Wouldn't it be better to fix the bug in udev to that it doesn't try to
access blocks that are past the end of the device, even if the partition
table seems to say they exist?
James Olson wrote:
The new version of the udev package in Fedora Core 4 updates is udev-071-0.FC4.2, and it tries to scan drive partitions to create
some new devices such as those in the new /dev/disk/by-id/ directory.
With dmraid you don't want it to scan your striped raid drives individually or it will generate errors as it tries to
seek past the end of the drive with the "big" partition on it. My raid
drives are /dev/hde, /dev/hdf, and /dev/hdg but you can modify it for
your own raid devices that you don't want udev to scan.
Here is the patch to the 50-udev.rules file in /etc/udev/rules.d/
# diff -Naur 50-udev.rules.orig 50-udev.rules
--- 50-udev.rules.orig 2006-01-31 08:09:15.000000000 -0800
+++ 50-udev.rules 2006-02-05 14:13:37.000000000 -0800
@@ -255,6 +255,9 @@
# skip rules for inappropriate block devices
KERNEL=="ram*|loop*|fd*|nbd*", GOTO="persistent_end"
+# skip dmraid drives
+KERNEL=="hde*|hdf*|hdg*", GOTO="persistent_end"
+
# never access removable ide devices, the drivers are causing event loops on open()
BUS=="ide", DRIVER!="ide-cdrom", SYSFS{removable}="1", GOTO="persistent_end"
_______________________________________________
Ataraid-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/ataraid-list
