tool for mapping away bad blocks on an external disk

Michael Powell nightrecon at
Wed Apr 20 15:30:27 UTC 2016

Matthias Apitz wrote:

> El día Monday, April 18, 2016 a las 08:55:34AM +0200, Matthias Apitz
> escribió:
>> Thanks for all the hints; I started last night with overwriting the full
>> disk with:
>> # dd conv=noerror if=/dev/zero of=/dev/da0 bs=1m
>> ...
> The dd ended as. I issued from time to time a kill -INFO to the dd to
> see its progress:
> # tail nohup.out
> ...
> 885016+0 records in
> 885016+0 records out
> 928006537216 bytes transferred in 35244.495687 secs (26330538 bytes/sec)
> 980612+0 records in
> 980612+0 records out
> 1028246208512 bytes transferred in 39028.251450 secs (26346202 bytes/sec)
> dd: /dev/da0: Input/output error
> 1154078+0 records in
> 1154077+0 records out
> 1210137444352 bytes transferred in 45969.382123 secs (26324858 bytes/sec)
> Does this mean I could use the first 1154077 blocks of 1m as partition,
> i.e. shrink it to this size and just ignore the rest?

I have done that on occasion; I actually have the extra storage drive in my 
webdev box partitioned that way now. It had a bad spot which the mfr's 
utility corrected, then a few months later a few more, then a large number 
of them splatted out but they were all in about the same 400MB space in the 
middle of the drive. So I have a small partition in front (with a few 
hundred MB cushion on either side) and a larger partition behind it, with 
the bad space in between unpartitioned.

Not a recommended way to go about things, but I get to use the drive until I 
buy a replacement. I've procrastinated on that as it has now been stable for 
almost a year like that. Generally when you start seeing dead sectors the 
remap zone is full and the first few sectors going bad begin to snowball into 
larger numbers fairly quickly. Failing magnetic media is nothing to waste 
time over as it is only going to cause data loss. But you can do this - I 
just don't really recommend it as it generally is a waste of time because 
once the media begins to fail it will continue to do so. ymmv


More information about the freebsd-questions mailing list