On Fri, May 4, 2012 at 12:27 PM, Kapetanakis Giannis
<[email protected]> wrote:
> On 03/05/12 22:56, mxb wrote:
>>
>> I'd suggest you to experiment with the src and try to rollback if_pfsync.c
>> to,
>> say rev. 1.179,
>> then roll forward with revisions until you can pinpoint one which breaks
>> it.
>>
>> //maxim
>
>
> Ok, I've traced the problem back to r1.180 of if_pfsync.c
>
> If I remove 1.180 and apply 1.181, 1.182, 1.183, 1.184
> then everything works fine.
>
> According to my understanding 1.180 introduces the following two problems:
>
> 1) When one firewall reboots, the other one is initiating a bulk transfer
> thus demoting his carp/pfsync groups.
> I don't understand why this is needed.
>
> 2) When the firewall boots it sets his demotion counter on groups
> carp/pfsync to 0 (instead of 1) before his bulk transfer
> is finished.
>
> The diff bellow removes 1.180
> Feel free to send me diffs to try.
>
> regards,
>
> Giannis
>

hi,

yes, there's no doubt that this is a problem.

btw, can you try the unpatched kernel and put a switch between
the firewalls, i.e. connect pfsync interfaces through the switch.
will that work for you?

Reply via email to