Re: Processes stuck in unkillable D state (2.4 and 2.6)

["Rob Mueller" <robm@fastmail.fm>]

> Well, you've got two procs waiting for pages but it isn't entirely clear
> why they aren't getting them.  There have been quite a few fixes in this
> area since 2.6.4, how hard is it for you to upgrade?

We can upgrade, and probably will soon. It was just that this problem has 
been occuring for a long time (as I mentioned, since 2.4.18) and this is the 
first time I'd managed to capture some sysreq-t output for this problem, and 
I thought that the infornation might be in some way immediately useful. 
Given that it's stuck in io_schedule, and the long term nature of the 
problem, does that suggest a hardware/device driver issue maybe? It just 
surprised me that the backtrace for both procs showed they got stuck going 
through the exact same code path, wasn't sure if it was a coincidence or a 
sign of something definitely going funny in that code path.

Anyway, we're currently doing some stress tests, and will probably upgrade 
in the next week or two to a newer kernel. I'll see what happens then.

Rob