I have a PXE-based HTTP kickstart routine with an updates.img file. On
one set of systems, the updates.img file never gets downloaded and the
fixes for raid5 corruption it contains never get applied. I've verified
that the web server never sees the request for the updates.img file,
even though it serves the request for the kickstart file and the
subsequent requests for the product.img and netstg2.img files.
If I purposefully break the config and manually enter the kickstart URL,
then it fetches the updates.img as one would expect.
Checking our Cisco logs, the interface does go down for six seconds
after fetching the kickstart file, which I can only assume is anaconda
resetting the interface, and then right after that, it issues requests
for product.img.
Has anyone ever seen this? The server being installed onto is a
dual-Intel gigabit machine that functions flawlessly otherwise. The
machine being installed off of is a dual-Broadcom Opteron box. An
almost identical (servers serving the files are e1000-based Intels)
configuration elsewhere functions flawlessly.
Any information would be appreciated,
Philip
