Re: [PATCH] fix spurious OOM kills

[Chris Ross]

Marcelo Tosatti escreveu:
> If its not the case, increasing the all_unreclaimable "timer" to a higher value
> than 5 seconds will certainly delay the OOM killer such to a point where 
> its not triggered until the VM reclaiming efforts make progress.
[...]
> Chris, can you change the "500*HZ" in mm/vmscan.c balance_pgdat() function
> to "1000*HZ" and see what you get, please?

Changed. FWIW it's been running happily for hours without a single oom, 
including the normally guaranteed build UML test. I'll leave it running 
and see how it goes. The daily cron run is a usually a popular time for 
killing off a few essential daemons (ntpd, sshd &c), in fact I think the 
OOM Killer actually looks forward to it :)

Regards,
Chris R.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/