Description of problem: While migrating all domU's from one machine to another, I found that if the source machine started with 6 domU's, the live migration would fail. On the source, if I xm console the VM that failed migration, I see this output: WARNING: g.e. still in use! WARNING: leaking g.e. and page still in use! WARNING: g.e. still in use! WARNING: leaking g.e. and page still in use! netif_release_rx_bufs: 0 xfer, 43 noxfer, 213 unused WARNING: g.e. still in use! WARNING: leaking g.e. and page still in use! To recover I have to xm destroy on the source; then I can xm create on the dest. I don't know if there is a hardware relation here, but the source and dest machines are both Dell 6950 w/ 4 AMD dual-core opteron and 16GB RAM. Other cases I tried: - Migrating a different VM (failed w/ 6 running on source) - Migrating a VM with only 5 on source - succeeded Version-Release number of selected component (if applicable): All dom0 domU completely current as of 9/25/08: kernel-xen-2.6.18-92.1.13.el5 xen-3.0.3-64.el5_2.1 How reproducible: Always for me Steps to Reproduce: 1. Run 6 domUs on a single host, at least 1 on shared storage 2. Live migrate shared VM from this host to another Actual results: domU crashes with above message on console Expected results: successful migration Additional info:
Another report of the "leaking" messages in rhts log at: http://rhts.redhat.com/cgi-bin/rhts/test_log.cgi?id=5596480
I've just tested this on RHEL5.3, and this issue appears to be fixed. Versions: kernel-xen-2.6.18-128.el5 xen-3.0.3-80.el5 I moved 3 VMs to a server that already hosts 4 (for a total of 7), and was able to migrate them all back without incident.
Thanks for the update. Closing as fixed in the current release. Please reopen if it reoccurs.