We have confirmed that we do have the patch you mentioned.
Additionally, please note that the bug you mentioned describes a slightly different situation.
In the bug you mentioned the FIP was never migrated to the host where the VM resides.
This is *not* what we're seeing.
In our case the FIP *is* properly brought up in the host where the VM resides (as expected), but it is never removed from the snat namespace (not expected).
So, that does look like a slightly different bug.
For what it's worth, we have several hypervisors, most of them run with 'agent_mode=dvr', but two of those are running in 'agent_mode=dvr_snat'.
Is there anything specific we can check on our end to help with troubleshooting this?
Hi Miguel,
Thanks for your reply.
We have confirmed that we do have the patch you mentioned.
Additionally, please note that the bug you mentioned describes a slightly different situation.
In the bug you mentioned the FIP was never migrated to the host where the VM resides.
This is *not* what we're seeing.
In our case the FIP *is* properly brought up in the host where the VM resides (as expected), but it is never removed from the snat namespace (not expected).
So, that does look like a slightly different bug.
For what it's worth, we have several hypervisors, most of them run with 'agent_mode=dvr', but two of those are running in 'agent_ mode=dvr_ snat'.
Is there anything specific we can check on our end to help with troubleshooting this?
Thanks!