start going awry more quickly.

 

[NugentS]

Exactly. Buy a new drive and replace. Then trash the crap out of the old drive and see if it appears good. At that point either throw it away or keep it for a rainy day when another drive craps out. I have drawers full of spare drives, some of which have been swapped in just this scenario, I then badblocks the drive for a week and if its still running at the end it goes back in the drawer as potentially good and spare

 

[danb35]

Was this thread necro'd by a post that's since been deleted?

 

[NugentS]

Definately been necro'd - but looks vaguely valid

 

[Mike amrijo]

Yes, unreadable sectors are real errors. They may be something that can be corrected. An overwrite is the classic way to find out.



That's because it isn't a TrueNAS message. It's coming from the SMART on your hard drive. It's one of many potential errors an intelligent drive can throw.



Depends on your level of ZFS protection and your risk level acceptability. On a RAIDZ3 pool, for example, a pending sector developing is probably no big deal. On a pool with no redundancy, it represents potential data loss.



A single pending sector is like your car not starting the first time you turn the key. It's unusual, it could very well be something amiss, but it could also just be a one-off nonissue. It's worth keeping an eye on and making a note of.

TrueNAS is mostly focused on how ZFS reports the health of the pool. The problem with that is, if you have ten 20TB HDD's but only 1GB of data on your pool, ZFS is only seeing the state of a fraction of a percent of your raw disk space, and can only detect problems with that fraction. There is no proactive scanning of unused disk space by ZFS. SMART is the classic way that this shortcoming is addressed. Two different mechanisms addressing two different-but-related disk health issues.



The developers have not been particularly interested in delving into this level of complexity. It's a bit of an art form, and they tend to avoid the ambiguous stuff, or paint over it a bit. The remediations for pending sectors are all "risky" in a direction that the developers don't seem to care to support. The classic "overwrite the sector and see if it comes back to life" is a bit like handing people a can of gas, a pack of matches, and saying "go burn down those weeds over there," because if not done properly, you can ruin a ZFS pool when you forget to put "count=1" on your dd options.

The official recommendation has typically been to replace the drive, though this may be a bit overly paranoid. It's certainly a great point at which to acquire a spare drive though, just in case things start going awry more quickly.

Thanks for your reply -all great info!
[I'm running a Z2 array so I should be good]

The point of my post was a response to "Why is this question is constantly being asked?"

Simple answer. - Because it is not clearly answered in the documentation and the error message is a little confusing to new / less technical users.

Possible solution - add a trouble shooting section to the documentation that incudes the basic smart errors and basic recommendation.
example

Error -
Currently unreadable (pending) sectors.

Recommendation
Replace drive.