> I am not sure how linux will deal with
> losing swap space that is in use. 
> 
Well, now that you asked. Had just that failure last week on one of 
my non-raid hosts. The system is up and responds to some port queries 
but will not connect to telnet. The error message on the console was 
something like 'hole in blah blah... /dev/hda3' which is the high 
priority swap device. I tried doing a make swap on the device and 
starting up again, but it failed again in a few hours. I think the 
disk is history at this point, at least the tracks in the swap space.

Fundamentally, the kernel is hung. It is not smart enough to 
recognize the fault and use the other swap partition.

Michael

Reply via email to