I'm trying to track down a problem with a set of machines. They are all using Trenton single-board computers, and they are all having issues. I will post what I have.

The first set are running Trenton XPTs with dual 3.06GHz Xeons, 4GB of RAM. They have dual onboard Intel Gigabit Ethernet interfaces (running the E1000 driver). It has an offboard Adaptec 2110S board which is running the RAID array (RAID 1 mirroring with a hot spare using the i2o driver).

The system is running CentOS 4.2 with a custom-compiled 2.6.9 kernel built from the RPM.

At random times, when copying large amounts of data over to the drive(s), the networking will lock up, and ifconfig will show something similar to the following:



This has occurred on two boxes we are trying to bring up in production, so we set up a test box. The test box originally had an IDE drive and no RAID card, and I was unable to repeat the problem. As soon as we replaced the IDE drive with the 2110S and a SCSI drive, the problem occurred within minutes of my attempting to dump data to it (I was using approximately 7GB of test data).

Has anyone ever seen or heard of interactions between the E1000 and i2o drivers? I have been searching on google for the past 2 or 3 days, and haven't found anything. Right now, we are going to put the IDE drive back in the test box and leave the Adaptec controller in there, but try writing to the IDE to see whether the problem is with the presence of the Adaptec or whether it is an issue of writing to the SCSI drive.

Any suggestions would be deeply appreciated.
--Storm