[Gllug] dd and iotop weirdness

[general_email at technicalbloke.com]

On 30/11/10 09:51, Chris Bell wrote:
> On Mon 29 Nov, general_email at technicalbloke.com wrote:
>
>    
>> What on earth is going on?
>>      
>    
>> Yours confused,
>>
>> Roger.
>>      
>     Was the bootable live CD trying to access swap space on the first drive?
>
>    


No, the drives were gpt/ufs formatted raid 1 volumes from a bsd box, 
there was no swap space on them. Also they were a fair way in and one 
was aleady partially wiped (and the bootsector would have been the first 
to go) :/ I did remove the drive that stopped, scanned it for errors and 
it found none, although the other slower one did turn up several bad 
sectors. Much as I would have liked to get to the bottom of it I was 
pushed for time so I put the two drives in two separate machines and all 
was fine from there on in, I've returned the slower faulty one to the 
retailer now.

Still I'm puzzled as to how one sata drive, even if it has faults, could 
knock the other offline. I would chalk it down to coincidence if it 
wasn't instantaneous (the LED in the caddy went out as soon as I started 
ddrescue). I was wondering if it might have anything to do with "watch". 
I was surprised to see the output from both watches appear in the same 
console even though I started them in separate terminals, that suggests 
to me some kind of shared state / class attributes that might allow the 
processes to affect each other. Having said that I thought watch worked 
by sending "signals" via the operating system rather than by any shared 
memory shenanigans. TBH I don't know nearly enough about either watch or 
signals to make any informed comments so I'll stop rambling now!

Thanks to everyone else who replied BTW :)

Oh and also, I have experienced a similar situation trying to run 
several large rsync jobs simultaneously on one box in the past, to the 
extent that I'm a little cagey about doing it now. Henceforth I haven't 
run into it for a while so I'm a bit sketchy on the details but I seem 
to remember it happening when starting a local rsync while I had an ssh 
one already running, or maybe it was the other way round. Anyway, I'm 
worried I might have developed a superstition there, is multiple 
simultaneous rsyncing of large volumes (>100GB) something people do 
without problem on a regular basis?

Roger
-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug