[j-nsp] SRX5k problem

[OBrien, Will]

The fabric carries traffic between the nodes, so it's my immediate suspect on the traffic loss.

Are your connections configured as standard Reth interfaces? Are you using some form of igp?
In active/active mode, I've seen some traffic loss, but most of it was due to ospf taking time to select the new pathway.




On Sep 3, 2013, at 9:54 AM, R S wrote:

we have a remote cluster, hence no direct connection is provided but a switched connection (through L2 infrstructure - a vlan basically)
yes dual REs
why do you suppose a problem on fabric links ?




> From: ObrienH at missouri.edu<mailto:ObrienH at missouri.edu>
> To: dim0sal at hotmail.com<mailto:dim0sal at hotmail.com>
> CC: juniper-nsp at puck.nether.net<mailto:juniper-nsp at puck.nether.net>
> Subject: Re: [j-nsp] SRX5k problem
> Date: Tue, 3 Sep 2013 14:19:35 +0000
>
> Failover works fine on my 5800 cluster. I use direct connections for fabric and control.
> It sounds like you're losing traffic in Zmode. I'd start by taking a serious look at your fabric links.
> Do you have dual REs in each chassis for the double control links?
>
>
> On Sep 3, 2013, at 7:34 AM, R S wrote:
>
> > We are having a geographic SRX5800 chassis cluster with two redundancy-groups (RG0 and RG1).
> >
> > We are having many issues while performing the failover of the RG1, with and without preemption. The issue is that when the RG1 becomes active on the node1, real traffic is being lost. Also with preemption, when the node0 becomes again the active for the RG1, lot of traffic is lost.
> >
> > There are dual Ctrl and Fabric link both through a Layer2 infrastructure, one couple through MX960 and one couple through EX8200.
> >
> > Any similar nightmare ?
> >
> > Tks
> > _______________________________________________
> > juniper-nsp mailing list juniper-nsp at puck.nether.net<mailto:juniper-nsp at puck.nether.net>
> > https://puck.nether.net/mailman/listinfo/juniper-nsp
>