Dear Nils,
error messages means that restore of network device inside container was failed.
Setting on the host should not be related to this issue in this particular case.
Please try to upgrade kenrels on your nodes, probably it helps.
I do not remember this concrete problem, but we have fixed few CPT related
issues in last kernels,
so probably we have fixed problem observed on your nodes too.
If not -- please report us about this problem via openvz bugzilla.
Happy new year to you, and thank you,
Vasily Averin
On 07.01.2015 14:59, Nils Effenberger wrote:
> Happy new year folks,
>
> we've got an openvz cluster with two nodes running debian 7.6 and
> 2.6.32-openvz-042stab092.3-amd64.
> Networking of the containers is done via bridged vlan interfaces.
> vlan-raw-device is a bond interface in mode 1 (failover) over eth0 and
> eth1.
>
> The interfaces where added in the containers via
> vzctl set CTID --netif_add=eth0,,,,<bridge>
> this triggers an unmodiefied /usr/sbin/vznetaddbr
>
> When I try to live-migrate a container to the other node I run into
> following error:
> root@openvz1:~# vzmigrate -v -f=cpu -r no \
> --live --keep-dst openvz2 651700
>
> [...]
> Container is mounted
> undump...
> Setting CPU units: 1000
> Setting CPUs: 1
> Configure veth devices: veth651700.0
> Adding interface veth651700.0 to bridge br507 on CT0 for CT651700
> Restore error, undump failed: No such device
> Error: netdev 4115 rst failed -19
> Error: rst_restore_net: -19
> Container restore failed
>
> Container is unmounted
> Error: Failed to undump container
> Resuming...
>
> Relevant kernel messages of the destination node:
>
> Jan 6 12:48:09 openvz2 kernel: [12206877.941140] CT: 205623: checkpointed
> Jan 6 16:01:36 openvz2 kernel: [12218496.823161] CT: 651700: started
> Jan 6 16:01:37 openvz2 kernel: [12218497.648247] device veth651700.0 entered
> promiscuous mode
> Jan 6 16:01:37 openvz2 kernel: [12218497.648601] br507: port 2(veth651700.0)
> entering listening state
> Jan 6 16:01:37 openvz2 kernel: [12218497.659436] CPT ERR:
> ffff881f75271000,651700 :netdev 4115 rst failed -19
> Jan 6 16:01:37 openvz2 kernel: [12218497.659706] CPT ERR:
> ffff881f75271000,651700 :rst_restore_net: -19
> Jan 6 16:01:37 openvz2 kernel: [12218497.702244] br507: port 2(veth651700.0)
> entering disabled state
> Jan 6 16:01:37 openvz2 kernel: [12218497.719653] device veth651700.0 left
> promiscuous mode
> Jan 6 16:01:37 openvz2 kernel: [12218497.719934] br507: port 2(veth651700.0)
> entering disabled state
> Jan 6 16:01:37 openvz2 kernel: [12218497.883689] CT: 651700: stopped
>
> NETIF variable of the container:
> NETIF="ifname=eth0,bridge=br507,mac=00:18:51:F6:C9:BB,host_ifname=veth651700.0,host_mac=00:18:51:0E:74:94"
>
> bridge on the source node:
> root@openvz1:/etc/vz# brctl show br507
> bridge name bridge id STP enabled interfaces
> br507 8000.0018510e7494 yes veth651700.0
> vlan507
> bridge on the destination node:
> root@openvz2:~# brctl show br507
> bridge name bridge id STP enabled interfaces
> br507 8000.002590f456cf yes vlan507
>
> If you need further information let me know.
>
> BTW: Live migration works on a secend cluster with the same configuration
> without problems. The only difference is the kernel version:
> 2.6.32-openvz-042stab094.7-amd64
>
> And I'm pretty sure that it was working last year and "I didn't do anything"
> ;)
>
> Thanks in advance for any feedback.
>
> Regards,
> Nils
>
> --
> Nils Effenberger
> UNIX-Administration
>
> n@work Internet Informationssysteme GmbH Wandalenweg 5, 20097 Hamburg,
> http://www.work.de E-Mail: effenber...@work.de, Tel: +49 40 2388090 -
> Fax: +49 40 23880929
>
> HR B 61 668 - Amtsgericht Hamburg, Geschaeftsfuehrer: Jan Diegelmann
> _______________________________________________
> Users mailing list
> Users@openvz.org
> https://lists.openvz.org/mailman/listinfo/users
>
_______________________________________________
Users mailing list
Users@openvz.org
https://lists.openvz.org/mailman/listinfo/users
