Red Hat Bugzilla – Bug 149149
fsck fails - no devices node - fcpconf.sh does not verify if device nodes are created
Last modified: 2010-03-18 12:16:16 EDT
Description of problem:
/sbin/zfcpconf.sh does not verify if the device nodes are created. The fsck
required in fstab on that activated devices fails.
Adding a "sleep 10" at the end of zfcpconf.sh solves this problem on my systems.
Version-Release number of selected component (if applicable):
(on s390x as well)
Boot from dasd, add zfcp devices using /etc/zfcp.conf and add the devices into
fstab with checking required.
fsck.ext3: No such file or directory while trying to open /dev/sda1
successful checking of device
I added a "sleep 10" at the end of zfcpconf.sh and this work around the problem
on my system. Values of 3 or 5 seconds were not enough.
For me it looks like udev is to slow to create the device nodes (maybe our
storage has some impact on this as well).
Only waiting for the creation of /dev/sda1 is not a solution as well, because
this would hang the system for ever if the device is not available at all. Maybe
a further analysis of the sysfs subtrees is required before returning from
Created attachment 111229 [details]
Ok, i just talked with our udev guru here, and it's most likely an udev problem.
I'm reassigning it to udev for now.
Read ya, Phil
which version of udev is installed?
I think I have found a solution in the sysload package (http://www.ibm.com/developerworks/linux/linux390/sysload.html) - they use the udevsettle utility (or udevstart when it is not available) - udevsettle watches the udev event queue and exits when all events are processed, udevstart walks through /sys and creates the nodes for the devices it founds.
As RHEL-4.9 is last update for RHEL-4 and it is not suitable for new features
and should address only security, performance and critical issues, I'm closing
that bugzilla WONTFIX. If this functionality is still missing in RHEL-5, feel
free to clone that bugzilla against it.