FreeNAS 11.1-BETA1 Now Available

[Mirakal]

Tomorrow morning.

Heart beating rapidly..:D

 

[diskdiddler]

Thanks.


You don't happen to know, if there's any negative impact, for those of us who had to manually add the tunables to our system to fix the networking do you? Should we remove those strings before updating? (since we shouldn't have had to add them in the first place, to my knowledge)

 

[Jailer]

I imagine we're going to see BETA2 any minute now.

It's available.

https://download.freenas.org/11.2/STABLE/BETA2/x64/

 

[diskdiddler]

I forget, can we reboot during a scrub ? Its the first if the month now, so my system is vbusy for 36 hours

 

[anmnz]

I forget, can we reboot during a scrub ?

Of course you can. The scrub will resume, pretty much where it left off, after the reboot.

 

[diskdiddler]

Actually someone yelled at me that a scrub a month is bad on my disks (I'm not sure if I agree but so be it, they've lasted a while)
So I changed to 1,3,5,7,9,11 - so it wasn't scrubbing after all.

Phew.

 

[Jailer]

Actually someone yelled at me that a scrub a month is bad on my disks

Hogwash. I've run 2 scrubs a month for nearly 4 years and my server is still running strong.

 

[diskdiddler]

I did one a month for nearly 3 years, I guess the only reason I mainly succumbed to their suggestion was I lost usable disk speeds for 36 hours a month, now it'll only be 36 hours every 2 months.
I still run a full smart test, per disk, once a month.

 

[jgreco]

I still run a full smart test, per disk, once a month.

I can't tell if you think that's a good idea or a bad idea.

I run a short every four hours and a long twice a week. The SMART tests may actually be better to run as the HDD manages these and is supposed to keep them from interfering with the disk I/O demands of the system, and does the entire disk linearly. Scrubs do not analyze the unused portions of the pool. I would rather know sooner that a problem is developing.

Hogwash. I've run 2 scrubs a month for nearly 4 years and my server is still running strong.

Scrubs may be a bit tougher on the drives because there tends to be more seeking. Hard drives should be able to do massive amounts of read activity without incurring a significant shortening of life expectancy. Some people have pointed to the recent trend to define data center HDD's in terms of workload TB written per day. This is mostly a side effect of SMR impacts and is for writes, not reads.

 

[diskdiddler]

I can't tell if you think that's a good idea or a bad idea.

I run a short every four hours and a long twice a week. The SMART tests may actually be better to run as the HDD manages these and is supposed to keep them from interfering with the disk I/O demands of the system, and does the entire disk linearly. Scrubs do not analyze the unused portions of the pool. I would rather know sooner that a problem is developing.

The full smart test is still going to strain a disk, when I'm asking to read or write and a disk has been in the middle of reading all 5TB for 5 hours, the heads need not continue to thrash around performing the test. Ideally the disk would go offline entirely.

I'm happy with monthly SMART tests, it's VASTLY MORE than I ever gave disks before I had a NAS (no smart checks ever, then wonder why a disk fails 'suddenly' losing data)
So yes a good idea, once a month seems reasonable enough.

Scrubs may be a bit tougher on the drives because there tends to be more seeking. Hard drives should be able to do massive amounts of read activity without incurring a significant shortening of life expectancy. Some people have pointed to the recent trend to define data center HDD's in terms of workload TB written per day. This is mostly a side effect of SMR impacts and is for writes, not reads.

Honestly I agree with you, seems reads, really shouldn't impact the disk too much, but at the end of the day, the actuator still needs to work, the head is performing a function, it's utilisation more than idle.
Either way, since moving to FreeNAS, I've managed to have a reliable store of data and confidence in my hard drives.

In the 'old days' before FreeNAS I would go and buy a nice big 2TB hard disk and fill it 65% of the way, not knowing that 80% onwards was utterly worthless, damaged sectors, debris, just poorly formed magnetic surface, who knows - the 'end' of the disk was faulty on more than one occassion for me, but I NEVER encountered it until the last minute.
Now, I'm 99.9% confident my disks are reliable and I'm told of a problem with the disk itself, BEFORE I've ever tried to write to that sector (!!)
Love it, very happy with FreeNAS as a storage OS

 

[jgreco]

The full smart test is still going to strain a disk, when I'm asking to read or write and a disk has been in the middle of reading all 5TB for 5 hours, the heads need not continue to thrash around performing the test. Ideally the disk would go offline entirely.

I guess I don't care if the mechanism is "thrash[ing] around performing the test" if it isn't being asked to do anything by the host platform. It gives an opportunity to remediate the issue as it allows me to try to force a sectror reallocation, and if that fails, then I can pull the drive. I have a small pile of unreliable disks I've removed from service this way.

But I've been accused of being paranoid before.