Here's the problem I'm having. I have rhel5 2.6.18 kernel, and PowerPath 5.0.1 patch b022 on a Dell 2950. If I have the /dev/emcpower device or an emcpower device LUN wrapped in LVM, in the fstab file, the host fails to boot complaining it cannot find the device.
Looking at the rc.sysinit file, it seems the host runs an fsck on all the filesystems in fstab (if they have that option enabled) before PowerPath is loaded. So, any partition that's in fstab, that has the fsck option set, won't work. If I set the fsck option to 0, which I don't want to do, it works fine.
**Looks like another forum member Celso Capao, has run into the same problem, using the same work around**
Mounting the PowerPath devices with "_netdev" option permits boot-time fsck on PowerPath pseudo-devices (and also on Logical Volumes created out of PowerPath pseudo-devices). The fsck is performed by S25netfs after network services are up (PowerPath is already started by then).
The fsckorder (6th column on /etc/fstab) may be marked as necessary (a value greater than 0 performs boot-time fsck).
is _netdev requirement something new that has to be done with PP5, we have never used _netdev with PP4.5 ..but i wonder if this is something we have to pay attention to when we upgrade to PP5.
The PP5.1 release notes are now available on PowerLink: from p30: === The _netdev option in /etc/fstab PowerPath devices should be mounted with "_netdev" option in /etc/fstab to permit boot-time fsck. For RHEL 5 and 5.1, in an iSCSI setup, PowerIscsi is run as service S24. This configures the PowerPath devices and makes them available to S25netfs to perform fsck on them. ===
the Release notes are non-specific regarding RHEL version, I have checked the OPT that this bug was reported to EMC and it was on the RHEL5 platform... I will do more reseach into previous cases and see if RHEL4 is affected the same way... Will update you shortly.
jem3
6 Posts
1044
0
Posted January 10th, 2008 10:00