I have quite a few drives, and I’ve RMA’d one of them when its SMART log was reporting “less than 24 hours to failure, backup all data” that was a no brainer. But I have a few others that generate read errors in the dmesg and some times those cause mergerfs to lockup, and I have to cold boot the machine, which I don’t like doing. I’m tracking which drives cause issues, but the SMART log isn’t so dire, double digits of recorded error events.
So when is it time to pull the drive and ask for a replacement?
It’s been many years since I had to RMA a drive, I remember them wanting me to try and run some sort of software diag from them to try and talk with the disk and get an error report. Don’t really know what the process is in today’s time. If they will send you a replacement on a CCard and not charge you when they get the old one back.
I just had an 18tb drive that was acting up, with long response times (Seagate). It had a yellow under “Current Pending Sector Count” in Crystal Disk Info. I downloaded Seatools, pulled it from the node and put it on another PC, did the “full scan and fix” option, It took more than a day to finish (super long, but whew), but the drive seems happier, the yellow is gone. Time will tell if issues return.
I was hating the idea of RMAing after my WD experience.
Having one such pending sector is not really a problem. In the best case, few hashes will be destroyed in one plot (out of billions), in the worst that one plot needs to be replotted (after fixing it the way you did), and drive is as good as before.
However, it is good to monitor such drive for some time to make sure that those pending sectors will not be coming more often. If they do, there is no point fighting them, as at some point the spare sector table will be exhausted.
It was 183 if I remember correctly, and everything I read said it was RMA-able at that point …probably. But a drive in the node is worth holding on to if at all possible, so the attempt at a (not so quick fix) was worth it, hopefully.
Even at that it took a couple dasy to offload all the plots to various available ssds and other small drives I had laying around to save the plots on it that I had done, before trying the ‘fix’.
That is not good. Based on what @seymour.krelborn posted before, when he RMAed his Seagate drive, they did a backup of all his plots and sent those back to him, although, he went through some hoops with that process. So, most likely right now you will get back most of your plots recovered, otherwise you will end up with a dead drive that will need to be replotted.
Reallocation sector count is your typical gage of HDD health and I’ve had drives with low single digit number work for a long time without a failure and others which showed few sectors and these multiply few thousand times within 24h before the drive was essentially unreadable. I would not trust any HDD with any number above 0 in this SMART attribute. I actually had one manufacture argue with me that this value is not necessarily an indicator of failure and can be fixed with their special scan and “fix” tool but I think what that does is just remaps the bad sector with a sector from reserved space while hiding the possible growing problem until your warranty runs out. I recommend running SMART extended test once a month to keep tabs on HDD health and possibly more often on suspect drives.
I have RMA’d maybe a dozen Seagate EXOS Drives. I’ve only had 1 or 2 that completely failed and I wasn’t able to pull the plots off. Otherwise, once they start acting up (either I hear them knocking) or they periodically lose connection, I’ll proactively RMA them but transfer the plots to another drive first if possible.
The process with Seagate is pretty straight forward; you do it online and send it in (you pay for shipping). The turn around time has been pretty good. I get a notification when they receive it, then I get a notification when they send the replacement. I’d say on average they ship out a replacement within about 3-4 days after receiving it.
I’ve had one replacement drive that was DOA. That was a bit of a process because I felt I shouldn’t have to pay for shipping to return it. I was finally able to get Segate to send me a drive first and include a shipping label. But that process was a little painful dealing with support over email. Never quite sure if you’re dealing with Bots or real people in support emails - it’s like they don’t read the details.
I had only 1 RMA with Seagate. It was for an external 14 TB USB drive.
They paid for shipping in both directions.
They also gave me free data recovery, where the recovered data (the recovered plots) were placed on two more drives (presumably, they did not have a 14 TB drive on their end).
So all shipping was free, the data recovery was free, and the two extra drives were free.
Aside from a shipping mistake on their end (which made things a headache), their RMA process exceeded my expectations.
I had no idea that my warranty included free data recovery. They brought that to my attention when I spoke to them.
Do all of their drives include free data recovery in the warranty?
Note that I did not use the automated process. I spoke to support personnel. That should not change the outcome. But maybe it is why I got the free shipping in both directions, and maybe that is the only way to check for and request free data recovery and get a free replacement drive?
Note that I also had two warranty claims with Western Digital. They made the process as painful and time consuming imaginable. They did not charge anything for shipping in either direction. But it was still a nightmare dealing with them.