User timeouts on 10.20

[Eric Malzahn]

Hello,

I've got a D360 running 10.20 connected to a Cisco switch. The switch connects to redundant cisco routers (HSRP). 1 of these routers can see the D360's IP address with no problem. The other from time to time will lose the connection. Users that are primary to this router will also lose connectivity. There are several other servers (Sun, HP 10.20 and 11, NT) that have no problem. We have the latest HP patches for 100BT card. My network team is running out of ideas. Could this possibly be a 10.20 patch or other issue that I have overlooked?
Thanks,
Eric Malzahn

[Ceesjan van Hattum]

Hi,
Since i've recently found out that by "ndd" IP periodically tests if the gateways are available by probe active gw's AND the dead ones, it might be that in your case the probe is of and therefor loses it's connection.

Check its probe status:
ndd -get /dev/ip ip_ire_gw_probe
If 0, then no check
If 1, then checks are on.

In case this does the trick, look at the settings in nddconf:
nddconf entry example:
TRANSPORT_NAME[0]=ip
NDD_NAME[0]=ip_ire_gw_probe
NDD_VALUE[0]=1

[Ron Kinner]

Sounds like a misuse of HSRP. The S in HSRP stands for Standby not Sharing.

With HSRP only one router is active on the LAN at one time. The other is just sitting there listening to hello packets from the active router to make sure that the prime router hasn't died. He is not really supposed to do any talking.

Your hosts are pointed at a virtual router with an ip different from either router A or B. Only the active router is supposed to respond to traffic to that address. The standby one is just supposed to sit there and do nothing. Both routers are supposed to be able to get to any possible destination so the fact that the standby router doesn't want to talk should not bother anyone.

Your having users on the standby router is weird. You might want to look at doing a little reorganizing or at least add a link between the two routers.

OK enough lecture. Why do some PCs work and others not?

Could be the way they get their default route.
Proxy ARP
RIP
IRDP
OSPF


Take a look at
netstat -rn
on the ones that work and the ones that don't. If it doesn't have a Default route pointing to the virtual router's ip address then it uses proxy ARP. If it points to the real IP address of a router then it's not setup to make use of HSRP.

Ron

[Johan Gjestvang]

Hi, you can't do ndd on 10.20

10.20 do not have a gw probe
mechanism like 11.0



Johan Gjestvang

[Eric Malzahn]

I should clarify:
The routers are using HSRP and are attached but we have some VLAN's active on 1 router and some on the other router. Users in VLAN's on router 2 have issues getting to the server. Users in VLAN's on router 1 have no issues. Also, some users on router 2 have no problems at all getting to the server. I for instance, am on router 2 and I have no problems. A coworker also in the same VLAN will timeout occasionally. I looked at netstat -rn and both her and I have identical configurations.

We were able to workaround the issue by making VLANS with heavy usage of this server active router 1.

Router 2 continues to have the same issues, and it is only with this 1 HP 10.20 box. We will do a full patch analysis on it next week and we may upgrade IOS on both routers as well.

Any additional suggestions would be appreciated.

Thanks,
Eric