Jet Direct Boxes stop responding

[mikecel79]

We've had this problem for quite some time now but never seem to be able to find a fix for it. We have about 20-30 HP JetDirect boxes that unless data is sent to them almost constantly they will not respond to any IP traffic. Once they stop responding you have to power the unit off and then back on to get ti to respond.

For now we've taken to running a batch file that runs constantly and pings them all the time. Not the best solution but it works for now. I'd like to find out why these stop responding. It isn't limited to just JetDirect boxes either. We have a few time clocks that use IP and have the same problem.

The network is setup like this. The main switch is a Cat5513 with a Supervisor II card. It has a 100mbit fiber blade that goes out to each Cat1900 (8 of them). Most of the devices that stop responding are on the 1900 switches. The Cat5513 has 2 VLANs on it. The 1900s are on their own VLAN.

Any ideas where to start looking?

[glen.grant]

Well this almost sounds like a problem we used to have with some printers ,what we did to help the situation is to set the mac address aging time to 1 week instead of 5 minutes , for most cases this fixed the problem , don't know why specifically unless the printers had trouble with arp requests . We only saw this problem on 2924's and 1924's .

[mikecel79]

Right after I posted this I found that setting on the Cat1900 switch. I changed it to just over 1 week. I'll be monitoring it to see if that solves the problem. It makes sense though since the JetDirect boxes don't initiate any kind of IP session, they only receive. If nothing is sent in 5 minutes wouldn't the switch remove the address from it's table?

[konigl · Accepted solution]

You are correct: under normal aging-time, five minutes of inactivity by your JetDirect would cause its dynamically-learned MAC address to be purged from the mac-address-table.

Extending the aging-time is one way to tackle the problem; but in theory the address could still be dropped eventually if no one uses the printer for a really really long time.

Another way to attack the issue is to note the JetDirect's MAC address, and create a static mac-address-table entry on the switch it plugs into. The upside to this is, the entry will never time out. The downside to this is, if you or someone else moves your JetDirect to a different switch port. So if you do this, put a description on the port as well, so you know what's supposed to be plugged in there; and document it externally, so people know not to move the cable.

If you choose the "static MAC entry" approach, don't forget to save your running-config else you will lose the static entry on next reboot.

Hope this helps.

[mikecel79]

Thanks for the info. Stuff gets moved around in that area pretty frequently so using static address won't work too well for us.

One last question. These are 1900 switches. Do they need to be restarted for this to take effect?

[konigl]

The CLI command

mac-address-table aging-time 86400

sets the length of time the switch keeps dynamic MAC addresses in memory before discarding to 86,400 seconds, or one day. Valid range is from 10 to 1,000,000 seconds (a little over 11.5 days).

The command takes effect instantly. You can verify this by entering the CLI command

show mac-address-table aging-time

Any configuration changes are saved to NVRAM automatically, within about a minute after having been executed successfully.

Hope this helps.

[ralphbender]

I have experienced similar problems with some of our JetDirect boxes. Our scenario is 1900 hanging off Remote 2500/4000 routers. The remote routers then hanging off Distribution routers. I have found a subtle configuration difference which is not always displayed in the Command Line Interface's config file (show run). When you go to Menu driven console and press [S] System, then [P] Network port, I discovered that switches with a setting under network port gave me abovementioned problem, but when I set to [N]one, the problem was rectified.