Bug 129528 - error dialog when setting up auto-partitioning
Summary: error dialog when setting up auto-partitioning
Alias: None
Product: Fedora
Classification: Fedora
Component: anaconda
Version: rawhide
Hardware: All
OS: Linux
Target Milestone: ---
Assignee: Jeremy Katz
QA Contact: Mike McLean
Depends On:
Blocks: 133260
TreeView+ depends on / blocked
Reported: 2004-08-10 03:29 UTC by Jens Petersen
Modified: 2007-11-30 22:10 UTC (History)
2 users (show)

Clone Of:
Last Closed: 2004-09-29 18:32:12 UTC

Attachments (Terms of Use)

Description Jens Petersen 2004-08-10 03:29:27 UTC
Description of problem:
When I try to install rawhide over a previous rawhide install
I get an error dialog from the autopartitioning screen.

How reproducible:
every time

Steps to Reproduce:
1. Start installing rawhide.
2. Get to the autopartitioning screen.
3. Press Next.
Actual results:
Dialog pops up with

  Error Partitioning
  Could not allocate requested partitions:

  Unsatisfied partition request
  New Part Request -- mountpoint: None uniqueID: 6
    type: physical volume (LVM) format: 1 badblocks: None
    device: None drive: ['hda'] primary: None
    size: 0 grow: 1 maxsize: None
    start: None end: None migrate: None origfstype: None


A second attempt at auto-partitioning gives a python backtrace.

Expected results:
No errors.

Additional info:
Partitioning manually with diskdruid works fine.
This started happening some weeks ago I think.

Comment 1 Jeremy Katz 2004-08-10 15:18:04 UTC
What is your partitioning setup originally?  What options are you
selecting as far as which drives, etc?  Could you attach the traceback
you get?

Comment 2 Jens Petersen 2004-08-11 08:19:57 UTC
The original partitioning is just a default autopartition
from a previous rawhide install, or the defaults from diskdruid
- so nothing special.

I'll try to retrieve the backtrace when I have time.

Comment 3 Jeremy Katz 2004-09-28 02:33:57 UTC
I'm pretty sure that this is fixed in CVS as of now (dupe of bug 131325)

Comment 4 Jens Petersen 2004-09-29 18:32:12 UTC
Confirmed fixed in rawhide-20040928.  Thanks.

Comment 5 Matthew Miller 2005-07-26 12:56:39 UTC
Bug #131325 is restricted. I'm seeing a similar problem in FC3 and CentOS 4.1
(ie RHEL4 u1) but not in FC4. Any clues as to what the fix is? Thanks!

Comment 6 Paul Nasrat 2005-07-26 13:56:06 UTC
anaconda maintains a pretty active ChangeLog you can use:

        * autopart.py (doAutoPartition): Fix LVM traceback if going back
        from not enough space (#131325)

cvs is your friend - in this case the change seems to be in one file, cvs log
suggests the following:

cvs diff -r 1.147 -r 1.148 autopart.py

It'd be helpful to know the steps you are taking and the setup of the machines,
in case it is not the same - the reproducer for bug #131325 was which should be
in RHEL4:

Remove none (use free space)
Failure message
Change to remove all partitions. 
Get traceback.

Comment 7 Matthew Miller 2005-07-26 14:13:39 UTC
Err, yes, that's nice about the changelog, but as mentioned, the bug referred to
there, #131325, is still restricted just like it was an hour ago.

The problem happens on a bunch of new IBM 345 rackmount servers, with RAID'd
disks (LSI controller). Autopartitioning fails immediately, with no need to
choose remove none and then go back, so it may be a different issue.

It also happens when using a kickstart script with --clearpart all.

Comment 8 Paul Nasrat 2005-07-26 14:23:01 UTC
True, but with the reproducer and the diff you should be able to see what's
going on :)

Please file a new bug for your issue.

Comment 9 Matthew Miller 2005-07-26 14:37:22 UTC
That line (from cvs diff -r 1.147 -r 1.148 autopart.py) is already in the
RHEL4u1 anaconda version, so that's not it. Clearly it's *something* fixed since
FC4, though, and there's rather a few changes to that file since then.

I'll file a new bug shortly.

Comment 10 Matthew Miller 2005-08-18 13:09:19 UTC
New bug is #164633, by the way. And for what it's worth, the "cvs diff -r 1.147
-r 1.148 autopart.py" change is already in centos4/rhel4u1, so that ain't it.

Note You need to log in before you can comment on or make changes to this bug.