NetworkNick.net

ProCurve has released a plethora of new product and features…here are the updated MIB’s for all you SNMP guru’s.

I attended the HPTF back in June, and didn’t get a chance to write about my favorite part of the week. Packet/protocol analysis is the last word for troubleshooting problems in our world. I use Wireshark (formerly Ethereal) every week. There is so much to know about the software, it’s use, and analysis features. No one I’ve met knows more about protocol analysis than Laura Chappell. She was the highlight of the HPTF, I attended 3 of her 2 hour sessions. If you’ve attended any of Laura’s protocol analysis training or seminars, I probably don’t need to tell you how awesome she is. If you don’t know who she is, hop over to :

Check this out. It’s a Perl script (hence cross platform) that not just pulls configs, but also archives and has the option to only pull them if they’ve changed. Check out the options:

Direct download.

The blokes over at the M$ NAP blog are talking about up and coming NAP plans, including 802.1x deployment scenarios. Check it out here. I find all of this interesting, as it coincides heavily with M$’s “lets play nice with TNC” push lately. All in the interests of going up against the C-word of course.

While onsite with a partner today we did a plugfest with a 5406 and 4506.

We used 802.1s, with multiple instances, and ran the 4506 and 5406 as the VRRP owner for the VLAN’s that were mapped to the instances that each switch was root for. We also did some basic OSPF interop, with redistribute static..

VRRP between the 2 works flawlessly. Fails over instantly. Absolutely a cinch to configure.

Here is the 4506 config and the 5406 config, running all standards based protocols for redundancy, load balancing, and dynamic routing.

Who said that we can’t just all get along?

‘Nuff said.

For all of you open source loving, FreeRADIUS users out there, here are some sample configs for both the server and switch. Also, FreeRADIUS is out for Windows (freeRADIUS.net) too!

I’ve been knee deep in SNMP lately for a very large client. I’ve come to realize that it isn’t as simple as I thought (or the name implies). Oh, don’t get me wrong, the foundations of SNMP are very simple. It’s the execution! The enterprise specific MIB info, the integration of information into the MIB….it’s endless! And lord forbid there is an issue with the way the network device formats enterprise specific information.
I figured we should break SNMP down quick for those who still think it’s “simple”:

We’ve got 3 subsystems:
Master agent
Subagent
Management station or NMS

We’ve got five core protocol data units (PDUs):

1. GET REQUEST – used to retrieve a piece of management information.
2. GETNEXT REQUEST – used iteratively to retrieve sequences of management information.
3. GET RESPONSE – used by the agent to respond with data to get and set requests from the manager.
4. SET REQUEST – used to initialise and make a change to a value of the network element.
5. TRAP – used to report an alert or other asynchronous event about a managed subsystem. In SNMPv1, asynchronous event reports are called traps while they are called notifications in later versions of SNMP. In SMIv1 MIB modules, traps are defined using the TRAP-TYPE macro; in SMIv2 MIB modules, traps are defined using the NOTIFICATION-TYPE macro.




Other PDUs were added in later versions, including:

1. GETBULK REQUEST – a faster iterator used to retrieve sequences of management information.
2. INFORM – an acknowledged trap

Okay…so where is all the data kept?? MIB’s OID’s and SMI! Woo hoo. Sounds exciting and not so simple. Maybe they’ll change it to NSSNMP. Don’t worry, we’ll save that for part 2…

There is a new set of ProCurve MIB’s that isn’t on the public site yet. Lots of people have asked for them, so here you go…

Ever been locked out of the console on an HP ProCurve switch? The switch still operates perfectly, you can still send SNMP set/get’s, but no CLI session…? There are a variety of reasons for this to occur, from having too many open telnet sessions, to a hung console session. You can try sending “break” or type in a CTRL-q command, since it is possible that the you inadvertently stopped the output stream of characters (this can be done with a CTRL-s command).
But most of the time only way to get out of this mode is a reboot.

What!!?! Frustrating right?

Don’t worry, there’s a better way to protect yourself from this type of problem/inconvience/irritation. This will require you to be proactive, but nothing like being prepared! By default the console inactivity time is set to “0″ or infinite. Set the console inactivity timer to a non-default value, say 10 minutes. Since they have set the inactivity timer on their switches, they have not experienced any console/serial port hangs.

Syntax:

console inactivity-timer < 0 | 1 | 5 | … >
Usage: console inactivity-timer <0|1|5|10|15|20|30|60|120>
Description: Set the number of minutes of no activity detected on the console port before the switch terminates a communication session. ‘0′ means disable inactivity timer.
Default is 0.
Supported Values:
• 0
• 1
• 5
• 10
• 15
• 20
• 30
• 60
• 120

This command only takes affect after the switch has been rebooted, so make sure to schedule a window to upgrade all of your switch configs with this invaluable little tip.