Description of problem: There is a "default" logical network on RHV Host, which is reported by: virsh net-list It is the result of an updating rhn-virtualization-host, which has dependency with libvirt and libvirt-daemon-config-network Version-Release number of selected component (if applicable): RHV 4.1.9 from: redhat-virtualization-host-image-update-placeholder-4.1-6.0.el7.noarch to redhat-virtualization-host-image-update-placeholder-4.1-9.1.el7.noarch How reproducible: every time Steps to Reproduce: 1. yum update 2. yum install rhn-virtualization-host # using version 4.1.6) 3. check using: virsh -r net-list Actual results: [root@host:~/home]# virsh -r net-list Name State Autostart Persistent ---------------------------------------------------------- ;vdsmdummy; active no no default active yes yes vdsm-rhevm active yes yes Expected results: There is no default network. Additional info: it can be easily removed using: https://access.redhat.com/solutions/27195
The problem here is not with the RHVH itself, but that installing rhn-virtualization-host from Satellite appears to pull in versions of system packages which are out of step with the RHVH image. The 'default' network is being created by the %post scripts there. Updating versions of any package shipped in the RHVH image outside of tested hotfix packages in the optional channel is not strictly a supported use case. A better resolution to this would be to ship rhn-virtualization-host in the RHVH optional channel, so we can ensure these match. Or... Can Satellite break the dependency on libvirt? It doesn't seem to be necessary, which was confirmed by mskrivanek's comments in the other open bug. This is the root cause of the problem. For now, '--nodeps' is even a better suggestion than a chance in RHVH code, until we can determine whether or not this dependency can be broken. We may be able to find a workaround in RHVH, but it will not be nice, since there is no resident RHVH code running which can pick this up on a running system, unless we added a systemd timer just for this use case.
Test Version: RHV 4.1.9 from: redhat-virtualization-host-4.1-20170914.1.x86_64.liveimg.squashfs to redhat-virtualization-host-image-update-4.1-20180126.0.el7_4.noarch.rpm Test Steps: 1. Clean install redhat-virtualization-host-4.1-20170914.1.x86_64.liveimg.squashfs with kickstart file. 2. yum install redhat-virtualization-host-image-update-4.1-20180126.0.el7_4.noarch.rpm to upgrade rhvh to 4.1.9 version 3. # virsh -r net-list Name State Autostart Persistent ---------------------------------------------------------- ;vdsmdummy; active no no 4. Yum install rhn-virtualization-host-5.4.54-3.el7sat.noarch.rpm, and install the dependency packages: 1) rhn-virtualization-common-5.4.54-3.el7sat.noarch.rpm 2) libvirt-3.2.0-14.el7_4.7.x86_64.rpm 3) libvirt-daemon-driver-lxc-3.2.0-14.el7_4.7.x86_64.rpm 4) libvirt-daemon-config-network-3.2.0-14.el7_4.7.x86_64.rpm 5. # virsh -r net-list Name State Autostart Persistent ---------------------------------------------------------- ;vdsmdummy; active no no default active yes yes Result: QE can reproduce this bug, so qe_ack
Not identified as a blocker for 4.2.7, moving to 4.2.8
rhn-virtualization requires "libvirt", not a specific version, any idea where it pulls this libvirt from ? it does not exist in the rhvh channel - what channels do you have enabled ? If this is indeed a bug, the only way I could think of fixing it is by adding a %triggerin to redhat-virtualization-host that deletes the default network, but I think it's risky.
Closing as insufficient data, if you can provide the needed info please feel free to reopen