Hi,

We have a 48 node Linux Cluster running redhat but we seem to getting performance issues on the server. The server has 2 nics (1 for the public network(eth1) & the other to connect to the nodes(eth0)). When I run the following command on both nics I get the following :

dmesg | grep eth0
divert: allocating divert_blk for eth0
eth0: Intel(R) PRO/1000 Network Connection
e1000: eth0 NIC Link is Up 1000 Mbps Full Duplex

dmesg | grep eth1
divert: allocating divert_blk for eth1
eth1: Hamachi GNIC-II type 10911 at 0xf89b3000, 00:e0:b1:45:67:ca, IRQ 28.
eth1: 64-bit 33 Mhz PCI bus (60), Virtual Jumpers 30, LPA 0000.
eth1: Too much work at interrupt, status=0x0001.
eth1: Too much work at interrupt, status=0x0001.

Any ideas on what : eth1: Too much work at interrupt, status=0x0001?????. It is running gigabit (fibre)

Thanks

There are (or at least were; are you that lucky?!) some bugs introduced by someone doing some "clean-ups" inserted in the driver for hamachi, specifically in hamachi.h. See if there are any driver updates out there for your NIC. There might be various other reasons as well but less probable. Your NIC may be just screwed for the time being or the fiber physical connection is not good. However i assume is the crappy driver. Or if you recently added some other PCI card in your box there might be an IRQ conflict out there. You need to check everything or just change the NIC

Regards
Andy