Description of problem:
I'm trying to automate the install of RHEL images to a test box via LVM. All is
fine if anaconda is completely in charge of the entire LVM setup, but things
fall apart if you attempt to install to a new logical volume on an existing
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Install to a volume group
2. Install a second image using the attached kickstart file
Anaconda dump (attached) while setting up filesystems.
The partitioning setup that is causing trouble is:
partition /boot --fstype=ext3 --noformat --onpart=/dev/sda5
volgroup --noformat shard
logvol / --fstype ext3 --name=i686-rhel4 --vgname=shard --size=8000
logvol swap --fstype swap --name=swap --vgname=shard --size=4000
The /dev/sda5 partition has already been created. The "shard" volume group
already exists, and "--noformat" is documented as meaning "Use an existing
volume group and don't format it."
Created attachment 111626 [details]
Created attachment 111627 [details]
Anaconda dump file
This bug is specifically applicable to creating installs to be used as virtual
hosts, eg: under Xen.
Falling back to creating an install image externally then copying it to an LVM.
A lame response, but is this fixed in an update release of RHEL4, or perhaps
even in RHEL5? It's been quite a while since RHEL4 GA and we've fixed a ton of
bugs since then.
Stephen - It seems to me like this one probably ought to be fixed by now at least in RHEL5, though I do not know for sure. The state of the lvm bugs is kind of a mess. For now I'm going to close this as WORKSFORME, but feel free to reopen this against RHEL5 if you are continuing to see problems. Thanks.
It's conceivable that it has been fixed. But the installs I've done recently have all been from the anaconda GUI, not from kickstart, it's quite possible I'd not have noticed if the bug were still there.
I'll certainly reopen if I see it recur.