/azp run

@shi-su i see vnet orch test case 7/8/9/11 are failing in the PR job. Can you please check ? I ran job again and it is still failing. Looks to be consistent failure.

cc @prsunny

@shi-su i see vnet orch test case 7/8/9/11 are failing in the PR job. Can you please check ? I ran job again and it is still failing. Looks to be consistent failure.

cc @prsunny

Ack checking

@shi-su i see vnet orch test case 7/8/9/11 are failing in the PR job. Can you please check ? I ran job again and it is still failing. Looks to be consistent failure.

cc @prsunny

This is a very weird failure. It seems that the same nexthop group gets recreated when a new route is pointing to it. This is an unexpected behavior since the nexthop group should be reused. I did not find any clue why the same nexthop group could get created multiple times. Interestingly, this failure only seems to happen in sairedis pipeline vs test while it is not observed in swss pipeline. On the other hand, I do not think there is anything in sairedis that may affect the behavior. Still investigating.

I tried to overwrite swss deb package with the one built from swss pipeline. The failure does not happen anymore after doing so. This seems to indicate that the swss package built from swss and sairedis pipelines are somewhat different. Any idea about what might cause such a difference?

@shi-su i see vnet orch test case 7/8/9/11 are failing in the PR job. Can you please check ? I ran job again and it is still failing. Looks to be consistent failure.
cc @prsunny

This is a very weird failure. It seems that the same nexthop group gets recreated when a new route is pointing to it. This is an unexpected behavior since the nexthop group should be reused. I did not find any clue why the same nexthop group could get created multiple times. Interestingly, this failure only seems to happen in sairedis pipeline vs test while it is not observed in swss pipeline. On the other hand, I do not think there is anything in sairedis that may affect the behavior. Still investigating.

@shi-su may be it is not binary based of your code . Can we put breakpoint and see the code flow when it is failure.

Actually, passes and failures are both pretty consistent from my observation.

Actually, passes and failures are both pretty consistent from my observation.

The test cases constantly fail with the deb package from sairedis pipeline and they constantly pass with the package from swss pipeline..

I suspect the problem is because the variable weight here https://github.com/Azure/sonic-swss/blob/master/orchagent/nexthopkey.h#L22 is not initialized for nexthops in vnetorch. Created this PR sonic-net/sonic-swss#2096 and let's check if it resolves the problem.

/azp run

I suspect the problem is because the variable weight here https://github.com/Azure/sonic-swss/blob/master/orchagent/nexthopkey.h#L22 is not initialized for nexthops in vnetorch. Created this PR Azure/sonic-swss#2096 and let's check if it resolves the problem.

@shi-su thanks the change worked..