Mailing List Archive: nsp: foundry

FESX ARP problem

Index | Next | Previous | View Threaded

rsm at fast-serv

Apr 15, 2011, 8:29 AM

Post #1 of 3 (489 views)
Permalink

FESX ARP problem

I've been scratching my head all morning on some arp issues on an FESX (prem)
that routes a handful of VE's.

The scenario is that when an ARP entry expires, only certain source IPs (on
the WAN side) can trigger an ARP request and subsequent re-entry into the ARP
table. There's no pattern for the source IP...just that some trigger ARP and
others don't.

So for example, Joe at 1.2.3.4 cannot reach his box, but as soon as Bob at
pings the machine, Joe can reach his box again.

Current table size fluctuates between 580 and 620 entries with a system-max of
4000 and there are no static entries.

#sh arp | i entries
Total number of ARP entries: 591

#sh default values | i arp
ip arp age:10 min bootp relay max hops:4 ip ttl:64 hops
ip-arp 4000 64000 4000
ip-static-arp 512 1024 512

Any ideas?

~Randy

_______________________________________________
foundry-nsp mailing list
foundry-nsp [at] puck
http://puck.nether.net/mailman/listinfo/foundry-nsp

andreas at larsen

Apr 16, 2011, 8:49 AM

Post #2 of 3 (464 views)
Permalink

Re: FESX ARP problem [In reply to]

I think I have seen this on some older code of the FESX 424 we where running
below 4.0. What version are you running ?

// Andreas

On Fri, Apr 15, 2011 at 5:29 PM, Randy McAnally <rsm [at] fast-serv> wrote:

> I've been scratching my head all morning on some arp issues on an FESX
> (prem)
> that routes a handful of VE's.
>
> The scenario is that when an ARP entry expires, only certain source IPs (on
> the WAN side) can trigger an ARP request and subsequent re-entry into the
> ARP
> table. There's no pattern for the source IP...just that some trigger ARP
> and
> others don't.
>
> So for example, Joe at 1.2.3.4 cannot reach his box, but as soon as Bob at
> pings the machine, Joe can reach his box again.
>
> Current table size fluctuates between 580 and 620 entries with a system-max
> of
> 4000 and there are no static entries.
>
> #sh arp | i entries
> Total number of ARP entries: 591
>
> #sh default values | i arp
> ip arp age:10 min bootp relay max hops:4 ip ttl:64 hops
> ip-arp 4000 64000 4000
> ip-static-arp 512 1024 512
>
> Any ideas?
>
> ~Randy
>
> _______________________________________________
> foundry-nsp mailing list
> foundry-nsp [at] puck
> http://puck.nether.net/mailman/listinfo/foundry-nsp
>

rsm at fast-serv

Apr 16, 2011, 9:50 AM

Post #3 of 3 (464 views)
Permalink

Re: FESX ARP problem [In reply to]

On Sat, 16 Apr 2011 17:49:28 +0200, Andreas Larsen wrote
> I think I have seen this on some older code of the FESX 424 we where running below 4.0. What version are you running ?
>
> // Andreas

4.0.0 ... it ran fine for 3 years until about 2 weeks ago; 1020 days uptime as of now.

I also started noticing MAC's dropping off the arp table long before they expire, and even with static ARP entries traffic still doesn't always pass. And it seems to be singling out only a couple out of several dozen VE's. The only thing that stays stable is downing the VE and adding a static route to the VRRP neighbor, I'm about to just pull it and let everything fail over.

I'm guessing CAM corruption, hopefully not faulty hardware.

~Randy

Index | Next | Previous | View Threaded

Interested in having your list archived? Contact Gossamer Threads