Forum
Welcome, Guest
Username: Password: Remember me
  • Page:
  • 1
  • 2

TOPIC:

Issues with XCP-ng 8.0 and VM migration 4 years 6 months ago #1884

  • Fabio Brizzolla
  • Fabio Brizzolla's Avatar Topic Author
  • Offline
  • Posts: 8
Scenario: 2 hosts with fresh and up-to-date XCP-ng 8.0 installation + XOA deployed as test VM (hosted and running on Pool Master)

Issue: live migration from pool master to the slave host results on unresponsive VM, that ends up rebooting itself after an Error Event like this: "Internal error: xenopsd internal error: Domain.Emu_manager_failure('Received error from emu-manager: xenguest Invalid argument')"

Steps to reproduce: just deploy XCP8, run the "Install-NoSAN" and deploy XOA. Try to move from one host to another.

Hope it can be fixed soon.

Regards
Attachments:

Please Log in or Create an account to join the conversation.

Issues with XCP-ng 8.0 and VM migration 4 years 6 months ago #1885

  • Salvatore Costantino
  • Salvatore Costantino's Avatar
  • Offline
  • Posts: 722
We coincidentally have the exact scenario in our lab while v8 is being tested with 4 copies of an xoa VM. All simulated crash scenarios and live migrations are working fine so far.

can you confirm that both hosts are connected to the iscsi SR? I would suggest also raising the issue on the xcp-ng forum in case there is a problem with their new emu manager.

Please Log in or Create an account to join the conversation.

Issues with XCP-ng 8.0 and VM migration 4 years 6 months ago #1887

  • Fabio Brizzolla
  • Fabio Brizzolla's Avatar Topic Author
  • Offline
  • Posts: 8
Thank you so much for the quick response Salvatore!

Yes, both hosts are connected to the iSCSI SR, as you can see in the attached image.

I've ran a new 'yum update' and find out some new up-to-date packages pending installation. Applied them ans did a new test with the same results: the VM hangs and reboots on the destination host.

As you can see I`m doing all my tests using nested virtualization (XCP VMs inside VMware Workstation).

I'll assume that could be a potential reason to the failures before open a ticket to Olivier's guys from XCP-ng.

Regards!
Attachments:

Please Log in or Create an account to join the conversation.

Issues with XCP-ng 8.0 and VM migration 4 years 6 months ago #1888

Having similar issue here, I think also with 7.6 but will test tomorrow again...

At first I was expecting this to be related with my hardware but now running as two libvirt VMs (xcp-ng 8 + halizard nosan installed with 2.1.4 installer)

DRBD is in sync, HA is disabled, manual mode enabled, storage visible on both, VM is Centos 7.6.1810 with tools installed, no ISO attached, not assigned to any xen host, no idea what else could be off..

When trying to migrate VM (xen02 -> xen01) it fails with:
Sep 16 17:03:54 xen01 xapi: [error|xen01|5015 ||backtrace] Raised Server_error(INTERNAL_ERROR, [ xenopsd internal error: Xenops_migrate.Remote_failed("error from remote: Xs_protocol.Enoent(\"directory\")") ])

