[NETVIRT-463] L2GW Connection Utils does not see connected HWVTEP device Created: 06/Feb/17  Updated: 19/Oct/17  Resolved: 04/Apr/17

Status: Resolved
Project: netvirt
Component/s: General
Affects Version/s: Boron
Fix Version/s: None

Type: Bug
Reporter: Victor Pickard Assignee: Unassigned
Resolution: Duplicate Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified
Environment:

Operating System: All
Platform: All


Issue Links:
Duplicate
duplicates NETVIRT-474 L2GW: Null Pointer Exceptions in Phys... Resolved
External issue ID: 7744

 Description   

I have a HWVTEP emulator connected to ODL. When the HWVTEP manager is set to point to the ODL Controller, both controller and HWVTEP see the node as "connected". Configuring L2GW and L2GW connections works, as long as the set-manager config is performed on the HWVTEP AFTER ODL is up and running.

In this same setup, if I unstack/stack (using devstack), and leave the HWVTEP manager config in place, the connection between the HWVTEP and OVSDB on the control node stays connected (via vtep-ctl show, and dumping MDSAL OVSDB connection info for hwvtep dev).

Now, if I configure a L2GW and L2GW connection, the config will not be pushed from the ODL control node to the HWVTEP device. Debugging in the logs shows that L2GW Connection utils doesn't see that the HWVTEP device is connected, even though the HWVTEP DB shows connected, and OVSDB SB MDSAL shows connected.

I don't have the log handy at the moment, will add it to this bug next time I see it.

Suneelu has observed the same behavior. The thinking is that there is possibly some timing issue between OVSDB Connection manager and OVSDB clients.

The workaround for this bug, is to del-manager on HWVTEP device, wait about 30-45 secs, then do a set-manager on HWVTEP device. At this point, L2GW and L2GW Connection config will be sent from ODL to HWVTEP device.



 Comments   
Comment by Sam Hague [ 03/Apr/17 ]

vic, Suneelu, is this still an issue?

Comment by suneel verma [ 04/Apr/17 ]

This should be fixed after the patch.
https://git.opendaylight.org/gerrit/#/c/53337/

Comment by Victor Pickard [ 04/Apr/17 ]

(In reply to Sam Hague from comment #1)
> vic, Suneelu, is this still an issue?

I have been testing with patch https://git.opendaylight.org/gerrit/#/c/53555, and I no longer see this issue.

What I saw during my testing (without above patch) is that multiple updates would arrive from hwvtep device, resulting in callback(s) with multiple updates.
During processing of the updates, an unhandled exception would occur (NPE), and any other updates would not be processed (Physical_Switch update, in case of connection update, resulting in L2GW cache not being updated).

Generated at Wed Feb 07 20:21:37 UTC 2024 using Jira 8.20.10#820010-sha1:ace47f9899e9ee25d7157d59aa17ab06aee30d3d.