Another Seagate HDD Bites It
Poor quality of Seagate disks is a well-know issue for people using spinning disk storage at home. I hear their enterprise HDDs aren’t bad, but that isn’t what we purchase.
My sample size is very small. From 1990 – 2005 I went out of my way to purchase Seagate HDDs. They lasted for the sizes I bought. Used some 320G Seagate disks in an array for 7+ yrs and NONE of those failed. They made quality HDDs.
Until around 2007 and the 750GB and larger sized disks came out. Since that time, Seagate non-enterprise disks have had a quality issue. Every quarter, Backblaze publishes their failure information which clearly shows that desktop seagate HDDs fail at a much, much, much higher rate than those from WD or Hitachi or Toshiba or … pretty much any other vendor. The numbers speak for themselves and are sufficient to recommend avoiding Seagate.
So what does this have to do with anything? A few years ago, I needed storage. It was during a period after major flooding in Asia had effectively stopped the manufacturing and HDD prices were very high. I’ve been buying 2 HDDs at a time since around 2002 (when I got backup religion ) after loosing about 80% of our data. Bought 2 Seagate 2TB HDDs in a weak moment, before doing research on the specific model. 2 hrs after the purchase, I had buyers remorse. One of those failed 1 year and 5 days after being installed. The other was running until this morning – 2.5 yrs of use according to SMART data. The warranty for both was 1 year. Way to stand behind a product. The first failed disk was replaced by a Toshiba, which is still working.
That was the last of the big Seagate HDDs running here. My data is worth the slight extra cost of a different HDD vendor. All the other vendors have exceeded my expectations when it comes to HDD lifetime. It is just Seagate which has left me disappointed and only since 2007-ish.
Usually, I’d take this as a forced opportunity to increase the size of the storage, but quality 4TB HDDs are a little pricey today. Usually wait until some HGST drives are on sale. Can’t wait this time and microcenter does have the Toshiba I need in stock.
Goodbye Seagate. Good riddance.
Replacement – I love Linux!
The replacement disk arrived and this morning (Saturday) the old disk was replaced. It was extremely easy. Easier than setting up a new array, because mdadm rocks.
- Determine the failed disk. My array has 4 drives – 2 WD Red, 1 Seagate and 1 Toshiba (in that order from top to bottom). Checking dmesg and parted and gparted made it clear the issue was with the Seagate. The other Toshiba was a replacement for another dead Seagate a few years ago. Bleh.
- Pull the failed disk and replace it with a new disk – of the same size or larger. I had to shutdown the system for this. Not a hot-swap array for $120 all those years ago. ;) This was the hardest part of the disk replacement effort. While I had it open, vacuumed the dust from inside the array.
- Clone the source disk partition table to the new disk (target). fdisk, sfdisk aren’t safe with GPT disks, so use sgdisk. Best to have this program installed before it is needed. Just install it now, if you don’t have it.
sgdisk -R /dev/sdc /dev/sdd
sdd is the source. sdc is the target. The devices changed names after a reboot, so beware. Also note that sgdisk options aren’t in the order you expect. Double check these BEFORE you press enter. - The GUID was cloned above, which isn’t good for a RAID disk. Force a new GUID to be generated for the disk:
sgdisk -G /dev/sdc - Verify that the new disk has all the partition(s) you needed.
sudo parted -lYes? All ok? - Add the partition on the new disk to the SW-Array.
mdadm --manage /dev/md1 --add /dev/sdc3 - Watch the sync progress
$ more /proc/mdstat Personalities : [raid1] [linear] [multipath] [raid0] [raid6] [raid5] [raid4] [ra id10] md1 : active raid1 sdc3[3] sdd3[2] 1943010816 blocks super 1.2 [2/1] [_U] [===>.................] recovery = 16.9% (328619776/1943010816) finish=234.5min speed=114719K/sec
Note that the boot drive is NOT part of this array. If it were, I’d need to install grub after all the partition cloning were done, but before the partition was added to the array. grub-install is the command. Don’t know if update-grub is needed or not.
I wrote this more to remind myself how easy it was later than to help others. Used Tim’s tutorial to get started. I didn’t fail or remove the bad disk before replacement, as shown on Falco’s website.
Many people don’t partition their RAID drives. I do, so that other systems and people have a chance to see there is something on the disk – it isn’t empty. Also, I partition a few wasted partitions “just in case” I need them in the future. a) UEFI partition; b) 10G partition for unknown future uses. Sometimes having just a little more space is handy. That’s all. Plus 10G is large enough for a Linux install, should that be needed.
