100 Days

[adrianwi]

There's a lot to be said for sticking around on an older version of a while ;) Quite happy I rolled back from the replication and LSI firmware issues in the early 9.3.1 release as I've had a problem free 100 days uptime on 9.3.

I know these are probably fixed, but I've got ownCloud, Plex, openVPN, VirtualBox (Win 10 VM) and Calibre all running just perfectly, so I'm sticking with the 'if it ain't broken' mantra for a little longer :D

 

[Mlovelace]

There's a lot to be said for sticking around on an older version of a while ;) Quite happy I rolled back from the replication and LSI firmware issues in the early 9.3.1 release as I've had a problem free 100 days uptime on 9.3.

I know these are probably fixed, but I've got ownCloud, Plex, openVPN, VirtualBox (Win 10 VM) and Calibre all running just perfectly, so I'm sticking with the 'if it ain't broken' mantra for a little longer :D

Looks like your /dev/da3 could use some help ;)

 

[adrianwi]

Haha. It's been doing that sporadically since before my last FreeNAS upgrade. I've got SMART reports scheduled periodically and a spare drive lying around for when it increases beyond 8 :D

 

[m0nkey_]

I'm not one for caring about up-time. I would also replace that drive before it's too late.

 

[DrKK]

Most/many of us would replace that drive at the FIRST sign of an offline sector.

M0nkey and I think it's a Seagate. :)

 

[random003]

I got you beat. :p
http://postimg.org/image/mx8f8arun/


Multiple HBA cards. Updating firmware is a pita so I left it as is.



I also disagree about the need to replace that drive. If you have redundancy and backups I don't see the need as long as it performs well.

 

[Mirfster]

Multiple HBA cards. Updating firmware is a pita so I left it as is.

How so? "sas2flash" is in FreeNas and takes under a minute....

I also disagree about the need to replace that drive. If you have redundancy and backups I don't see the need as long as it performs well.

Like living on the edge eh? :D

 

[depasseg]

Side note: Today was the kids "100th day of school" celebration. I thought you were talking about that.

 

[Starpulkka]

I have some stickers on each freenas machines, and as an example i can easily see that one hdd has 75 reallocated sectors, so i keep "track" that hdd. . But it's quite nice to have redundancy and backups, so you dont get that instant awful feeling when you are about to lose data. I know that if you try to refurbish that hdd by using spinrite or badblocks on another machine, pool might not accept it back if it has not enought sectors. And i still use half of my machines FN8, so i kindof understand your current status..

 

[DrKK]

"Let me drive extra fast, because I have my seatbelt on".

 

[adrianwi]

Duplicate

 

[adrianwi]

I did think about switching out straight away but thought I'd see if it got any worse. It hasn't and at times doesn't even report it as a problem. Every now and again it does, but only the same 8 sectors it first reported about 7 months ago.

It's one of 7 in RAIDZ2 and the whole pool is backed up to another machine so I'm reasonable comfortable with things. As soon as I see more than 8 I'll be changing it.

Any yes, it's a Seagate with just under 2 years left on the warranty. Others are Seagate along with the 5 in the backup machine, and no issues with them, yet

 

[toadman]

Sweet. If it's got warranty left could you not just replace it and RMA for a new one?

 

[Bidule0hm]

I know that if you try to refurbish that hdd by using spinrite or badblocks on another machine, pool might not accept it back if it has not enought sectors.

Actually bad sectors are remapped and the limit is usually 65535 sectors so unless you go over that limit (then I don't even know what happens, but at this stage the drive is trashed beyond any usability anyway...) you don't lose any sectors.

And in the hypothetical case where you lose sectors there's the swap space that can be reduced a bit so the smaller drive can be used ;)

 

[Starpulkka]

Actually bad sectors are remapped and the limit is usually 65535 sectors so unless you go over that limit (then I don't even know what happens, but at this stage the drive is trashed beyond any usability anyway...) you don't lose any sectors.

And in the hypothetical case where you lose sectors there's the swap space that can be reduced a bit so the smaller drive can be used ;)

There was a history on wd greens that you replaced bad one on brandnew "same" size one pool did not accepted it back as it was like 1 sector smaller than others in pool.
Havent tested that if you reduce swap size from, does hdd get accepted back to pool (but i dont understand how smaller swap size affects avaiable sectors on entire hdd isn't whole hdd sector amount be still same??), plus if others hdd have bigger swap size. But i put it on my todo list. Currently having fun testing speeds on bigger block sizes versus lz4 compression on. So far looks that 128 KB blocksize is and old relic. And hdd bad speed setting's is happening again when there was clone pc era.

If you wanna be smart you can change sector amount ahead (before makin pool) on a hdd make it appear a smaller, and if hdds lose more avaiable sectors you can edit those hidden sectors back to avaiable and again pool accepts it gladly. This was used manytimes on old times as fooling data forensics as they just press enter and found nothing on avaiable sectors. ;)
Edits:some bad spelling and bad translations finnish to english

 

[adrianwi]

My 100 days only made it +8 and then da6 decided to start throwing uncorrectable sector errors. Started with 8, like da3, but after about 24 hours jumped to 16 so that was pulled and a new drive resilvered. Ready to RMA back to Seagate, and when the replacement arrives I'll swap out with da3 and RMA that one too.

 

[jgreco]

Sweet. If it's got warranty left could you not just replace it and RMA for a new one?

You probably don't want to do that for a drive that's merely reporting a few sectors bad. That can be a stable situation and is certainly not a sign of impending doom - just a flag to keep an eye on it. An increasing count is a sign of problems. The drive a manufacturer sends you in exchange for an RMA, however, is NOT likely to be a new drive.... rather, it is very likely to have failed catastrophically and had to be sent back by a previous owner. It could have been in an external enclosure and gotten baked or dropped. It might have gone through a hundred thousand load cycles. Who knows what happened to it. It is likely to have had a lifetime experience that is substantially worse than merely developing a few bad sectors. You are probably much better off with a stable disk that has a few bad sectors that'll remap the next time they're written.

Some of us won't even bother putting an RMA'd disk back into production.

 

[Bhoot]

You probably don't want to do that for a drive that's merely reporting a few sectors bad

I hate it when the dot in the right corner is red with "1 sector currently pending (unreadable)" But yes I wait for a few sectors to get unreadable. My system had 16 stable sectors unredable whereas another disk started from 5 and went to 600+ in a a matter of 2 days. A cold spare was used and the disk was rma'ed. The disk with 1 sector unreadable now is upto 16. I am thinking of replacing it with the disk received after rma.