Just want to try something different, and the WD80EFZX is the only other low power 8TB drive available. Locally, it costs about CAD$60 more than the Seagate.

I paid the USD$30 ransom for a hassle-free RMA with 2-way shipping included. The replacement Seagate 8TB drive arrived today via UPS, and has been inserted into the system's backup array (currently re-syncing 4TB of data to it).

All good for now.

This escapade forced me to finally deal with the 99% full backup array, by replacing one of its 4TB drives with the RMA replacement 8TB drive. Now the backup array is the same size (20TB) as the main array. Yay!

That turns out to have been US$30 wasted. The replacement drive has already developed numerous bad sectors -- I guess they must've just reformatted somebody else's failed drive and shipped it to me.

So.. off to the local shop again later today for another WD Red drive.

Cheers

This is very useful info Mark! It seems Seagate's bad rep is still not totally unwarranted. Just when I was in doubt of giving them another chance because I'm looking for an 8 GB drive specifically for backup reasons... (mainly because of the price) I guess I'll just bite the bullet and spend the extra cash to buy me a WD 8 GB drive.

I do admit I have fewer complaints with WD. Not zero, but a lot fewer. The only reason for the Seagate drives here (4TB and 8TB) was that they became available much earlier and are much less pricey than WD's similar models.

Now that WD has 8TB, they do appear to be the better choice.

A non zero and increasing Reallocated Sector Ct and Seek Error Rate on a brand new drive is a bad sign, right ?

Just picked up 3 4TB WD drives to upgrade my ZFS mirror. I'm swapping the first drive in and it is showing errors

Those couldn't be caused by a dodgy connection I assume ?

They are all WD40EZRZ (I'm now regretting not picking up three different model drives like I did last time).

Reallocated sector count should be zero. Otherwise replace the drive.
The Seek Error Rate is a normal non-zero thing. Ignore it.

Bad cabling will produce UDMA CRC errors rather than the others.

Cheers

..but noticeably noisier than the Seagate drives they replaced.

The 8TB Seagates were silent. Nothing there, any noise from them was masked by the very gentle airflow of the case fans in the HTPC box.

The WD Red 8TB have a noticeable low frequency "thumping noise" that sounds kinda like the kid across the street (300' away) bouncing his basketball.

Very subtle, and distracting. And those two drives are already on silicone isolation mounts.

Someday..

What would a sketchy power feed cause?

Your guess is as good as mine there. Clean power is a necessity for digital, and without it all kinds of weird stuff can happen.

Cheers

Thanks

I plugged the second drive into the same SATA cable/power and successfully wrote 2TB to it, so I guess I'll be returning the first drive. I still need to test the third one.

At this point in the process I'm wishing I'd used FreeNAS for my ZFS array rather than Ubuntu. The ten step process to replace each drive is a little tedious and error prone (I've already accidentally over written a partition table on one disk).

It would be a great deal easier if I wasn't being a cheapskate and not using the same 3 drives I'm mirroring under ZFS to also boot Ubuntu using a standard Linux mirrored array.

- add new drive
- copy partition table from existing drive
- randomise guid partition labels on new drive
- resize the ZFS partition on the new drive from ~2TB to ~4TB
- dd the boot partition over from old drive
- fail and remove the old drive from the Linux RAID mirror
- add the new drive to the Linux RAID mirror
- replace the old drive with the new drive in the ZFS mirror
- run grub install
- wait 10+ hours for the Linux mirror and ZFS to rebuild

Repeat three times and hope at the end of it all that ZFS correctly expands into the new space...

Yeah. Sharing the boot drive into a RAID leads to all sorts of complications. Usually a USB stick or small SSD is used for that.

Cheers

When I setup this server doing ZFS on the Linux root disk wasn't really practical (it looks like it is still a pain in the arse, though more workable). So if I wanted redundancy on my root partition, which I did of course, then using standard Linux RAID for the root partition was my only option.

I probably should have used FreeNAS, which would have made full ZFS easy. However when I was building the server FreeNAS wasn't as well suited to general server stuff as it is now (I run various VMs etc on the box).

You can all laugh when I blow away the ZFS pool accidentally...

I am looking forward to testing how easy it is to get one of my removed ZFS drives to boot on other hardware. That is after all the main reason I mirror rather than use higher RAID levels.

And.. the other Seagate 8TB drive here has also now failed.
Off to the shops again when they re-open Tuesday for yet another WD 8TB Red. Hopefully these ones will last longer than just a few months.

