The World’s Longest Console Cable – Smartflex Router Series

console_cable_header

As everybody knows, console cables and their corresponding interfaces are a scourge on a network administrators existence. They are always too short, COM-ports get renumbered, Baud-rates are unknown, etc. By their very nature, you also have to be physically present in the vicinity of the box you’re trying to configure. Today we’re going to change all that, by constructing a wireless console cable adaptor with almost infinite range.

I can almost here people shouting something about SSH and Telnet to their screens, as well as something about the Airconsole. First and foremost, telnet and SSH are both wonderful, after you’ve configured the device, but for initial setup, they sadly can’t be used. Airconsole is quite nice, and I use it almost daily, but it’s useful when you want to sit somewhere comfortable when configuring a device in the same room as you, not so much when you want to be comfortable on a beach half a world away from the offending device.

This we’re going to change today, with one device, the B+B SmartFlex router.

Continue reading

Advertisements

How to Stack HPE/Aruba 2920 Switches – HP Networking Series Part 4

Stack

The 2920 series switches from HPE/Aruba has a special feature that no other switches in the old ProCurve/E-Series line-up has; dedicated stacking interfaces. On other HPE switches, stacking merely facilitates “simplified” management, by allowing you to manage any switch in the stack from the stack commander’s CLI. On the 2920 series however, you can use dedicated stacking interfacing, giving you 2 20 Gbps links to other switches in the stack and merging two to four switches into one coherent switch. This is quite similar on how stacking works on some Cisco models, but the specialized stacking interfaces gives you higher speeds between the members. In this article, I’ll show you have to set up the commander switch and how to add switches to the stack once it’s created. I’ll also go into how it affects the configuration of the switches, which is quite crucial if you want to stack switches that are already in production.

Continue reading

Using the reload command with HP Switches – HP Networking Series Part 2

When configuring network hardware over the network itself it’s very easy to issue a command that breaks the network connection to the switch, requiring you to visit the switch and connect to it with a console cable to fix the issue. This is not a big problem if the switch is just down the hall in a closet somewhere, but in today’s world a switch that you’re configuring might be hundred of miles away from where you’re sitting.

Of course this whole issue can be sidestepped by just thinking through what you’re doing before hitting the enter key, but in the real world this is not always an available luxury.

The solution to this problem is to cleverly utilize the “reload” command in combination with the “write” command.

Continue reading

How to set up STP on HP Switches – HP Networking Series Part 1

Setting up Spanning Tree on HP switches is really easy, after you’ve done it a couple of times, but getting it configured can seem like a daunting task for beginners, espcially in a production network. Here are some simple steps on how to get it up and running in no time. Note that enabling STP might cause small network outages, so don’t do this in a production network if you haven’t tried it before! All commands are written within “citation”-marks, so that’s what you need to enter into the CLI on your switch. For completeness I’ve added all the commands in order at the bottom of the post, which might be handy if you’re just looking for a CLI-reference.

Continue reading

First look at the Raspberry Pi

Yesterday my Raspberry Pi arrived, and today I had some time over to set it up and have a first look.

First of all, I don’t see this becoming a desktop machine for anybody, it’s just too slow to be able to even render webpages fast, never mind streaming HD video, at least the Raspbian version, I haven’t tried the media PC oriented distributions yet, but since they almost certainly lack support for Netflix, HBO and the like, I’m not so confident that they will be much better.

Anyhow, where I think the Raspberry Pi will indeed be useful are areas like automation and tasks were you need to interface with the real world. A good example is an low cost stratum 1 NTP server. The Raspberry has more than enough processing power to handle thousands of NTP querys per second, and it’s very easy to interface it with a serial GPS receiver for example.
Another great application would be custom made protocol converters for industry applications, which mostly convert a serial line protocol to something that can be shifted over Ethernet.

I’ll look into the hardware aspects of the Raspberry during the following weeks and get back to you guys with an update. One thing I’d really like to build is a I/O board for it, so we’ll see about that. Or maybe a NTP time server coupled to a big clock of some sort (maybe even the relay clock I’ve been planning to build for years now).

Annoying PoE Errors

Procurve has you covered

Last week I updated the core switches at one of our sites to two new fancy HP 5406zl. Along with all the nice new features that came with these two beasts was PoE on all ports. All was good and well, until today, when the logs on one of the switches started filling up with 3 errors about PoE delivery on a specific port.

After a lot of cable-tugging and google, it turned out that a subcontractor had installed a small V1810-8G switch in the network. And among other things, these small buggers can be fed by PoE on Port 1. But since it was hooked up with a wall-wart, that collided with the PoE-feed. One would think that HP could put in a bit of error checking on the PoE side so that a device that already have a functioning power supply does not request a PoE-feed.

So, if you’re struggling with a log full of PoE errors (I got 110 000 events in half a week), do a quick google if the connected device supports PoE.

Professional Rack Installation = 8 Hours of Labeling


DELL Server im Rack

Originally uploaded by BlaM4c

Last week I spent close to 8 hours printing and applying labels to different cables in a packed and live rack. And it didn’t help that the rack runs our virtual servers, ie. our most mission critical servers.
The rack itself consists of 3 identical 1U servers from Dell, 2 Gigabit Dell switches trunk’d into each other running separate VLANs for iSCSI and “normal” traffic. The EqualLogic HDD-array that provides the storage for the servers runs 2 controllers, both with a connection to both switches. Then there’s 5 CAT6 cables for each server providing the iSCSI, server management, virtual server NIC etc.

