RHBA-2010:0435 - Bug Fix Advisory

Topic

Updated vdsm packages that fix several bugs are now available for Red Hat
Enterprise Linux 5.4 and Red Hat Enterprise Linux 5.5.

Description

The Virtual Desktop Server Manager (VDSM) is a management module that
serves as a Red Hat Enterprise Virtualization Manager (RHEV-M) agent on
RHEV-Hypervisor (RHEV-H) or Red Hat Enterprise Linux hosts.

Note: this update has been tested & is supported on Red Hat Enterprise
Linux 5.4 & 5.5 (with all appropriate post-GA 5.4 & 5.5-specific updates).
(BZ#587055)

Bugs fixed with this update include:

• tap device MAC allocation does not influence the software bridge MAC &

external connectivity. VDSM should not use the MAC address assigned to tap
devices/vnet interfaces by the TAP/TUN driver. (BZ#589060).

• in some circumstances the Storage Pool Manager (SPM), which manages host

& virtual disk relationships via a master file system mounted under
[pooldir]/mastersd/master/, failed to start after a power down due to
corruption of the master file system's Logical Volume (LV). This update
addresses this in two ways. For new installations, the SPM master file
system LV uses ext3 by default. For existing ext2-baesd setups, this update
adds a journal (by adding a .journal file to the file system's root
directory) before mounting the file system. (BZ#586474)

• resume from suspend was broken on iSCSI/FCP storage. (BZ#586479)
• installation over clean Red Hat Enterprise Linux failed since addNetwork

fails when bonding was not insmoded. (BZ#586492)

• VDSM uninstall during service init resulted in wrong editing of "sudoers"

file, leading to a deadlock. (BZ#586491)

• if a virtual machine (VM) was stopped almost immediately after it was

started, TUN (network TUNnel virtual kernel driver) interfaces setup by the
VM remained in place. With this update, TUN device interfaces are taken
down in such an event. Note: this is primarily a presentation fix. A
restarted VM can re-use such forgotten TUN instances without problem.
(BZ#586485)

• when using the RHEV-M interface to manage High Availability (HA) VMs,

power down requests were not honored. Consequently, some HA VMs restarting
automatically after an issued Shut Down command. Power down requests are
now honored when Shut Down commands are sent & all HA VMs power down as
expected. (BZ#586480)

• in some circumstances, when a QEMU process disconnected, a VDSM exception

UnboundLocalError: local variable 'prompt_index' referenced before
assignment

was thrown and the associated VM crashed. As of this update, the
prompt_index variable is referenced correctly, the UnboundLocalError is
avoided and VMs no longer crash when a QEMU process disconnects.
(BZ#586477)

• in some circumstances, if a VM was unable to run at launch, a vdsm bug

caused the newly-created qemu process to remain alive even though VDSM no
longer reported the VM's presence & "destroy" was called. With this update
the errant qemu process is killed as expected in such a case. (BZ#586475)

• on servers under high load some long operations, such as the

de-activation of an old master storage domain after creating two iSCSI
storage pools & setting the second pool as the new master, would fail with
a ProtocolError traceback such as:

ProtocolError for 0:54321/RPC2: -1

For this update, the VDSM time-out was increased from 8 to 60 seconds,
ensuring long operations such as this have time to complete. (BZ#586486)

• race condition after LV extend caused Error Message with (Error code:).

(BZ#586490)

• two vdsm.spec file errors (encountered during development) were

corrected. A missing distribution (dist) tag was added & a typo in the
version tag was corrected. (BZ#589175 & BZ#590449)

• running "unmount -f" caused an Input/Output Error (EIO) if two mounts

shared the same cache (eg, if two NFS storage domains were exported from
the same directory). A "--nosharecache" option has been added to the mount
command to obviate this. (BZ#591478)

RHEV-M 2.1 users with Red Hat Enterprise Linux hosts should install this
package. Alternatively, RHEV-M can install the new package automatically.

Solution

Before applying this update, make sure all previously-released errata
relevant to your system have been applied.

This update is available via the Red Hat Network. Details on how to
use the Red Hat Network to apply this update are available at
http://kbase.redhat.com/faq/docs/DOC-11259

Affected Products

• Red Hat Virtualization 3 for RHEL 5 x86_64

Fixes

CVEs

(none)

References

(none)