... this is worrying. My server main storage system still has an 8GB Seagate in it (the other one failed as you may remember from several posts above). I am hoping these failures are caused by high temp, because that is what I do NOT have any longer after adding fan and making a hole in my home server case. Otherwise, I'll be replacing the Seagate soon I am afraid.

It looks OK to me. What do you guys think?
Runnig at 33°C .

http://www.ilquen.it/download/miscellanea/Seagate8GBSMART.jpg

Perfectly healthy.

My drives run closer to 42C, still plenty cool enough for temperature to be a non-issue.

Cheers

It turns out I have been a bit hasty in condemning the Seagate 8TB drives.

Yes, the first one that died really was a bad drive.
But the replacement for it, as well as the other (original) Seagate 8TB drive, are both still fine.

They do both report the same strange S.M.A.R.T. values -- a sign of something strange, but given how little vendors document their stuff I'm not 100% sure those values really are "bad".

But the kicker for me was both the replacement, and other, drive misbehave like crazy in the backup array here, which has a JMicron JMB363 SATA port multiplier. That setup works flawlessly with all other drives I've used in it: WD Green 2,2.5,3.0,4.0TB drives, various Hitachi and Maxtor models, smaller capacity Seagate drives, and Seagate 4TB drives.

But put the Seagate 8TB drives in there, and the result is a series of semi-random SATA errors with no meaningful diagnostics. So today, I flipped the switch on the array to change the interface from eSATA to USB3. All is now fine again. Slightly slower as one would expect, but no more mysterious failures.

And given how funky JMicron stuff has been in the past, and their extreme reluctance to provide programming documentation for their chips, I'll have to score this one as a JMicron rather than Seagate issue.

Cheers
-ml

The last of the 8TB Seagate drives here has failed, and now all are out of service, with various models of WD 8TB drives having replaced them one by one.

Friends don't let friends buy Seagate drives.

I read that somewhere; now I just have to remember to apply it!

Any issues with the WDC drives other than the noise?

The WDC drives I buy are all 5400rpm or so, and not noisy. Every one of the four WDC 8TB drives is working perfectly thus far, as are my much much older 2TB, 3TB, and 4TB WDC Green drives that I use for various oddball purposes.

There are still three working Seagate 4TB drives in a backup array here that have vastly outlived their 8TB cousins.

Spinning rust is only used for non-critical stuff here: media files and secondary backups. All of our boot/system/working drives are SSDs, and have been for much of the past 10 years.

EDIT: Heck, I recently picked up a small stack of 120GB SSDs at CAD$30 each, to use (with an adapter cable) in place of similarly priced (and MUCH slower) USB sticks!

Woah, where did you get those SSDs?

I should knock on all the wood around, but so far the WD drives in my Synology have been running for about 5 years, and for 4 of those it was running Surveillance Station with two cameras.

Local computer shop had a bunch of HP M700 SSDs, 120GB, at CAD$30/each. I bought five of them. They appear to be exactly the same as the HP S700 SSDs, but probably named "M700" by mistake before somebody realized HP already had a different product called M700. That's my theory.

Kingston also markets $30 120GB SSDs (Amazon, and my local shop), but they are a little bit slower.

These are all part of the new wave of "DRAM-less" SSDs --> omitting the DRAM buffer saves on price, but costs somewhat in endurance. The ones I got claim 70TB endurance.

They're just so useful in combination with a CAD$15 UASP capable USB3-SATA adapter cable. I put the Win10 from my new XPS 13 onto one of them, and boot it from USB to check for dock/device firmware updates.

Hmmm. My Synology DS416play is currently populated with 2x WDC WD40EFRX-68WT0N0 (4TB) and 2x Seagate ST8000VN0022-2EL112 (8TB), configured as a single Synology Hybrid RAID (SHR) pool. I also have a single Seagate 8TB drive (same model) as a cold spare. Should I be concerned?

Aside: I've only used 2.2TB of the 14TB available.

I’ve now had three of my WDC 4TB drives fail. Thankfully they are in a ZFS 3 way mirror and they have failed one at a time.

I’ve ended up replacing them with exactly the same model each time, purely because I’ve been stung in the past buying another xTB drive just to discover it is a tiny bit smaller tan my other xTB drives and so can’t be added to the array !

All the failures have been basically the same, drives just don’t appear in the SATA bus anymore.

Now that's just plain weird to me! Mine all die from "media errors" (bad sectors). They're fine, fine, fine, then one bad sector, then lots more within days. That's the usual pattern for a particle landing on the platters and then getting pushed around by the heads.

My usage situation probably does differ a lot from many others here, so drives that fail for me quite regularly may be fine for others with different usages.

