Bug 1891437
Summary: | Reboot of host with ovs interface lefts the interface in activating state | ||||||||
---|---|---|---|---|---|---|---|---|---|
Product: | Red Hat Enterprise Linux 8 | Reporter: | Ales Musil <amusil> | ||||||
Component: | NetworkManager | Assignee: | NetworkManager Development Team <nm-team> | ||||||
Status: | CLOSED DUPLICATE | QA Contact: | Vladimir Benes <vbenes> | ||||||
Severity: | high | Docs Contact: | |||||||
Priority: | unspecified | ||||||||
Version: | 8.3 | CC: | acardace, atragler, bgalvani, dholler, fge, lrintel, mburman, mtessun, peljasz, rkhan, sukulkar, thaller, till, vbenes, ymankad | ||||||
Target Milestone: | rc | Keywords: | Reopened, Triaged | ||||||
Target Release: | 8.0 | ||||||||
Hardware: | Unspecified | ||||||||
OS: | Unspecified | ||||||||
Whiteboard: | |||||||||
Fixed In Version: | NetworkManager-1.30.0-2.el8 | Doc Type: | If docs needed, set a value | ||||||
Doc Text: | Story Points: | --- | |||||||
Clone Of: | Environment: | ||||||||
Last Closed: | 2021-03-04 10:43:58 UTC | Type: | Bug | ||||||
Regression: | --- | Mount Type: | --- | ||||||
Documentation: | --- | CRM: | |||||||
Verified Versions: | Category: | --- | |||||||
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||||
Cloudforms Team: | --- | Target Upstream Version: | |||||||
Embargoed: | |||||||||
Bug Depends On: | |||||||||
Bug Blocks: | 1920330 | ||||||||
Attachments: |
|
Created attachment 1724128 [details]
nmcli commands
Hi, I'm having similar if not identical problem. I have both(and probably other relevant components pulled in) NetworkManager and openvswitch2.11 from ovirt-4.4-copr and cannot confirm if the problem is exclusive to ovirt-4. I use NM to set up OVS bridges and have a pretty straightforward setup I believe. Each time after system reboot one (always the same) interface fails to start and stays in a weird state, actually it gets duplicated. ovs0-int-10.3.3 d9c9268f-408c-4035-9684-97e4eaf92e18 ovs-interface ovs0-int33 (colored yellow) ovs0-int-10.3.3 d9c9268f-408c-4035-9684-97e4eaf92e18 ovs-interface ovs0-int33 (colored red) whereas the rest of the bridge: ovs0-int-10.1.1 be5e79e7-2e4e-47ed-97d0-270c76e43b9a ovs-interface ovs0-int11 ovs0 9cf72215-eebc-4f7c-9466-b697f09e1bda ovs-bridge ovsbr0 ovs0-port-11 1577414e-7eeb-41bc-a66f-7b7b596285d0 ovs-port ovs0-port11 ovs0-port-33 5465d3c5-9a0c-4f78-89c0-de94b0294d0e ovs-port ovs0-port33 ovs0-port9 b9c2b7cd-897e-4726-8fc7-e2f7c7aab067 ovs-port ovs0-port9 ovs0-port9-physical 14fb0b0e-f7bb-4248-9295-34f561058f4d ethernet enp7s0f1np1 ovsbr0-libvirt0-int f217cd6c-976b-4002-b191-9f15271602cd ovs-interface ovsbr0 ovsbr0-libvirt0-port 7c2d30bd-0019-4cdb-af05-3df666005261 ovs-port ovsbr0 And then only another reboot with either: a) deletion of the interface prior to reboot and after the reboot creation anew b) removal of /etc/openvswitch/conf.db prior to reboot. When all is good and whole bridge is up & running then bridge looks like this: 004e89ac-4c45-4a33-9347-c1d343889ceb Bridge "ovsbr0" Port "ovs0-port11" tag: 11 Interface "ovs0-int11" type: internal Port "ovsbr0" Interface "ovsbr0" type: internal Port "ovs0-port9" Interface "enp9s0f3" type: system Port "vnet0" tag: 33 Interface "vnet0" Port "ovs0-port33" tag: 33 Interface "ovs0-int33" type: internal ovs_version: "2.11.0" Journal does not tells much, at least not by default .... device (ovs0-int33): Activation: starting connection 'ovs0-int-10.3.3' (d9c9268f-408c-4035-9684-97e4eaf92e18) .... When I try to 'c up' manually nmcli just "time outs". nmcli shows: ... GENERAL.STATE: activating ... and also: ... GENERAL.STATE: deactivating ... at the same time as the interface is duplicated so are GENERAL entries. I'm on Centos8(no Stream) and with kernel-ml. It feels (I have no lab where I could fiddle more) that this should reproduce. I have a second system with slightly different hardware and this same problem occurs. Feel free to ask for more info/logs. many thanks, L. A workaround which is bit less invasive then a reboot: 1) [root@dzien ~]$ rm -f /etc/openvswitch/conf.db 2) [root@dzien ~]$ systemctl restart openvswitch.service 3) [root@dzien ~]$ systemctl restart NetworkManager 4) [root@dzien ~]$ { _IF=ovs0-port9-physical ; nmcli c d $_IF; sleep 2; nmcli c u $_IF; } # seems that (at least in my case) the physical iface needs such a "reset" to make network able to access it again Hi, this looks related to bug 1861296. During a reboot, NM is stopped and tries to delete the interfaces that were added to the ovsdb. However, it fails to do that, and therefore some ovs interfaces are present in the ovsdb at the next boot, conflicting with the configuration that NM wants to do. Hi Ales, please retest with NetworkManager-1.30.0-0.8.el8 once it's available in the repositories. Ales, could you try reproducing the issue with the build mentioned by Beniamino in the above comment. Thanks! Hi, I tested this change and it seems like the issue is gone. Thanks Closing per above comment. closing as duplicate *** This bug has been marked as a duplicate of bug 1861296 *** |
Created attachment 1724127 [details] Trace log Description of problem: After configuring multiple ovs interfaces on single host, the host reboot wi8ll cause those interface to be stuck in activating state. Version-Release number of selected component (if applicable): NetworkManager-1.26 How reproducible: 100% Steps to Reproduce: 1. Create multiple ovs interface that are connected by bridge to physical interface 2. Reboot the host 3. Actual results: The ovs interfaces are stuck in activating state Expected results: The ovs interfaces should be up and running Additional info: