On Tue, Aug 10, 2021 at 1:13 AM Laurent Vivier <lviv...@redhat.com> wrote: > > v2: use validate_features() to disable the guest driver rather > than setting vring.num to 0. > > With failover, when the guest virtio-net driver doesn't support the > STANDBY feature, the primary device is not plugged and only the virtio-net > device is kept. Doing like that we can migrate the machine and > keep the network connection. > > But in some cases, when performance is more important than availability > we would prefer to keep the VFIO device rather than the virtio-net one, > even if it means to lose the network connection during the migration of > the machine.
I think we still need to seek a way to recover the network after migration. Thanks > > To do that we can't simply unplug the virtio-net device and plug the > VFIO one because for the migration the initial state must be kept > (virtio-net plugged, VFIO unplugged) but we can try to disable the > virtio-net driver and plug the VFIO card, so the initial state is > correct (the virtio-net card is plugged, but disabled in guest, and > the VFIO card is unplugged before migration). > > This change doesn't impact the case when guest and host support > the STANDBY feature. > > I've introduced the "failover-default" property to virtio-net device > to set which device to keep (failover-default=true keeps the virtio-net > device, =off the other one). > > For example, with a guest driver that doesn't support STANDBY: > > ... > -device virtio-net-pci,id=virtio0,failover=on,failover-default=on \ > -device vfio-pci,host=$PCI,id=hostdev0,failover_pair_id=virtio0 \ > ... > > [root@localhost ~]# ip a > 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN qlen 1 > link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 > inet 127.0.0.1/8 scope host lo > valid_lft forever preferred_lft forever > inet6 ::1/128 scope host > valid_lft forever preferred_lft forever > 2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state > UP q0 > link/ether 26:28:c5:7f:14:24 brd ff:ff:ff:ff:ff:ff > inet 192.168.20.2/24 brd 192.168.20.255 scope global eth0 > valid_lft forever preferred_lft forever > inet6 fe80::2428:c5ff:fe7f:1424/64 scope link > valid_lft forever preferred_lft forever > # ethtool -i eth0 > driver: virtio_net > version: 1.0.0 > firmware-version: > expansion-rom-version: > bus-info: 0000:04:00.0 > supports-statistics: no > supports-test: no > supports-eeprom-access: no > supports-register-dump: no > supports-priv-flags: no > > ... > -device virtio-net-pci,id=virtio0,failover=on,failover-default=off \ > -device vfio-pci,host=$PCI,id=hostdev0,failover_pair_id=virtio0 \ > ... > > [root@localhost ~]# ip a > 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN qlen 1 > link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 > inet 127.0.0.1/8 scope host lo > valid_lft forever preferred_lft forever > inet6 ::1/128 scope host > valid_lft forever preferred_lft forever > 2: enp2s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP > qlen 100 > link/ether 26:28:c5:7f:14:24 brd ff:ff:ff:ff:ff:ff > inet 192.168.20.2/24 brd 192.168.20.255 scope global enp2s0 > valid_lft forever preferred_lft forever > inet6 fe80::2428:c5ff:fe7f:1424/64 scope link > valid_lft forever preferred_lft forever > [root@localhost ~]# ethtool -i enp2s0 > driver: i40evf > version: 1.6.27-k > firmware-version: N/A > expansion-rom-version: > bus-info: 0000:02:00.0 > supports-statistics: yes > supports-test: no > supports-eeprom-access: no > supports-register-dump: no > supports-priv-flags: no > > With guest driver that supports STANDBY, we would always have: > > [root@localhost ~]# ip a > 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group > defau0 > link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 > inet 127.0.0.1/8 scope host lo > valid_lft forever preferred_lft forever > inet6 ::1/128 scope host > valid_lft forever preferred_lft forever > 2: enp4s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state > UP gr0 > link/ether 26:28:c5:7f:14:24 brd ff:ff:ff:ff:ff:ff > inet 192.168.20.2/24 brd 192.168.20.255 scope global noprefixroute > enp4s0 > valid_lft forever preferred_lft forever > inet6 fe80::2428:c5ff:fe7f:1424/64 scope link > valid_lft forever preferred_lft forever > 3: enp4s0nsby: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel > master0 > link/ether 26:28:c5:7f:14:24 brd ff:ff:ff:ff:ff:ff > 4: enp2s0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq master > enp4s0 st0 > link/ether 26:28:c5:7f:14:24 brd ff:ff:ff:ff:ff:ff > [root@localhost ~]# ethtool -i enp4s0 > driver: net_failover > version: 0.1 > firmware-version: > expansion-rom-version: > bus-info: > supports-statistics: no > supports-test: no > supports-eeprom-access: no > supports-register-dump: no > supports-priv-flags: no > [root@localhost ~]# ethtool -i enp4s0nsby > driver: virtio_net > version: 1.0.0 > firmware-version: > expansion-rom-version: > bus-info: 0000:04:00.0 > supports-statistics: yes > supports-test: no > supports-eeprom-access: no > supports-register-dump: no > supports-priv-flags: no > [root@localhost ~]# ethtool -i enp2s0 > driver: iavf > version: 4.18.0-310.el8.x86_64 > firmware-version: N/A > expansion-rom-version: > bus-info: 0000:02:00.0 > supports-statistics: yes > supports-test: no > supports-eeprom-access: no > supports-register-dump: no > supports-priv-flags: yes > > Laurent Vivier (1): > virtio: failover: define the default device to use in case of error > > include/hw/virtio/virtio-net.h | 1 + > hw/net/virtio-net.c | 49 +++++++++++++++++++++++++++++----- > 2 files changed, 44 insertions(+), 6 deletions(-) > > -- > 2.31.1 > >
