I have an Intel D845PEBT2 mobo with onboard IDE and SATA controllers. There are no disks attached to the SATA controller. There is no way to disable the SATA controller in the BIOS. When FC1-test3 boots, it goes thru a several minute delay while it tries to find (non-existent) disks on the SATA controller. This is not very accceptable. You either need to greatly shorten the probe time (to a few seconds at most), or provide a way to permanently disable probing on that controller.
A workaround for this bug is to add 'hde=noprobe hdg=noprobe' to the boot options. This bug seems related to #85828 for Red Hat Linux. I noticed the delay with RH9, but it was about only about a third as long. This appears to be because FC test3 does multiple retries w/ 30-second timeouts for each SATA controller (ide2 and ide3 on a D845PEBT2) while RH9 only does one each.
I have experienced the same problem with FC2 test3, with a Silicon Image SiI3112 on-board controller. I have one SATA drive connected tot he first channel but nothing on the second. I get the same ~30 second wait on booting while the kernel scans for a non-existent drive on the second channel.
The adding of hde=noprobe hdg=noprobe solve the problem of long boot time. This method still does not solve the loading of the "smartd" service failure at boot up. Is there any way to solve the "smartd" problem? Thanks.
Try to change your /etc/smartd.conf to your needs.
This is fixed in fc3test3. Thanks.