L2 switch monitoring issues with IP pings

[Giovanni Trapasso_1]

Hello,

we are monitoring our L2 switches using GFI and pings. the gfi pings are set if the ip pings fail twice within a minute to mark the device as down. But when looking at the switch uptime everything looks good.

I am not so worried about this because these are L2 switches, hp 2810 and dell 5324, and the IP address is not used for anything except management and monitoring of the switchs. As stated the uptime does not register an outage, I have spanning tree running and it is stable, the web console does not register anything. I also setup a procurve manager server which is not showing any interface issues. CPU util on the switches are quite low. but gfi is alerting intermittanty on various switches in my network.

attached is the network diagram. My root stp switch is one layer up. the 2 top switches connect to the root switch via fiber. the gfi server connects to the 2 bottom middle set of switches using switch fault tolerant network teaming.

anyone see anything like this? any suggestions? other than Procurve manager any suggestions on a better switch Monitoring server?

[Pieter 't Hart]

look at attached diagram.
This could be the active spanning-tree topology.
your ping may need to pass 5 switches before reaching the destination.
look if you can increase the timeout of your ping in gfi.

anyway how does a "manual" ping respond?

about the server,
is it network team "fault tolerant" only or faul-tolerant with "transmit-load balancing"?
I don't think the last is supported over two switches.

[Giovanni Trapasso_1]

took a look at your diagram, and basically my network is layed out that way. The diagram is a screen shop from procurve manager, I guess it isn't that clear, but they are both the same. I guess I can connect the 2 top switches together to reduce 2 hops.

The manual pings from one of the switches the server is connected to I usually get 1 ms, but I did get a 25 ms response. From the server itself I get 1 ms to 15 ms reponses.

The fault tolerant setup is switch falut tolerant. 2 network interfaces, 2 different switches, one active and one standby.

[Pieter 't Hart]

our company uses WUG (same functionality)

important to know only one nic is active.
it's also good information the ping from the switch has stable response.

but with manual ping i meant from the same server that's running GFI.
This to check if any network connection has problems, like half/full duplex mismatch.

run a ping -t for some time
- to the first switch
- to a switch 2 hops away
- to a switch 4 hops away
etc.

see if this responds differently.
match any misses with the log of GFI.

[Giovanni Trapasso_1]

Thanks for the info.

I did the pings from the server with the -t option, this morning before e-mailing, and while looking into this issue for the past few weeks. pings are pretty consistant. switch and interface interface are set to 1 Gb full duplex, auto negotiation.

I setup a HP Procurve management server and it did not report any major issues. I will take another look at what's up gold.