Ping Timeouts

[Bill Limberg]

All of our servers are plugged into a ProCurve 2824. I was noticing some disconnects between two of our servers, so I started continuous pings between servers to look into the issue. From these pings I'm seeing a lot of request timeouts. Normally there are atleast 30-40 in a 24 hour period. Right now I narrowed my testing between two of our new servers (HP 380 and 360 both less than 6 months old). I replaced the cables on both servers with new ones, but that didn't help.
Both servers are running at 1000FD. I haven't done any special configuration on the switches; it's all pretty much default at this time. I am running the latest 2824 software.
Port counters all look good. Zero errors/collisions/etc. The syslog information in PCM doesn't show anything out of the ordinary either.

At this point I'm not sure what else to look at. Any ideas would be appreciated.

[Preston Gallwas]

Bill,

If you console or telnet in and "show log" what do you see?

You say the config is pretty much default: what is the Firmware version?

We run 2 2848's for our servers - at one point our contact at HP suggested disabling LACP unless we needed it (conf t, no int all lacp)
and that has resolved a lot of 'port drop' issues for us

In addition, if possible it may be wise to hardcode the duplex (which may be difficult, becuase i believe the only 2 choices are "auto" and "auto-1000"...so this may not apply)

Do the NIC's on those servers have the latest drivers? What PLSP are you running? For netware, we've seen issues with the old Q57.LAN drivers causing all sorts of problems.

[Bill Limberg]

If I do a 'show log' i see the same messages that are on the syslog server. Nothing out of the ordinary.

The switch is running version I.08.71

I'll try disabling LACP tonight, it is not required.

I haven't tried setting duplex speeds yet. I was hoping to avoid that if possible.

NIC drivers were up to date when I first noticed this problem, but I see some updates out there that I can try as well. I'll try to install those tonight while I'm at it.

We do have one old Netware server here and I did consider that might be causing problems, so I moved it to a different switch. That didn' help however.

Thanks for the post. I'll make some changes and see how it looks after that.

[Preston Gallwas]

Incidentally, have you possibly rolled your NIC drivers back?

It sounds like if it is across thebaord it is likely LACP (in my experience at least)

[Bill Limberg]

I tested pinging on about a half dozen servers and see the issue with all of them. So I'm going to assume this is an issue on every port on the 2824.

[Preston Gallwas]

See if disabling LACP helps. In addition to that, you may try upgrading your firmware

[Bill Limberg]

Nothing has helped yet. No LACP. Manually set speed and duplex. New cables. Going to install newer printer drivers this weekend.

[Les Ligetfalvy]

Are these servers NIC teamed? Do you have ISL trunks? Are you running STP? You might want to run a network sniff and watch for STP irregularity.