[Xorp-hackers] xorp svn VRRP centos

[Ben Greear]

On 02/26/2010 11:41 AM, Eric S. Johnson wrote:
>
>
>
> greearb at candelatech.com said:
>> Can you try the attached patch and see if it works better?
>> This compile tested only.
>> I'd also like to see the error logs from the run with this patch if
>> that's convenient.
>
>
> Hey. Nope. Failed again. here is what seems to be relevant from the logs.
> It doesn't make sense to me.
>
> Im gonna hack on this a bit now and see if I can figure out what is happening.
>
>
> [ 2010/02/26 14:24:41  ERROR xorp_fea:26641 FEA +684 fea/data_plane/ifconfig/ifconfig_set_netlink_socket.cc set_interface_status ] WARNING:  Settting interface status using netlink failed on: eth0.21 but ioctl method worked.  Will use ioctl method from now on.
> [ 2010/02/26 14:24:41  ERROR xorp_fea:26641 FEA +426 fea/data_plane/ifconfig/ifconfig_set.cc push_interface_begin ] Interface error on eth0.21: WARNING:  Settting interface status using netlink failed on: eth0.21 but ioctl method worked.  Will use ioctl method from now on.
> [ 2010/02/26 14:24:41  FATAL xorp_fea:26641 MFEA +538 fea/mfea_node.cc vif_update ] Assertion (vif_index != Vif::VIF_INDEX_INVALID) failed

Hard to see how my change would have created that assert unless it's a race.  I'd add more debugging around that vif_update() method logic
to see why the fatal assert hits.

I also modified fea & interface logic a lot in my tree, (though not necessarily fixing
or making this particular problem worse).

If there's an easy way to reproduce this, maybe let me know your config & commands?

Thanks,
Ben

-- 
Ben Greear <greearb at candelatech.com>
Candela Technologies Inc  http://www.candelatech.com