How often should Smart and Scrub tests be done?

[mike360x1]

First, I want to apologize in advance if there is a similar thread to mines that addresses this problem. I recall finding a thread similar to mines but I can't seem to locate it.

In any case, What kind of schedules do you guys have for reoccurring scrub and S.M.A.R.T tests? And what kind of setup do you have going on for your array? Do you stagger your smart tests for each disk?

A monthly scrub test with a bi-weekly long smart and a weekly short smart test is what I'm thinking.
The Array I am running is a 5x3TB WD Red setup in RAIDz2. The freenas machine is new so I don't have much experience with these tests.

Thanks in advance,
Michael L.

 

[Ericloewe]

Standard practice is twice a month each, on alternate weeks (scrub -> SMART long test -> scrub -> SMART long test -> ...).

 

[danb35]

A monthly scrub test with a bi-weekly long smart and a weekly short smart test is what I'm thinking.

I would consider that the minimum. The range I've seen here is short tests every 1-7 days, long tests every 1-4 weeks, and scrubs every 2-4 weeks (being careful not to overlap when the long tests are happening). My own schedule is on the short end of that--short test daily, long test weekly, scrub every two weeks.

 

[Bidule0hm]

I have the schedule I use in the Misc section in this post: https://forums.freenas.org/index.ph...d-identification-and-backup-the-config.27365/ ;)

 

[BigDave]

A monthly scrub test with a bi-weekly long smart and a weekly short smart test is what I'm thinking.
The Array I am running is a 5x3TB WD Red setup in RAIDz2. The freenas machine is new so I don't have much experience with these tests.

Here's my settings for a 6 disk (3TB) RAIDz2 volume


How often you run the tests is purely a personal thing, I just tried my best to make sure the scrubs and tests don't conflict with each other.

 

[mike360x1]

Ok cool, so the consensus is bi weekly scrub and long smart test. and weekly short smart tests.

I won't be using my server for any critical applications or intensive tasks of any sort. I don't know if that makes the possibility of bit flips or other errors less likely to occur.

I might stick to monthly scrubs and bi weekly long tests. Unless there is a strong reason to do otherwise :)

 

[rpsip]

Here's my settings for a 6 disk (3TB) RAIDz2 volume
View attachment 12174
View attachment 12175
How often you run the tests is purely a personal thing, I just tried my best to make sure the scrubs and tests don't conflict with each other.

I know this is an old thread, but I'm revising my scrub schedule and was trying to figure out what others are doing. Mike360X1 is expecting to run his long/short SMART tests on different nights than his scrubs. He is attempting to do this by specifying the days of the month, but also specifying the days of the week. However, according to FreeNAS documentation, the Day of Week

Code:

field and the Day of Month field are ORed together: setting Day of Month to 01,15and Day of week to Thursday will cause scrubs to run on the 1st and 15th days of the month, but also on any Thursday

If I apply what the documentation says to the way Mike has set it up, Mike's scrubs are being performed on every day of the month - even on days he has not specified using the Day of Month field. This is not what Mike (or most other users) would expect. Am I misunderstanding, is Mike misunderstanding, or is the documentation incorrect?

The way to prevent this would be to not require the Day Week of Field when Day of Month is specified. However, Day of Week is required no matter what Day of Month is set to. Seems like a bug?

 

[BigDave]

The picture above is of my current SMART settings in the GUI Tasks tab,
the output below is the SSH results for upcoming SMART tests.
Please note that if you specify days, that Days of Week must all be checked or tests will be skipped
due to dates occurring on different days of the week throughout a 12 month period.

Code:

