Node provisioning fails consistently about kubevirtci

[rmohr]

Oh, I guess /dev/kvm is not there. The environments are meant to provide VMs on bare metal (developer machine or CI servers).

For as long as you do not plan to run VMs inside the started environments(3 levels of nesting is too much) you could also use them on VMs which have nested virtualization enabled.

[copejon]

@rmohr Thanks, that's pointed in what looks like the right direction. I've installed kvm. Also, centos7.5 is running in VMware on my OS X host w/ nested virtualization enabled.

I'm getting a new error now related to iptables. I'ved tried iptables -F as well as running with firewalld started and stopped. The result is the same -

[root@centos ~]# gocli run --random-ports --nodes 3 --background kubevirtci/k8s-1.10.3
{"status":"Status: Image is up to date for docker.io/kubevirtci/k8s-1.10.3:latest"}Error response from daemon: driver failed programming external connectivity on endpoint kubevirt-dnsmasq (2c6d779f54ed159c3ffe288156c916d448d2c537f5dd2c3bd5dbc82dd4080e42):  (iptables failed: iptables --wait -t nat -A DOCKER -p tcp -d 0/0 --dport 32847 -j DNAT --to-destination 172.17.0.2:8443 ! -i docker0: iptables: No chain/target/match by that name.
 (exit status 1))

[rmohr]

This one is indeed a docker/firewall issue. I think that if you start the current versions for rhel/centos in the wrong order firewalld overwrites some docker iptable rules. Restarting docker should do the trick.

[copejon]

Good news - that fixed the error
Bad news - another error. Some googling seems to indicate this is an error specific to using qemu/kvm on VMware machines, which I'm doing.

If you agree with the above, then I'll close the issue as it wouldn't relate to kubevirtci anymore.

[root@centos kubevirtci]# docker logs kubevirt-node01
3: tap01: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast master br0 state DOWN mode DEFAULT group default qlen 1000
    link/ether d2:12:b1:a3:5a:85 brd ff:ff:ff:ff:ff:ff
Creating disk "/var/run/disk/disk.qcow2 backed by /disk01.qcow2".
Formatting '/var/run/disk/disk.qcow2', fmt=qcow2 size=42949672960 backing_file=/disk01.qcow2 cluster_size=65536 lazy_refcounts=off refcount_bits=16
SSH will be available on container port 2201.
VNC will be available on container port 5901.
VM MAC in the guest network will be 52:55:00:d1:55:01
VM IP in the guest network will be 192.168.66.101
VM hostname will be node01
qemu-system-x86_64: error: failed to set MSR 0x38d to 0x0
qemu-system-x86_64: /builddir/build/BUILD/qemu-2.10.1/target/i386/kvm.c:1788: kvm_put_msrs: Assertion `ret == cpu->kvm_msr_buf->nmsrs' failed.

[rmohr]

If you agree with the above, then I'll close the issue as it wouldn't relate to kubevirtci anymore.

Works for me. If you find solution to fix this /dev/kvm thing, let us know there is something we can change to make this work.