Slow ddrescue recovery problem (slow reads? excessive retries?)

mventimi · 09-17-2010, 11:38 AM

Hi,

I'm in the process of recovering data from a bad HDD using ddrescue. I've done this a few times, and I'm pretty comfortable with using ddrescue. However, things are moving very slowly with the HDD I'm working on right now. ddrescue seems to grab ~200MB off the disk very quickly (~5MB/s), then gets hung up on the next ~65MB, reading at 1-2KB/s. When I look at /var/log/messages, I get this every 10s:

Code:

Sep 17 10:28:23 localhost kernel: ata5.00: exception Emask 0x0 SAct 0x1 SErr 0x0 action 0x0
Sep 17 10:28:23 localhost kernel: ata5.00: irq_stat 0x00060002, device error via SDB FIS
Sep 17 10:28:23 localhost kernel: ata5.00: cmd 60/00:00:98:31:f4/01:00:0d:00:00/40 tag 0 cdb 0x0 data 131072 in
Sep 17 10:28:23 localhost kernel:          res 41/40:00:98:31:f4/95:00:0d:00:00/6d Emask 0x409 (media error) <F>
Sep 17 10:28:23 localhost kernel: ata5.00: configured for UDMA/100
Sep 17 10:28:23 localhost kernel: ata5: EH complete
Sep 17 10:28:23 localhost kernel: sd 4:0:0:0: [sdb] 488397168 512-byte hardware sectors (250059 MB)
Sep 17 10:28:23 localhost kernel: sd 4:0:0:0: [sdb] Write Protect is off
Sep 17 10:28:23 localhost kernel: sd 4:0:0:0: [sdb] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA

From looking at /sys/block/sdb/stat, the longest reads appear to take about 100s to complete.

I'd like to speed up the reads. I'm going to go back and retry the bad blocks later, anyway, but I wanted to grab as much as I can from the disk on the first pass. I'm worried that all of the read retrying is just further damaging the disk.

Does anyone have any ideas on how I can either decrease the number of read retries, or decrease the amount of time before the reads time out? These answers, or any other suggestions, would be appreciated.

Thanks!

unSpawn · 09-17-2010, 01:08 PM

Could use dd_rescue which unlike ddrescue can read back to front. Did you decipher these SCSI warnings beforehand to get an idea of what's up?

mventimi · 09-17-2010, 01:25 PM

I decoded the SCSI errors, and they don't really have any useful information beyond indicating a pretty standard media error on the drive. This is what I'd expect on this drive.

I also tried dd_rescue, and it has the same problem with slow reads. I'm thinking that there are chunks of the disk that are damaged but still readable. The problem is that I actually get something back after a few retries, and the driver is happily doing the retries until it reads something. I'd like to be able to tell the disk driver not to retry, or to only retry for 1s, or something like that, so that I can grab as much good data as I can on the first pass.

unSpawn · 09-21-2010, 09:08 AM

IIRC the docs should show the amount of retries can be configured in both tools.

rob.rice · 09-22-2010, 07:03 PM

the drive is heating up
with marginal electronics the hotter they get the worse they work
there is no software solution to this the only thing you can do is
run it for a few minuets at a time and let it cool or even stick it in the fridg
to buy more time or run your computer with the case open and a fan blowing on
the hard drive's electronics