Next scheduled self tests (at most 5 of each type per device):
Device: /dev/da0 [SAT], will do test 1 of type S at Mon Jul  9 01:19:34 2018 CDT
Device: /dev/da1 [SAT], will do test 1 of type S at Mon Jul  9 01:19:34 2018 CDT
Device: /dev/da2 [SAT], will do test 1 of type S at Mon Jul  9 01:19:34 2018 CDT
Device: /dev/da3 [SAT], will do test 1 of type S at Mon Jul  9 01:19:34 2018 CDT
Device: /dev/da4 [SAT], will do test 1 of type S at Mon Jul  9 01:19:34 2018 CDT
Device: /dev/da5 [SAT], will do test 1 of type S at Mon Jul  9 01:19:34 2018 CDT
Device: /dev/da0 [SAT], will do test 1 of type L at Wed Jul 11 01:19:34 2018 CDT
Device: /dev/da1 [SAT], will do test 1 of type L at Wed Jul 11 01:19:34 2018 CDT
Device: /dev/da2 [SAT], will do test 1 of type L at Wed Jul 11 01:19:34 2018 CDT
Device: /dev/da3 [SAT], will do test 1 of type L at Wed Jul 11 01:19:34 2018 CDT
Device: /dev/da4 [SAT], will do test 1 of type L at Wed Jul 11 01:19:34 2018 CDT
Device: /dev/da5 [SAT], will do test 1 of type L at Wed Jul 11 01:19:34 2018 CDT
Device: /dev/da0 [SAT], will do test 2 of type S at Fri Jul 13 01:19:34 2018 CDT
Device: /dev/da1 [SAT], will do test 2 of type S at Fri Jul 13 01:19:34 2018 CDT
Device: /dev/da2 [SAT], will do test 2 of type S at Fri Jul 13 01:19:34 2018 CDT
Device: /dev/da3 [SAT], will do test 2 of type S at Fri Jul 13 01:19:34 2018 CDT
Device: /dev/da4 [SAT], will do test 2 of type S at Fri Jul 13 01:19:34 2018 CDT
Device: /dev/da5 [SAT], will do test 2 of type S at Fri Jul 13 01:19:34 2018 CDT
Device: /dev/da0 [SAT], will do test 3 of type S at Sat Jul 14 01:19:34 2018 CDT
Device: /dev/da1 [SAT], will do test 3 of type S at Sat Jul 14 01:19:34 2018 CDT
Device: /dev/da2 [SAT], will do test 3 of type S at Sat Jul 14 01:19:34 2018 CDT
Device: /dev/da3 [SAT], will do test 3 of type S at Sat Jul 14 01:19:34 2018 CDT
Device: /dev/da4 [SAT], will do test 3 of type S at Sat Jul 14 01:19:34 2018 CDT
Device: /dev/da5 [SAT], will do test 3 of type S at Sat Jul 14 01:19:34 2018 CDT
Device: /dev/da0 [SAT], will do test 4 of type S at Sun Jul 15 01:19:34 2018 CDT
Device: /dev/da1 [SAT], will do test 4 of type S at Sun Jul 15 01:19:34 2018 CDT
Device: /dev/da2 [SAT], will do test 4 of type S at Sun Jul 15 01:19:34 2018 CDT
Device: /dev/da3 [SAT], will do test 4 of type S at Sun Jul 15 01:19:34 2018 CDT
Device: /dev/da4 [SAT], will do test 4 of type S at Sun Jul 15 01:19:34 2018 CDT
Device: /dev/da5 [SAT], will do test 4 of type S at Sun Jul 15 01:19:34 2018 CDT
Device: /dev/da0 [SAT], will do test 2 of type L at Tue Jul 17 01:19:34 2018 CDT
Device: /dev/da1 [SAT], will do test 2 of type L at Tue Jul 17 01:19:34 2018 CDT
Device: /dev/da2 [SAT], will do test 2 of type L at Tue Jul 17 01:19:34 2018 CDT
Device: /dev/da3 [SAT], will do test 2 of type L at Tue Jul 17 01:19:34 2018 CDT
Device: /dev/da4 [SAT], will do test 2 of type L at Tue Jul 17 01:19:34 2018 CDT
Device: /dev/da5 [SAT], will do test 2 of type L at Tue Jul 17 01:19:34 2018 CDT
Device: /dev/da0 [SAT], will do test 5 of type S at Thu Jul 19 01:19:34 2018 CDT
Device: /dev/da1 [SAT], will do test 5 of type S at Thu Jul 19 01:19:34 2018 CDT
Device: /dev/da2 [SAT], will do test 5 of type S at Thu Jul 19 01:19:34 2018 CDT
Device: /dev/da3 [SAT], will do test 5 of type S at Thu Jul 19 01:19:34 2018 CDT
Device: /dev/da4 [SAT], will do test 5 of type S at Thu Jul 19 01:19:34 2018 CDT
Device: /dev/da5 [SAT], will do test 5 of type S at Thu Jul 19 01:19:34 2018 CDT
Device: /dev/da0 [SAT], will do test 3 of type L at Mon Jul 23 01:19:34 2018 CDT
Device: /dev/da1 [SAT], will do test 3 of type L at Mon Jul 23 01:19:34 2018 CDT
Device: /dev/da2 [SAT], will do test 3 of type L at Mon Jul 23 01:19:34 2018 CDT
Device: /dev/da3 [SAT], will do test 3 of type L at Mon Jul 23 01:19:34 2018 CDT
Device: /dev/da4 [SAT], will do test 3 of type L at Mon Jul 23 01:19:34 2018 CDT
Device: /dev/da5 [SAT], will do test 3 of type L at Mon Jul 23 01:19:34 2018 CDT
Device: /dev/da0 [SAT], will do test 4 of type L at Sun Jul 29 01:19:34 2018 CDT
Device: /dev/da1 [SAT], will do test 4 of type L at Sun Jul 29 01:19:34 2018 CDT
Device: /dev/da2 [SAT], will do test 4 of type L at Sun Jul 29 01:19:34 2018 CDT
Device: /dev/da3 [SAT], will do test 4 of type L at Sun Jul 29 01:19:34 2018 CDT
Device: /dev/da4 [SAT], will do test 4 of type L at Sun Jul 29 01:19:34 2018 CDT
Device: /dev/da5 [SAT], will do test 4 of type L at Sun Jul 29 01:19:34 2018 CDT
Device: /dev/da0 [SAT], will do test 5 of type L at Sun Aug  5 01:19:34 2018 CDT
Device: /dev/da1 [SAT], will do test 5 of type L at Sun Aug  5 01:19:34 2018 CDT
Device: /dev/da2 [SAT], will do test 5 of type L at Sun Aug  5 01:19:34 2018 CDT
Device: /dev/da3 [SAT], will do test 5 of type L at Sun Aug  5 01:19:34 2018 CDT
Device: /dev/da4 [SAT], will do test 5 of type L at Sun Aug  5 01:19:34 2018 CDT
Device: /dev/da5 [SAT], will do test 5 of type L at Sun Aug  5 01:19:34 2018 CDT

