I have a hard drive where I have a storj node. The disk has the warning marked. The disk has eight reallocated sectors. Last month it had no reallocated sectors. The node is fine.
Do you recommend me to leave the node without changing the hard drive or change the hard drive for another used one without reallocated sectors?
Hello @Vicente,
Welcome back!
Many Operators suggests to run until it die, however, it’s up on you.
You may also run a long S.M.A.R.T. test, some suggests that it can mark bad blocks beforehand:
Run-to-failure - It may keep going for years with an occasional fsck.
Or it may die tomorrow.
When it starts causing you problems too often: see if you can recover it to a new drive. Maybe you can continue, and maybe you have to make a new identity and start again. Storj circle of life 
Net growth is almost not existing for storj nodes these days and old nodes have accumulated a lot of precious long term data. Therefore I prefer to move to a new disk as soon as possible. If bad disks are out of warranty I am using them for chia until final death.
If the hard drive is still under warranty, replace it. The manufacturer knows that reallocated sectors indicates a failing disk and will replace under warranty no question.
Nope. It does not. There are trillions of sectors on the disk. It’s OK for some of them to go bad. Few bad sectors does not mean failing disk.
This depends on the vendor. But why would you bother for 5 bad sectors? Sending drive in, receiving replacement, etc, recycling boxes. Replace disk once it no longer reads/writes. When filesystem offlines the disk, because it ran out of spare sectors and continues to deteriorate. Not eariler. Doing it earlier is a waste of resources: unnecessary errands and carbon emissions recycling perfectly usable disk.
But what if it fails after my warranty expires? I can replace it now for free!
To that I say – Ignore warranty entirely. It’s not worth your time. Warranty is worth couple of bucks tops. It’s worthless. Buying used disks you save way more than warranty is worth anyway.
TLDR: disk is online and writes/reads data – keep using it. Disk is offline – replace it. Ignore SMART data altogether.
In theory storj could work with unhealthy disks but storagenode software is not designed to deal with read errors, it just exits on any read error. So too much work to keep such nodes running in my opinion.
You don’t get that nice “fresh-out-of-the-box-brand-new” smell from a sh drive.
If you want to miss that, go ahead, by used ones. 
Depends. I bought refurbished components from China, they used an odorant 
to make them smell as new, so do not think about this at all. Maybe it’s just a silica gel was with an odorant, but well.
The hard drive is working fine.
I think the most sensible thing to do is to leave the hard drive running. Check the hard drive in the next few weeks to see if more reallocated sectors appear.
Doing a long smart test does not seem like a good idea to me. It can take a long time. The node would be offline for a long time.
Would it be okay to migrate the node to another hard drive and do the long smart test? This would also waste a lot of time.
How big is your drive? For a nonactive drive, it took 30h for 24TB. Stop the node and do the test. Won’t take long. You won’t get DQed.
I use my disks for other data, not just StorJ. For me it makes sense to get a replacement to keep my “real” data safe. If it was just for a StorJ node, then I would probably leave it be longer.
I too share pool with storj that holds my personal important data (including documents, photos, and various backups).
However, I don’t trust any disks to keep my data safe, regardless of number of bad sectors. It’s not their job. Keeping my data safe is a job of a redundant filesystem. Not individual disks.
Right now I have two disks with 7-10 unrecoverable sectors. I sleep just fine at night.
The hard drive is 8tb
So it will take you like 10 hours for SMART long test, if you stop the storagenode and any other program that might use that drive. You should be fine.
Scanned the hard drive, no errors found.
Still 8 reallocated sectors.
