There are some servers on Ubuntu Server 10.04 with that particular hardware: output of hwinfo.
The problem is the hangup when machine experiences a traffic higher than 10Mb/sec for several minutes; only hardware reset can bring the machine back to life.
Before the hangup occures, dmesg displays a repeated message like that:
[106142.897269] r8169 0000:03:00.0: eth0: link up [106142.897280] NOHZ: local_softirq_pending 08 [106146.198770] r8169 0000:03:00.0: eth0: link up [106146.198782] NOHZ: local_softirq_pending 08 [106146.537895] r8169 0000:03:00.0: eth0: link up [106146.537907] NOHZ: local_softirq_pending 08 [106147.525354] r8169 0000:03:00.0: eth0: link up [106147.525365] NOHZ: local_softirq_pending 08
But, of course, the PHY is always connected.
Are there any known ways how to fix it?
It looks like this kernel bug.
You can either downgrade your kernel to one that didn’t exhibit the problem, or upgrade to one in which it is fixed (3.4+).
This work is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License.