Description of problem:

As reported by Matthias Bolte on the libvirt developer mailing list, the libvirt source code has reference leaks in several functions:

  http://www.redhat.com/archives/libvir-list/2010-June/msg00331.html

  remoteDispatchDomainMemoryStats()
  remoteDispatchDomainMigrateFinish2()
  remoteDispatchDomainIsActive()
  remoteDispatchDomainIsPersistent()
  remoteDispatchInterfaceIsActive()
  remoteDispatchNetworkIsActive()
  remoteDispatchNetworkIsPersistent()
  remoteDispatchStoragePoolIsActive()
  remoteDispatchStoragePoolIsPersistent()

This leads to incorrect information being passed back to a libvirt client program if the object it is working on has been recreated.

For example:

    virsh # pool-create-as images_dir3 dir - - - - "/home/images2"
    Pool images_dir3 created

    virsh # pool-info images_dir3
    Name:           images_dir3
    UUID:           90301885-94eb-4ca7-14c2-f30b25a29a36
    State:          running
    Capacity:       395.20 GB
    Allocation:     30.88 GB
    Available:      364.33 GB

    virsh # pool-destroy images_dir3
    Pool images_dir3 destroyed

    virsh # pool-create-as images_dir3 dir - - - - "/home/images2"
    Pool images_dir3 created

    virsh # pool-info images_dir3
    Name:           images_dir3
    UUID:           90301885-94eb-4ca7-14c2-f30b25a29a36
    error: Storage pool not found

    virsh #

In the above, the "images_dir3" storage pool was created, and libvirt generated a new UUID for it at creation time.  However, the old UUID is incorrectly still being used to refer to this recreated pool instance, causing the above "Storage pool not found" failure.


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

In RHEL 6 beta1: libvirt-0.7.6-2.el6.x86_64.rpm
In Fedora 14 (rawhide): libvirt-0.8.1-1.fc14.x86_64.rpm


How reproducible:

Every time.


Steps to Reproduce:
1. Compile a new virsh, with a call to virStoragePoolIsPersistent(pool) added to the virsh "pool-info" function cmdPoolInfo().
2. Create a directory "/tmp/testbug" on a server with libvirtd running.
3. Use the new virsh connect to the libvirtd server.
4. Inside virsh, create a new storage pool:

   # pool-create-as mypool dir - - - - "/tmp/testbug"
   Pool mypool created

5. Still inside the same virsh session, destroy the storage pool:

   # pool-destroy mypool
   Pool mypool destroyed

6. Still inside the same virsh session, re-create the storage pool using the same name:

   # pool-create-as mypool dir - - - - "/tmp/testbug"
   Pool mypool created

7. Attempts to access this pool by name (in this virsh session) will now fail:

   # pool-info mypool
   Name:           mypool
   UUID:           90301885-94eb-4ca7-14c2-f30b25a29a36
   error: Storage pool not found


Actual results:

Recreated storage pools are unable to be found correctly in a given client session.


Expected results:

Recreated storage pools should work "as per normal".


Additional info:

Matthias Bolte's patch submitted to the libvirt developer mailing list is known to fix this problem in git head:

  http://www.redhat.com/archives/libvir-list/2010-June/msg00331.html

This bug also affects the libvirtd in present Fedora 14 rawhide.