|
>
> I think that this approach seems quite sound. Have you tested it to see
> how it behaves?
Why do toy think it is sound. It is just more aggressive in connection
termination on the backup. I have tested it, and it behaves as expected.
In the worst case two more syncs will be sent however with direct
routing and ipip tunnel all the connections die with FIN_WAIT timeout
no-matter master or backup. I even sniffed the traffic no anomalies.
I will prepare it with comment and sign off.
What comes to the first patch I will fix the lock problem and send it
again.
Rumen
> For reference, I have attached my version of your two patches,
> which I have rediffed a little. Could you provide an appropriate
> comment and signoff for the second patch once you are happy with it?
>
|
