Bug 512680

Summary:	errors while altering RPM package database
Product:	[Community] Virtualization Tools	Reporter:	Marek Goldmann <mgoldman>
Component:	libguestfs	Assignee:	Richard W.M. Jones <rjones>
Status:	CLOSED DUPLICATE	QA Contact:
Severity:	medium	Docs Contact:
Priority:	low
Version:	unspecified	CC:	mbooth, rjones, virt-maint
Target Milestone:	---
Target Release:	---
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2009-07-21 12:51:15 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Marek Goldmann 2009-07-20 11:28:27 UTC

Description of problem:

When executing rpm --rebuilddb (or any other switch which alters package database), libguestfs is exitting with an error.

Version-Release number of selected component (if applicable):

Host:

[oddthesis@lolek-f11 ~]$ uname -a
Linux lolek-f11 2.6.29.5-191.fc11.i686.PAE #1 SMP Tue Jun 16 23:19:53 EDT 2009 i686 i686 i386 GNU/Linux

Guest:

Same as host.

[oddthesis@lolek-f11 downloads]$ rpm -qa | grep guestfs
libguestfs-1.0.61-1.fc11.i586
libguestfs-java-1.0.61-1.fc11.i586
libguestfs-javadoc-1.0.61-1.fc11.i586
libguestfs-devel-1.0.61-1.fc11.i586
perl-libguestfs-1.0.61-1.fc11.i586
ocaml-libguestfs-1.0.61-1.fc11.i586
libguestfs-java-devel-1.0.61-1.fc11.i586
ruby-libguestfs-1.0.61-1.fc11.i586
libguestfs-debuginfo-1.0.61-1.fc11.i586
python-libguestfs-1.0.61-1.fc11.i586


How reproducible:

Always.

Steps to Reproduce:
1. Grab disk image from: http://repo.oddthesis.org/dev/management-appliance.raw.gz
2. Launch guestfish: guestfish -iv management-appliance.raw
3. Remove old db lock files: sh "rm /var/lib/rpm/__db.*"
4. Try to rebuild database: command "rpm --rebuilddb"
  
Actual results:

Different errors:
- Error without information
- rpm segfault in different libs
- no vm86_info: BAD error

Expected results:

No error, RPM package database rebuild.

Additional info:

On 64 bit host (and guest), everything works great.

Comment 1 Marek Goldmann 2009-07-20 11:30:22 UTC

Output from guestfish: http://gist.github.com/150259

Comment 2 Richard W.M. Jones 2009-07-20 15:43:32 UTC

I can't reproduce this at all.  I tried:

 - 32 bit w/ HVM
 - 32 bit w/o HVM
 - 64 bit w/ HVM
 - 64 bit w/o HVM

and all worked perfectly.

I don't have any guests running on VMWare ESX.  Please try to
reproduce this on a baremetal host.  If it doesn't happen on
a baremetal host, then it's a bug in VMWare, so you should ask
VMWare to look at it.

> - rpm segfault in different libs
> - no vm86_info: BAD error

Both of these errors indicate serious, random corruption in
qemu's emulation (TCG in this case).  Maybe VMWare can't run
qemu correctly.

Since it seems you have general problems running binaries in this
configuration, I strongly suggest you use a binary such as
/bin/false or:

  main () { exit (0); }

and try to get that to fail, with a stacktrace, on a baremetal
host.  There are just too many variables when you're trying to
run some process as complex as 'rpm --rebuilddb' to know where
to begin.

Comment 3 Richard W.M. Jones 2009-07-21 12:51:15 UTC

I'm now convinced this is a dup of 502074.

*** This bug has been marked as a duplicate of bug 502074 ***