So after my last post, I discovered that one of the two new 3Tb Western Digital drives is throwing SMART errors, as is one of the older 2Tb Seagate drives. Well, the new one is brand new, just a few weeks old, and the older 2Tb one is just under a year old, so it’s still under warranty, so time to test the two RMA processes side by side.
I put in both RMAs on the same day. I had some problems with the Seagate web site, but I didn’t make careful note of the details and I forget exactly what the problem was. In both cases, I opted for the “advanced replacement” service where they send you the replacement drive first, and then you use that box to send back the defective one. I don’t recall either of them offering a more expedited version of the service.
The WDC drive took a few days to arrive. When the WDC one arrived, there was an option on their web site to click a single link and buy a UPS shipping label with the return address and RMA number and stuff all pre-printed. Very nice. When I went to the WDC support site dashboard, it already showed the new drive’s serial number as registered to me, and it had removed the defective one from my list of registered drives. The only problem: the dashboard showed the warranty period on the replacement one as expiring in 5 months. That’s a bit odd. I put in a ticket to ask about it and they said that when the defective one arrives back, they’ll update the warranty period back out to three years. We’ll see.
The Seagate one took over a week to arrive. The replacement has a big “REFURBISHED” label on it. I guess it’s unreasonable to expect a new drive to replace a year old drive, but one can live in hope, right? They sent me an email with instructions for returning it, including helpfully putting the return address and order number on page 4 of a 7 page email and suggesting I print it out and use that as a “mailing label”. That email also told me that I’d opted for “Ground Advanced Replacement” and if I’d opted for “Advanced Replacement” instead I would have gotten 2 day shipping on my replacement and a pre-paid shipping label for the return, all for $9.95. I don’t recall ever being offered this, or if I was, i wasn’t told how it differed from the free service. Still, the order confirmation is probably the wrong time to tell you what you should have ordered instead. Anyway, I guess I’ll be trudging off to the UPS store to get this shipped tomorrow.
Ok, now I’ve told you why Western Digital rules and Seagate drools, I’ll tell you about my replacment experiences.
When the first drive arrived, I shut down my computer, yanked the bad drive, put in the new drive, and rebooted. I got a message asking me if I wanted to start the RAID in degraded mode, and I did. Everything started up perfectly. I did the parted and mdadm magic to make the partitions on the new /dev/sdb and get it into the RAID, and after everything rebuilt it was right as rain. The number of odd DMA errors appearing in /var/log/kern.log went down to zero.
When the second drive arrived, I attempted the exact same thing. I shut down, yanked the bad one, put in the new one, and powered it up, and it refused to boot. Uh oh. Carefully checked the serial number on the drive to make sure it was the defective one. Checked in the BIOS to see if it seeing all the drives. But when I booted, I never saw the message asking me if I wanted to boot with the degraded RAID, it just hung. Put the defective drive on a spare SATA controller and booted, and it booted fine. Hmmmm. Used the appropriate mdadm commands to fail and remove the defective drive, and add the new one to the RAID. Tried grub-install, and it gave a non-fatal error about a device named “null”, but when I attempted to boot without the defective drive, got a grub error about being unable to find bios-i386-pc or something like that. Tried booting from all 4 disks, and got the same error. So I booted with the defective drive still installed, and waited 24+ hours for the RAID to finish rebuilding. Once it finished, I was able to do a grub-install and it didn’t give that strange error, and afterwards I was able to boot with the defective drive safely back in its shipping box. Phew.
Did you see this? All of my drives are Seagates. Blah. Gonna have to fix that…
http://techreport.com/news/25940/hard-drive-reliability-study-names-names
Well, I guess the nickname “Deathstar” for IBM Deskstar drives isn’t deserved any more.