greping for "directory" in xensource.log on other end:
Sep 16 17:03:54 xen02 xenopsd-xc: [ info|xen02|18 |Async.VM.pool_migrate R:1d034a6021b8|xenops_server] Caught Xenops_migrate.Remote_failed("error from remote: Xs_protocol.Enoent(\"directory\")") executing ["VM_migrate",{"vmm_url":"http://192.168.122.55/services/xenops?session_id=OpaqueRef:2c9d7835-709d-4583-8f75-c09ed96e2ccf","vmm_vgpu_pci_map":{},"vmm_vif_map":{},"vmm_vdi_map":{},"vmm_id":"c1b14927-7b6d-0c41-240e-398c2a6bcdaf"}]: triggering cleanup actions
Sep 16 17:03:54 xen02 xenopsd-xc: [error|xen02|18 |Async.VM.pool_migrate R:1d034a6021b8|task_server] Task 36 failed; Xenops_migrate.Remote_failed("error from remote: Xs_protocol.Enoent(\"directory\")")
Sep 16 17:03:54 xen02 xenopsd-xc: [debug|xen02|18 ||xenops_server] TASK.signal 36 = ["Failed",["Internal_error","Xenops_migrate.Remote_failed(\"error from remote: Xs_protocol.Enoent(\\\"directory\\\")\")"]]
Sep 16 17:03:54 xen02 xapi: [debug|xen02|2144 INET :::80|Async.VM.pool_migrate R:1d034a6021b8|xapi] xenops: not retrying migration: caught Xenops_interface.Xenopsd_error([S(Internal_error);S(Xenops_migrate.Remote_failed("error from remote: Xs_protocol.Enoent(\"directory\")"))]) from sync_with_task in attempt 1 of 3.
Sep 16 17:03:54 xen02 xapi: [error|xen02|2144 INET :::80|Async.VM.pool_migrate R:1d034a6021b8|xenops] Re-raising as INTERNAL_ERROR [ xenopsd internal error: Xenops_migrate.Remote_failed("error from remote: Xs_protocol.Enoent(\"directory\")") ]
Sep 16 17:03:54 xen02 xapi: [ info|xen02|2144 INET :::80|Async.VM.pool_migrate R:1d034a6021b8|xapi] Caught INTERNAL_ERROR [ xenopsd internal error: Xenops_migrate.Remote_failed("error from remote: Xs_protocol.Enoent(\"directory\")") ]: detaching networks
Sep 16 17:03:54 xen02 xapi: [error|xen02|2144 INET :::80|Async.VM.pool_migrate R:1d034a6021b8|xapi] xenops: VM.migrate c1b14927-7b6d-0c41-240e-398c2a6bcdaf: caught Server_error(INTERNAL_ERROR, [ xenopsd internal error: Xenops_migrate.Remote_failed("error from remote: Xs_protocol.Enoent(\"directory\")") ])
Sep 16 17:03:54 xen02 xapi: [error|xen02|2144 INET :::80|dispatch:VM.pool_migrate D:1a50e4613a9a|backtrace] Async.VM.pool_migrate R:1d034a6021b8 failed with exception Server_error(INTERNAL_ERROR, [ xenopsd internal error: Xenops_migrate.Remote_failed("error from remote: Xs_protocol.Enoent(\"directory\")") ])
Sep 16 17:03:54 xen02 xapi: [error|xen02|2144 INET :::80|dispatch:VM.pool_migrate D:1a50e4613a9a|backtrace] Raised Server_error(INTERNAL_ERROR, [ xenopsd internal error: Xenops_migrate.Remote_failed("error from remote: Xs_protocol.Enoent(\"directory\")") ])

I will try to look further tomorrow if time permits. Hope

Please Log in or Create an account to join the conversation.

Issues with XCP-ng 8.0 and VM migration 4 years 6 months ago #1889

  • Salvatore Costantino
  • Salvatore Costantino's Avatar
  • Offline
  • Posts: 722
Hi Jakub and Fabio,
it appears the one thing in common with both of your test environments is that you are running XCP under another hypervisor. I have not seen this condition in our v8 test environment running on HP HW. A mix of PV and HVM VMs (with and without tools installed) are all migrating without issue.

Please Log in or Create an account to join the conversation.

Issues with XCP-ng 8.0 and VM migration 4 years 6 months ago #1890

Hi Salvatore,

Initially my problem started on hardware, that's why I went to test with VMs.

Perhaps the platforms are lacking something essential?
Hardware I tried was two old Supermicro H8DM8-2 with 2xAMD Opteron 2354 CPUs
VM test was done on Fedora 30 under libvirt on a modern Core i5 CPU.

I will try to install xcp-ng 7.6 on the hardware today and test again.

Please Log in or Create an account to join the conversation.

  • Page:
  • 1
  • 2