0

As Chris Benninger says over at: http://www.benninger.ca/?p=58 lots and lots of people have the problem with Squeeze and xen4.0 telling them: 

launch_vm: SETVCPUCONTEXT failed (rc=-1)

but nobody seems to know what the solution is. I don't know either, but at least here, a solution might get recorded.

In my case, I can start one guest machine. An identical configuration for a second machine fails. Whichever one I start first, is the one that runs, the other gets the error.

I've got at least a dozen other systems (at my work) running great with Squeeze and 64-bit XEN, but not this new machine at home.

mcr
  • 233
  • 1
  • 10

3 Answers3

2

check http://wiki.debian.org/Xen "Configure Domain 0 Memory"

/etc/default/grub
GRUB_CMDLINE_XEN="dom0_mem=1024M"

/etc/xen/xend-config.sxp
(dom0-min-mem 1024)
(enable-dom0-ballooning no)

and no more "launch_vm: SETVCPUCONTEXT failed (rc=-1)" for me.

alf
  • 21
  • 2
0

I hit this, with a configuration for vcpus=2. I changed it to vcpus=1 and then the VM started OK.

I currently have 6 cores, cores 0&1 pinned to the Dom0, and then some CPU's pinned, some not. I had a total of 9 cores in use when I hit this. Changing to vcpus=1 made it 10 cores total.

Next time I reboot the DomU's will be pinned to cores 2-5, to see if that makes a difference. I feel like some limit is being reached here, but I'm not sure what that is.

Bill McGonigle
  • 1
  • I had 4 cores (single socket Xeon), no pinning, and vcpus=1 everywhere, and I could only run a single guest machine. – mcr Oct 31 '11 at 18:51
0

When I hit this, i had 3 guests running (4 cores available). When i stopped one, i could start the one that failed before. Unfortunately the stopped one ("halt" in the vm) was completely broken (ext4: mount failed, fsck killed all files, lvm failed to remove the volumes...).

Now i reduced the assigned memory, and it works (after recreating the broken image) - so perhaps check whether you assigned more memory than you actually have.

Stefan
  • 859
  • 1
  • 7
  • 18