tomkarpik.com

Man Up, Seagate (and WTF, Samsung?)

January 14th, 2009 :: 6 comments

I’ve probably owned every major brand of hard drive at one point or another, and I’ve only ever had one go bad on me in long-term use in the last 10-12 years — thus making me impartial when it comes to brands.

Sure, IBM had its Deathstars, Maxtor had those low-profile desktop drives (that my boss calls “Death Dealers”), Western Digital has had a problematic series IIRC, and so forth. All of these problems were mechanical, and did not affect all units of a particular model. There’s an IBM Deskstar 60GXP at my office that still works perfectly after all this.

And now we come to Seagate.

I’ve been exclusively buying Seagate desktop drives for the last few years because my local computer store has had good deals on them consistently, they’ve had an excellent reputation, good warranties, and a model nomenclature that is consistent and easy to read. Not all of them have been top performers in their generation, but they’ve mostly all been very quiet — which is more important to me the older I get. :-D

Then Seagate released the debacle that is the 7200.11 series.

A few weeks ago, I decided to finally do something with the 1U dual Opteron server that was sitting in my room for about two years. I bought a matched pair of Opteron 250’s, 4 GB of DDR ECC RAM (what a rip), and two 500 GB Seagate 7200.11’s to put in a mirrored setup.

I spent an hour or two a day for a couple of weeks setting things up, and when it was finally all ready, I drove it down to our datacenter in Tukwila, WA last Sunday — which is a good 3 hour drive. On Tuesday morning, the server was inaccessible. Didn’t even respond to pings.

I called up the Digital Forest guys and have them check up on the server, which I assumed had probably kernel panicked for some reason. Ditto on that. FreeBSD’s GMIRROR had freaked out due to the loss of ad10 (the second drive in the mirror), while the first (ad8) was having trouble with the “TIMEOUT - WRITE_DMA retrying” bullshit that some of us FreeBSD admins have had the pleasure of dealing with.

Brand new drives and they were already on the fritz? I raged hard. I had Digital Forest reboot the server, and it came back up, but the kernel was only detecting one of the drives at this point. An hour later, and it crashed again.

I went to NCIX that day, bought two Samsung HD501LJs to replace the Seagate 7200.11’s (which I no longer trusted), and drove down to Tukwila yet again. I get to the D.C., do my geek thing, start rebuilding the RAID mirror, etc., and within 35 minutes, one of the Samsungs craps out, and the server kernel panics again! I was incredulous. The hard drive had barely been in use for 2 hours before it started clicking wildly.

I gave up. I pulled the server from the rack, raged to Oolz about situation (who braved the noise and cold beside me with her trusty MacBook the whole time), and we took off. I am now the owner of four 500 GB drives, two of which are completely dead, one of which has one foot in the grave, and the last of which I don’t trust worth a damn.

Now, here are the WTFs …

The 7200.11 I have comes with “SD15″ firmware, which is pretty much known to be shit by everyone except Seagate (who is denying it). I wish I had known about this problem BEFORE I bought the drives, but I had a certain level of trust when it came to Seagate. I trusted that they were a company that tested their products, stood behind them, and actively worked on solutions if there was a problem found.

To this day, there has been no response from Seagate about the 7200.11 + SD15 issue, and if you search for SD15 in their Knowledge Base system, you get a single hit for a page titled “Does my drive need a firmware update?”, where Seagate proceeds to explain that they extensively test their shit, and that drives don’t usually need firmware updates.

REALLY? Because it seems that you’re aware enough of the situation to quietly link the search term “SD15″ to go almost directly to this particular article, but you’re not man enough to just go out and say “Okay, there’s definitely a problem with these drives, and we’ll replace them with updated models.”

The best part is that there’s nothing mechanically wrong with the 7200.11’s … it’s all programming error. The firmware issues a bunk command at some point, and locks the non-volative ROM into a useless state. The only way to fix the issue yourself is through some serious electronics/hardware hackery (ie. building your own interface, dicking around with some kind of command language, etc.).

If you Google for 7200.11 SD15 or Seagate SD15, you get reams of results. There’s a statistic going around that the initial failure rate of these drives is around 40% — and that’s not counting the delayed effect of drives in the wild that have yet to fail. Some fail within days, some within weeks, and some within months — but because it’s a software bug, almost everyone will come across it at some point, depending on their usage patterns. That’s the fucking frightening thing. Seagate is doing NOTHING about this. They just keep on selling 7200.11’s with SD15 firmware.

The other WTF is the Samsung HD501LJ that went bad on me within two hours of being installed. I was starting to hear good things about Samsung, and I decided to give them a go. Needless to say, I don’t trust them anymore either.

Western Digital has never failed me. I think I’ll be buying some Western Digitals tomorrow, returning these four drives (which have run me over $300 so far), and making another trip down to Washington.

What a waste of my god damned time and money. Seagate, fix your shit, and stop ignoring your customers, you assholes! Samsung, yours was probably a 1-in-1000 DOA, but that 1-in-1000 fuck-up caught me at a time when I had, and continue to have, NO patience. Sorry.