Re: Page alloc failures under network/disk IO load

[Robert Hancock <hancockr@shaw.ca>]

Dan Noé wrote:
> Francois Romieu wrote:
>> Dan Noé <dpn@isomerica.net> :
>> [...]
>>> I'm a bit confused because on another system (2.6.26.3) I never see
>>> messages like this despite having the same amount of physical RAM in
>>> each.  The 2.6.26.3 system is also under more active use, and has more
>>> userspace memory usage.  On that system:
>> Does this system use the same network card ?
> 
> Nope, good point :)  The system experiencing the allocation failures is
> using an rtl8169 gigabit NIC (at 1gbps), the 2.6.26.3 system is on a
> 100M connection with a gigabit Broadcom BCM5721 / tg3.
> 
> 02:00.0 Ethernet controller: Realtek Semiconductor Co., Ltd.
> RTL8111/8168B PCI Express Gigabit Ethernet controller (rev 01)
> 
> 04:00.0 Ethernet controller: Broadcom Corporation NetXtreme BCM5721
> Gigabit Ethernet PCI Express (rev 11)
> 
> I can see how this could account for different behavior, although
> presumably both drivers are calling __netdev_alloc_skb.

The Realtek driver is doing an order 1 allocation (an 8KB chunk). That 
is more likely to fail than an order 0. Are you using jumbo frames or 
higher than 1500 byte MTU on the network?