virtcontainers: Add vhost-user-net hotplug support#611
virtcontainers: Add vhost-user-net hotplug support#611sboeuf wants to merge 2 commits intokata-containers:masterfrom
Conversation
|
/cc @egernst |
|
PSS Measurement: Memory inside container: |
|
Build failed (third-party-check pipeline) integration testing with
|
|
CI is not happy |
3c92a3c to
9e2979f
Compare
|
@devimc should be fixed now! |
|
Build failed (third-party-check pipeline) integration testing with
|
|
Build failed (third-party-check pipeline) integration testing with
|
|
PSS Measurement: Memory inside container: |
Codecov Report
@@ Coverage Diff @@
## master #611 +/- ##
==========================================
- Coverage 64.69% 64.55% -0.15%
==========================================
Files 84 84
Lines 9558 9594 +36
==========================================
+ Hits 6184 6193 +9
- Misses 2726 2752 +26
- Partials 648 649 +1 |
|
Depends on kata-containers/govmm#39 |
The update of the govmm vendoring forced some changes in the code virtcontainers/qemu.go, related to the new flag shared-rw. In order to be able to share an image between several VMs, Qemu version 2.10 and newest need this flag to be enabled, otherwise the image is locked to be exclusively used by a single VM. Shortlog: 21504d3 qemu/qmp: Add netdev_add with chardev support ed34f61 Add some negative test cases for qmp.go 17cacc7 Add negative test cases for qemu.go 2706a07 qemu: Use the supplied context.Context for launching e46092e qemu: Do not try and generate invalid RTC parameters fcaf61d qemu/qmp: add vfio mediated device support 4461c45 disk: Add --share-rw option for hotplugging disks 6851999 qemu/qmp: add addr and bus to hotplug vsock devices Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
In order to fully move to network hotplug, we first need to support vhost-user-net hotplug. This commit relies on appropriate QMP commands to hotplug a vhost user network by first creating a new char device pointing to the socket path, by creating the appropriate network device, and finally by adding a virtio-net-pci device. Fixes kata-containers#610 Signed-off-by: Sebastien Boeuf <sebastien.boeuf@intel.com>
|
PSS Measurement: Memory inside container: |
|
@amshinde PTAL |
|
Build failed (third-party-check pipeline) integration testing with
|
| @@ -338,7 +338,24 @@ func (endpoint *VhostUserEndpoint) Detach(netNsCreated bool, netNsPath string) e | |||
|
|
|||
| // HotAttach for vhostuser endpoint not supported yet | |||
|
Need #758 to be fixed first. |
|
@raravena80 nope sorry, I'll work on it when I get some time. |
|
@sboeuf any updates? |
|
No, but this needs some time to be spent if we want it to be properly integrated with the device manager model. |
|
@sboeuf ping, any updates? (From your weekly Kata herder) |
|
Check 🏁 |
|
@caoruidong @sboeuf ping, from your weekly Kata herder. |
|
@caoruidong @sboeuf is this ready for merging? |
|
@sboeuf - could you update the branch to resolve the conflict? |
|
Ping @sboeuf. |
|
ping @sboeuf |
|
@sboeuf - shall we close this one? |
|
@jodh-intel yes let's close it and get back later to it if needed. |
Make NoPivotRoot depend on the actual FS type of `/` being rootfs, instead of agent being the init process. In this way, an initrd built without AGENT_INIT can still be used to run containers. Fixes: kata-containers#611 Signed-off-by: Marco Vedovati <mvedovati@suse.com>
In order to fully move to network hotplug, we first need to support
vhost-user-net hotplug. This commit relies on appropriate QMP commands
to hotplug a vhost user network by first creating a new char device
pointing to the socket path, by creating the appropriate network
device, and finally by adding a virtio-net-pci device.
Fixes #610
Signed-off-by: Sebastien Boeuf sebastien.boeuf@intel.com