IBM x360 server: Ran fine for a year on Redhat AS2.1 on the same subnet it's on now. Re-installed to Redhat AS3.0 and after about a month, the NIC started to go offline once per day or so. At the console, we see:

"e100_wait_exec_simple: failed"

You can bring the adapter back on-line by removing and re-inserting the ethernet cable, or by bringing the interface down and back up again. Originally running the e100 driver. Changed to eepro100 today as a test. Holding steady using the older eepro100 driver, but saw some "overruns" in ifconfig output.

Before the NIC goes off-line, we see a build up of "overruns" packets listed in "ifconfig eth0" output,such as the following:
RX packets:363057100 errors:0 dropped:0 overruns:452 frame:0

When the NIC goes off-line, that "overruns:" value may be around 12,000 packets.

Note also, that this same problem has happened on 4 servers, all identical hardware and software on the same subnet. Very random, but we've seen all four servers experience the problem at the same time a couple of times.

Thanks in advance for your suggestions.
Ron-

You might have more luck with the new e100 driver. download at http://prdownloads.sf.net/e1000/e100-3.3.6.tar.gz For this driver make sure to avoid NAPI mode (check the readme)

those wait exec simple failed messages are typically because of the card being unable to transmit due to heavy receive traffic.

Can you double check to make sure that your adapters are auto-negotiating correctly? use ethtool eth0 to check speed and duplex and autoneg settings.

Let us know how it goes. posting a message to e1000-devel at sourceforge would be not bad either.

Jesse (from Intel)