read error [Help] Received Warning About Drive Read Errors

[JordanCarr42]

I'm using a fully updated version of unRAID and have received a warning about read errors on a particular drive. According to SMART the drive has passed however as far as I know a failure only really covers an extreme form of failure and not more subtle ones. I have backups of the truly important data and I have parity set up so I'm not immediately worried about catastrophic data loss. I also have checksums of all the files through a plugin which I periodically download and save, from which there don't appear to be any corrupted files at least yet. The warning still has me worried and I will attach the SMART report and as some other information on the disk tab under array devices the disk lists "errors" as 168 with all other disks as 0.

 

Any advice would be greatly appreciated and if I have made a mistake in the posting of this in terms of appropriateness to the forum or some other rule then I would appreciate being politely corrected and directed to the appropriate location.

 

Thanks

 

WDC_WD40EFRX-68WT0N0-20180705.txt

[seagate_surfer]

Hello JordanCarr42, 

 

These numbers don't look bad, as you may already know some of the SMART details are just informational. If a hard drive says 'I will die soon', some other details tells us that's about to happen, for an explanation on each one you can check the SMART attributes Wikipedia for more: https://en.wikipedia.org/wiki/S.M.A.R.T.#ATA_S.M.A.R.T._attributes

 

We sometimes can identify write amplification problems, dying sectors and so on. Checking the columns if you see the threshold is 0 the attribute is only an information character. If threshold is 253 the attribute is only for testing reason. 

A typical attribute set could be:

• Attribute name: "Read Error Rate"
• Current: 253
• Worst: 253
• Threshold: 63
• Raw: 0

Scales could be ranged up to 100, 200 or 253. Often higher values are better than lower values. Even if the statistic is something like "read failure rate", a value of 200 is GOOD, whereas lower values are BAD. That is, the value you see there is not the raw error rate but a number representing how GOOD your value is.

 

The threshold marks the value at which the hard drive could fail, and here is where you are in the "safe side" because luckily you have not reached the threshold in any of the statistics you have shown. All that matters is that the "worst" and "value" is still greater than the value in the THRES column. The THRES indicates how far the value of that statistic would need to drop to represent a warning.

 

I did pay attention to Reallocated_Sector_Ct   with a value of P for prefailure warning in your case, it still doesn't necessarily indicate a problem. Some statistics might be entirely informative, rather than indicating a problem. The sector reallocation is just as normal and it is taken into consideration by hard drive manufacturers, all hard drives will have some "room" for this and the reason why it happens is just because the drive is in use, the hard drive reallocates the sectors which pretty much means that the information was moved to another sector and this one won't be used any longer. Now with that said, keep an eye on that one because if you see a trend of sectors failing quite often and numbers increasing every time more and more then you could be experiencing, as mentioned earlier, Write Amplification problems or Overload workloads, if this is the case just pay attention to what programs are writing in your drives as you may have something that writes more than usual or something causing information to be written twice or three times unnecessarily, or you may have had installed a hard drive that is not for the application given and thus exceeding its Load Cycles.

[JordanCarr42]

Thanks for the information. The drive is a WD RED is a 6 bay NAS. I have btrfs set to copy on write so that would increase the overall utilization and writing to the drive but it shouldn’t be too bad. I’ll look into my docker containers and see if there are things writing more than I would expect. I’ll look more into the meaning of the individual readings and see what ways I can mitigate wear

[seagate_surfer]

You are right! Avoiding copying when possible reduces writing to the hard drives, tasks like watching movies, reading PDF files, or viewing photos doesn't count as writing; that's reading. Copying files from another drive, downloading, editing files, or backing up, and so on require you to write. One of the settings that can also help you lower the amount of writes is hibernation (Not to be confused with sleep mode, which won't affect writing). Hibernate uses less power than sleep and when you start up the PC again, you’re back to where you left off (though not as fast as sleep), The sleep mode lowers the power consumption to a point where data is not lost but during hibernation, the contents of the system memory where currently running programs and information still being processed reside, are written to the computer's internal storage device before the computer is powered off, which translates to writing on the internal storage a hibernation file that's about the size of RAM, e.g. A computer with 4GB of RAM requires some 4GB of storage space to create a hibernation file. When the computer is restarted, the system reloads the saved content back into the system memory, and hence restores the computer to the state it was in before it was turned off. Once loaded, the saved content is deleted from storage. This becomes a large amount of gigabytes of storage space over time that can be reduced to 0 by disabling this feature in your system since many computers go into hibernation by themselves.