Re: Deadlock on fragmented memory?

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Thu, Oct 19, 2017 at 09:57:02AM +0200, Martin Husemann wrote:
> Hey folks,
> 
> I sometimes see my shark (arm v4, strongarm 110, 96 MB of ram, main
> characteristics: very *slow* disk access) "hang" with many processes
> blocked on "vmem". All of etc.daily is hit by that, lots of cron instances,
> and no new exec(2) work (or so it seems).

I'm also seeing this on Xen i386 guests doing pbulk builds, both
NetBSD 6 and 7. So it's not new. 
The disks access can be slowed down by concurent accesses.
These guests have several GB of ram, not I'm not sure the amount of RAM is
the key factor. From what I've seen it's virtual memory which is fragmented
too much, not physical memory. There is plenty of free space in the vmem, but
no chunk is big enough (from memory, hung processes are waiting for a 256kb
contigous chunk of VM space).


I've not setup a test machine yet to see if it's reproductible on bare metal.

I'd love to see a fix for this ...

-- 
Manuel Bouyer <bouyer%antioche.eu.org@localhost>
     NetBSD: 26 ans d'experience feront toujours la difference
--