Here, nothing stays spun up for more than a few hours at best. Most of the time my mechanical drives are powered off. They only get powered on and spun up when needed, perhaps once every couple of days. I guess the Seagate 8TB drives don't like that very much, despite the load/unload cycle count not being terribly high when they do die. Perhaps the spin up/down is dislodging some dust/particles or something that then kills them.

The WDC drives here thus far don't seem to suffer from that -- I really only ever replace them when I need higher capacity. I think I had one 3TB WDC die on me, but the rest have been fine for quite some time now. Maybe this on/off reliability comes from WDC's long history with external USB drives?

My 24/7 drives are all SSDs by Crucial/Micron. Most of our other SSDs are also by Crucial.

I have recently run across a completely dead (not showing on SATA bus) Mushkin 480GB SSD (w/Sandforce controller), and a dead Kingston V200 128GB SSD. Noteworthy for me, because dead SSDs are a very rare event around here. Both of those were older than the failed Seagate drives.

The Sandforce-based Mushkin would be recoverable if I had a factory firmware image for it (I don't), as shorting the internal jumper does bring it back to life in "manufacturing mode".

The Kingston V200 failed in a particularly nice way. No idea what its problem was, but it simply set itself to read-only and just accepts but ignores any attempts to write or erase it. The data on it was fully recoverable by simply mounting/reading it.

My mechanical drives aren't quite as quiet as yours, but they are pretty quiet.

My current server has three drives in it, over its five year life it has had four drives fail. One Toshiba and the three WDCs.

Some of the intermittently fell off the bus, some permanently failed to appear. A couple of them seemed to work when I plugged them into my external SATA dock.

Maybe there is something odd with my server hardware ?

But this is why I don't mess about with non mirror arrays. With my three mirrored drives in my ZFS I just don't stress about drive failures causing any real hassle. And drives are cheap nowadays

My previous servers, one Dell desktop machine and one Dell PowerEdge server with a five drive SCSI RAID array had zero drive failures in over 10 years of life (right up until the RAID card itself failed). I think I got very lucky...

I think we have six SSDs in use in the house, one Intel, one Apple (think it is a Samsung) and the rest Samsung retail ones. So far, zero failures on the SSDs.

I know I'm no expert in this area but that seems high to me. Your array has only had three drives at a time and you've had about one failure a year? I have four WD drives in my Synology, and they were getting written to 24/7 with the Surveillance Station running for three years, and the NAS has been operational for 4-5 years. So far I've never lost a drive (but now I'm going to kick myself for saying that).

Naturally, YMMV. I would say it could be environmental factors, but my Synology now lives in our laundry room where it's regularly 80 degrees (F). It's been humming away like that for two years now.

...I'd better order a drive to have on hand just in case...

It is certainly a much higher failure rate than I expect, but it could just be bad luck.

The Toshiba drive was effectively dead on arrival, it died a couple of weeks in.

It isn’t environmental, the server has been in the roof space above our kitchen. The temperature in there rarely moves out of the 20-25 C range and it is never humid.

In comparison my old servers lived in the garage at our old house, with the temperature ranging between 0 and 40 C with humidity often over 90%. The steel on the cases was rusty after 10 years, but all 8 drives kept on chugging along

For what it's worth, I have a number of drives that have got tens of thousands of hours of uptime on them in two different fileservers and no errors logged at all. For example, in my media server there are two Samsung HD204UI 2TB drives, one WDC WD20EARX 2TB drive, and two WDC WD30EFRX 3TB drives. Respectively, the raw power on hours values are:

56399
56803
51238
10236
15107

Assuming that these values actually ARE genuinely an hour count, one of those drives has been running for six and a half years continuously.

The other server is has six 1 and 2 TB Samsung drives in it. Last time I checked, a couple of those were over 80000 hours. Since it's in the shed as a semi-offsite backup, and the thing is unheated, it's at times registered drive temperatures according to SMART of -5 degrees or so

The only drive failure I've ever had in that either one was a Seagate drive. In both cases they lasted about a week longer than a year, then died in a matter of hours with creeping bad sector errors until they were totally shot. I was able to recover the data in time, but I haven't bought a Seagate drive since!

I've never had a Samsung drive fail,except for one that was DOA due to incredibly bad packaging from the supplier, and I've got about a dozen of them. None of the Maxtor or WDC drives have failed in the last seven or eight years either.

So it's also possibly just luck, but it's also fairly consistent

On the other hand, I went to a lot of trouble to make sure the drives are kept cool. -5 may be pushing it...