2

I discovered one of my servers running slower than usual today because of the checkarray launched by a cronjob installed by default on the server. This is Ubuntu 14.

In this particular case, I have just a single RAID1.

What is the purpose of this check? To flag a disk as faulty in case that checks turn out wrong and in the hope to minimize future problems while waiting to total failure?

How should I configure this checks to better fit my setup? If I should do that at all.

Is there a why to run this "nicer" that the idle mode? I am a bit concern about performance while system is on production due to intense I/O operations.

1 Answer 1

3

What is the purpose of this checks? To flag a disk as faulty in case that checks turn out wrong and in the hope to minimize future problems while waiting to total failure?

Pretty much. Without the check, the array will continue to operate until it encounters a fault. Issuing the check to the array periodically is done so that you can preemptively find these faults in advance, instead of waiting for them to happen. Or so the theory goes...

Is there a why to run this "nicer" that the idle mode? I am a bit concern about performance while system is on production due to intense I/O operations.

You can limit the I/O rate of mdadm to something that isn't overwhelming, although you are trading time for performance.

Or just schedule it for night, or some other time when the system isn't nearly as busy.

2
  • 1
    With the size of drives these day it can often take many hours to scan. The 4TB Raid1 in my home box takes ~16 hours.
    – Zoredache
    Feb 10, 2015 at 20:52
  • Ditto, I have 2x WD 3Tbyte Reds in a RAID 1 mirror. The check is done every Sunday evening, in the wee hours. It typically takes 8-10 hours. Feb 11, 2015 at 0:50

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .