Red Hat Bugzilla – Bug 452240
Unable to install Xen guest with multiple disks
Last modified: 2008-07-28 09:23:31 EDT
Description of problem:
When installing a guest virtual machine with more than one disk the partitions
on the xvd devices are not rescanned leading to an unhandled python exception.
Attached is the patch which we use to workaround this issue.
Version-Release number of selected component (if applicable):
Observed with 4.5 and 4.6 in a Xen guest domain.
Steps to Reproduce:
1. Create a RHEL 4 virtual machine with 2 disks
2. At disk partition page, select Autopartition.
3. At automatic partitioning page,select 'remove all partition on this system', and
select both disk for installation (xvda,xvdb)
4. Rest select default options in installation process.
Unhandled python exception. Screenshot attached (sorry its not terribly useful)
I suspect this is really a bug in parted but have filed against anaconda in the
first instance since this is the component we have patched as a workaround. I'm
not convinced two disks is the relevant factor here but that is how it was
reported to me.
Created attachment 309920 [details]
Workaround in anaconda
Created attachment 309921 [details]
Please attach the complete traceback to this bug report so we can see exactly
what the problem is. Unfortunately, you've only shown the top part which is the
least specific portion of the traceback. Thanks.
The repro steps given in the bug report I received are:
[1..5 are specific to XenServer, basiucally "create a guest with two disks, xvda
6. When RHEL 4.5 installation starts, At disk partition page, select Autopartition.
7. At automatic partitioning page,select 'remove all partition on this system', and
select both disk for installation (xvda,xvdb) (ref screenshot pic2)
8. Rest select default options in installation process.
9. Observe the result.
Unfortunately the only other information they gave was the screenshot which I
passed on to you. I'm attempting to reproduce now but not having much luck. I've
added a collegue who I believe has reproduced the issue in the past to the CC
Unable to reproduce on RHEL-4/U7 doing the steps outlined in comment#1 and comment#4
Closing on the basis of comment #5. If you are able to reproduce this with the
latest update release of RHEL4, feel free to reopen this bug with additional
information and we can take another look at it.
Created attachment 312770 [details]
anaconda traceback part 1/4
Created attachment 312771 [details]
anaconda traceback part 2/4
Created attachment 312772 [details]
anaconda traceback part 3/4
Created attachment 312773 [details]
anaconda traceback part 4/4
I've tried repro-ing this on 4.5 4.6 and 4.7 and I've found that the bug only
affects 4.5, so it must have been fixed in 4.6.
FWIW, I've attached the full traceback obtained from RHEL 4.5.