P.S.
My scrubs are every 19 days on the 7th and the 28th of each month. ;-)

 

[rpsip]

Looks like the behavior for SMART testing is AND, not OR.

 

[Mike05]

I have recently revisited my SMART schedules and found this thread to be useful. Like the OP, I was also wondering if the tests (short and/or long) should be staggered for each disk, or can they be tested simultaneously? My setup is 3 x 4TB in RAID 1 and 2 x 4TB in RAID 1. If staggering is recommended, is it more important for some RAID configurations over others?

 

[Ericloewe]

I was also wondering if the tests (short and/or long) should be staggered for each disk,

No, that would be extremely stupid. There is absolutely nothing to gain unless your server can't handle the additional power and heat, at which point you have bigger problems.

 

[CSP-on-FN]

No, that would be extremely stupid.

Off Topic ... With due respect to you Eric, that was a rather emphatic (and surely discouraging ?) dismissal of what I thought was a reasonable query from @Mike05.

Back on topic ... On whether or not to stagger long tests and short tests on specific drives:

@Ericloewe - (A sincere question) Is there's something in particular about staggering test durations which makes it a 'stupid' thing to do?

From my point of view - as someone who prefers not to run my system at full-pelt just because I know it can take it - my reply to Mike05 would be "You can stagger the tests if you wish, but there's no advantage in doing so."

But - I'm genuinely keen to know if there's any actual disadvantage in staggering them.

Colin P.

 

[Ericloewe]

Is there's something in particular about staggering test durations which makes it a 'stupid' thing to do?

Yeah, instead of reducing performance for a period x, you reduce performance for a period of n*x, for n disks. This assumes that all disks are about even both normally and during SMART tests and that they're evenly loaded in RAIDZ. For mirrors, the same is true, but the performance impact is smaller, since only one vdev at a time would be affected, so it may help in weird scenarios.

There's also the added difficulty in managing the whole thing. Instead of one task, you'd have n tasks, each with different settings! That is crazy to manage.

Off Topic ... With due respect to you Eric, that was a rather emphatic (and surely discouraging ?) dismissal of what I thought was a reasonable query from @Mike05.

Possibly... It does sound a bit aggressive. That's what I get for writing before properly waking up.

 

[CSP-on-FN]

Yeah, instead of reducing performance for a period x, you reduce performance for a period of n*x, for n disks.
(....)
There's also the added difficulty in managing the whole thing. Instead of one task, you'd have n tasks, each with different settings!

Ah! Gottit.
I see the logic now, thanks.
Colin P.

 

[Mike05]

Thanks Ericloewe for your response, and CSP-on-FN for backing me up. No offense taken.

Appreciate the clarification also. In my case, I decided to schedule two SMART tests, one for each volume.