TheBonsai's Blog

About the days and nights of TheBonsai

UPDATE: RAID5 and 3 (or 4) disks

July 4th, 2008 by TheBonsai

m00!

This week I finally bought 4 new 500GB harddisks to be used with my Escalade 8506-12. The Installation wasn’t a big deal – simple plug&play. Created the array… surprise! The array initialization stopped after about 5 minutes with tons of

Jul  1 17:12:00 gate kernel: 3w-xxxx: scsi3: AEN: ERROR: Drive error: Port #1.
Jul  1 17:12:00 gate kernel: 3w-xxxx: scsi3: AEN: WARNING: Sector repair occurred: Port #1.

in my syslog files.

I hoped the controller was okay when I swapped the drives on ports #1 and #0… and yes:

Jul  1 17:35:58 gate kernel: 3w-xxxx: scsi3: AEN: INFO: Initialization started: Unit #0.
Jul  1 17:37:54 gate kernel: 3w-xxxx: scsi3: AEN: ERROR: Drive error: Port #0.

Strike, it’s “just” a failing drive, not the controller! I got the S.M.A.R.T. errorlogs from the drive in question and was surprised: The first (power-on selftest) errors were logged immediately after the very first power-on – at lifetime zero! That drive was delivered broken!

Background:

These are not normal drives, this is the RAID edition, which is more expensive. Drives can fail, not a big deal, but I wouldn’t expect a failure at lifetime zero. Next time, I will buy Seagate drives again – it was the first time I tried something from Western Digital (for such a usecase), and immediately 25% of the drives were broken. Either I had luck for now, or Seagate’s drives are better (for the same/similar price!).

My Problem for now is, I can’t start initializing the array until I get the spare drive. 3 HDDs for that planned RAID5 is too less capacity and too less performance. So… waiting.

Sidenote:

On one reboot, the device name for the logical drives on that box swapped, before I had:

  • SDA: USB disk with my favourite Star Trek Voyager sequences
  • SDB: Logical drive of the Escalade controller

As I said, they were swapped after a reboot. I continued my performance tests and wondered why there’s only 20MB/s throughput on SDB (compared to 80MB/s before the reboot)… After writing some gigabytes of random data I finally realized that I destroyed the payload on my USB drive – ARGH!

Conclusions:

  1. I’ll buy Seagate again
  2. if you have important data, keep it away from me 😉

Have a nice day!

UPDATE-01:

Since the mentioned drive was broken, I brought it back. After 2 days I called ’em and asked about it. Answer: “I plugged it in, it works here.” HELL! He didn’t even know what a S.M.A.R.T.-log was… But I was able to convince him to order a new one, finally…

UPDATE-02:

The new disk arrived – the array is complete. Initialising now.

UPDATE-03:

After about 100 days of operation, the second disk begins to fail. Hell! I never saw such an amount of failures with Seagate! Hard disks can fail, yes. But this is a loss of 50% within 100 days

Additiponal complexity: I removed another hard drive for case maintenace (fan was broken), and the broken drive first showed up when I did an array rebuild – I can’t rebuild now :)

I won’t get yet another WD spare disk, I’m fed up. I’ll go with Seagate, I should have done that before…

UPDATE-04:

The LUN is gone after about 2 years. Too many media errors at the same time. These disks definitely were a waste of time and money. Thanks Western Digital. Going to get some Seagate Barracuda ES.2 drives, and all will be okay (I have good expirience with them, we use masses of this type at work, arrays with 64 or more disks – the disks are fine!). Stay away from WD!

This entry was posted on Friday, July 4th, 2008 at 06:15 and is filed under english, Hobby, Linux. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.

1 response about “UPDATE: RAID5 and 3 (or 4) disks”

  1. Usn said:

    That’s the problem with corner shops. Immediate availibitly (maybe), and little expertise. I know these dealers in general, and I know your dealer, and so I know what you are telling between the lines.

Leave a Reply