WD15EADS - ever going to be fixed, or do I take it behind the shed and put it out of my misery?

WD Internal Drives Desktop & Mobile Drives
1

I grabbed it a while back as a spare, tried to use it and found that I get a whopping 800k/sec.

Don’t “SATA controller” me, don’t “check for bad blocks”, everyone knows the EADS drives have stalling problems:

8,97 1 1 0.000000000 0 C W 1281714176 + 1024 [0]
  8,97 1 2 0.006678520 0 C W 1281715200 + 1024 [0]
  8,97 1 3 0.012108944 0 C W 1281716224 + 1024 [0]
  8,97 1 4 0.018799199 0 C W 1281717248 + 1024 [0]
  8,97 1 5 0.018828043 0 D R 8492168 + 8 [swapper/1]
  8,97 1 6 1.811002781 0 C R 8492168 + 8 [0]
  8,96 0 1 1.811318106 1853 A R 6393792 + 8 <- (8,97) 6391744
  8,97 0 2 1.811320062 1853 Q R 6393792 + 8 [btrfs-transacti]
  8,97 0 3 1.811334728 1853 G R 6393792 + 8 [btrfs-transacti]
  8,97 0 4 1.811341572 1853 I R 6393792 + 8 ( 6844) [btrfs-transacti]
  8,97 0 5 1.811368461 1853 D R 6393792 + 8 ( 26889) [btrfs-transacti]
  8,97 1 7 3.637386061 0 C R 6393792 + 8 (1826017600) [0]
  8,97 1 8 3.637453526 0 D W 1281718272 + 1024 [swapper/1]
  8,97 1 9 3.659819069 0 C W 1281718272 + 1024 [0]
  8,97 1 10 3.659868934 0 D W 1281719296 + 1024 [swapper/1]
  8,97 1 11 3.662621789 0 C W 1281719296 + 1024 [0]
  8,97 1 12 3.662643300 0 D W 1281720320 + 1024 [swapper/1]
  8,97 1 13 3.665554551 0 C W 1281720320 + 1024 [0]
  8,97 1 14 3.665580950 0 D W 1281721344 + 576 [swapper/1]
  8,97 1 15 3.667292502 0 C W 1281721344 + 576 [0]  8,97   1       16     3.667312546     0  D   W 1281721920 + 1024 [swapper/1]  8,96   0        6     3.637757606  1853  A   R 6480000 + 8 <- (8,97) 6477952  8,97   0        7     3.637759562  1853  Q   R 6480000 + 8 [btrfs-transacti]  8,97   0        8     3.637773739  1853  G   R 6480000 + 8 [btrfs-transacti]  8,97   0        9     3.637780094  1853  I   R 6480000 + 8 (    6355) [btrfs-transacti]  8,97   1       17     5.470261601     0  C   W 1281721920 + 1024 [0]  8,97   1       18     5.470339821     0  D   R 6480000 + 8 (1832559727) [swapper/1]  8,97   1       19     7.266953054  3883  C   R 6480000 + 8 (1796613233) [0]

 That’s blktrace in linux - 4th colum is time.  “D” is dispatch, “C” is complete.  _ONE POINT EIGHT SECONDS_ to return a single IO.  This is typical behavior for these drives, with bursts of speed where they actually work for a bit, then stall stall stall stall.   It’s not spinning down, it’s just not doing anything.  Bad batch of servos, can’t seek properly?  Who knows.

I’ve followed thread after thread on these **bleep** drives, and nobody has ever had an answer.   Send back your new drive, get a used drive instead.  If I _WANTED_ a used drive I would have _BOUGHT_ one at a discount.

2

It seems odd to me that the delay is consistently 1.8 sec, even for writes.

I would scan the drive from start to finish using MHDD. You may need to temporarily reconfigure your BIOS for legacy or IDE compatibility mode. MHDD will identify any “slow” blocks, ie those that require more than 500ms to be read. If the resultant bad block map has a checkerboard pattern, then this would tend to indicate a “weak” head.

Alternatively, you could try using dd to copy blocks of 10MB, say, to dev/null, from various zones on the disk. Then look for a pattern in your blktrace log. If there is no pattern, and the slowdowns are truly random, then this will at least eliminate a bad head as a possible cause.