All in all I think I labeled 5 cables per server (15), 3 per EqualLogic module (6), 3 iSCSI cables for some other servers and 2 trunk cables. In both ends. So if my math is correct, that amounts to 30+12+6+4= 52 labels on just network cables. Then the servers needed 2 labels each (6), the EQ-box plus labels identifying the controller modules (3), the 2 UPSs and the 2 switches. That’s 13 more labels. So all in all I did over 65 labels that day.

Now, I’m perfectly clear on why this rack needs labels just about everywhere. Imagine on cable comes loose when you’re fiddling with something. Then if that cable end is labeled with which port it goes into, then reconnecting it correctly becomes a very minor task, and can be done quite swiftly.

The problem I had with this task is that the week before, 2 technicians from Dell spent the whole week setting up the rack. It would have been a lot easier for everyone if they’d bothered labeling every cable as it went into the rack in the first place instead of me carefully tracing each and every cable in the operational rack.

More Bloody Printers

Today I got into another argument with a network printer. I installed a new computer at a remote location, and they had their own Samsung network printer that I had to connect manually.

Most people that aren’t familiar with Samsungs can’t probably understand why this would be a problem, just go into printers, do a network search, and add the thing. This is indeed the correct procedure for standard network printer, like a Canon for instance, but not for Samsungs.

The last Samsung I had to install was pretty complicated as well, it didn’t show up in the network listings, as that particular model didn’t use IP-addresses but some weird UDP broadcast system that you needed a special “admin” tool to find and configure, both when installing the printer and when installing them for the user. This one, however, was a bit different.

This blasted thing has it’s own IP address, but, when you go to install it, it just smiles at you for 5 minutes and then Windows complains that it can’t connect to the printer. And as I didn’t remember the awkwardness of installing the last one, I spent the next 15-30 minutes troubleshooting the network connections, which consisted mainly of loose handmade network cables (remember my last post?).

Anyhow, finally, after checking cables and other network gear, I remembered the other printer and downloaded the drivers. I tried installing them on the user account by selecting “Run as…” instead of running it under my account, which resulted in the installer freezing 10 minutes in when it called some other part that tried to run without admin privileges (they probably didn’t think of programming in a fail safe or making sure the other process inherited the privileges of the first one). This got me even more frustrated (this all happened just before lunch time) as I hate having to install printer software under the admin account just to find out it doesn’t add the printer for the users and it won’t work to install them there either, then you need to upgrade the user to admin for a while and it’s just such a PITA.

Anyhow, after installing the blasted thing under my admin account, it finally worked, and miracle of all miracles, the installer added the printer for all users, be they domain or local.

So, the point of my endless rant is that why can’t some manufacturers make TCP/IP printers that can be added through the Windows dialogues, when some apparently can, and WHY do they choose these insanely complicated workaround tools for this? I can understand it when it’s one of these UDP machines I’m dealing with, as there’s an actual need for that tool, and it’s quite handy as well (try upgrading a server shared printer with a fixed IP to a static IP from the DHCP and you’ll see what I mean), but WHY OH WHY do they do it with the ones that use TCP?

The Magical “No HTTP-traffic” CAT5 Cable

Now I’m really starting to doubt myself. For 2 weeks have I been troubleshooting a Vista machine that just got connected to the network. I’ll start to from the beginning;

2 weeks ago we had a contractor add a double CAT5 cable between two building so a computer in the second building would get internet access. The contractor finished his job, and I was about to go get some patch cables to wrap up the gig when I get a call that one of the teachers (the location is a school) had already scrounged up some cables, connected the computer to the wall outlet and the other outlet to the switch, but he couldn’t get it to work.

I went over without any patch cables thinking this would be an easy task, he probably connected the patch cables to different outlets in both ends. When I get there I notice that he used some of those old style cables, you know, just a crimped on 8P8C connector at the end of the cable, no dust cover, no injection molded strain relief. Determined to get this to work fast I first check his connections, no luck there, he actually used the green outlet in both ends (Lexcom outlets). Thinking that the contractor might have forgotten to fault-check the cables (Lexcom seldom has any connection problems so sometimes the rookies tend to get lazy), so I switch over to the red outlets, still no luck.

I then proceed to blame the switch, it was a cheap dumb desktop model and I knew it has had some problems before that could only be solved by a hard reset. Luckily I had another dumb 8 port in my car (Zyxel, very nice gear for this kind of things) that I knew worked so I switched it out. Suddenly the connection went live, and I thought I had finally found the problem, not as easy as I thought, but relatively simple and now I had fixed another problem I knew was just waiting to happen (the switch).

But, when I get over to the computer in question, things get really weird. The computer gets a physical connection (blinking lights on the NIC), and Windows even manages to get an IP. I open IE8, and it finds Google just fine, but hangs midway through. I try pinging our HQ’s gateway, no problem there, I then try to ping google.com, and it resolves the IP just fine, but then hangs. I then remember that our network doesn’t allow outgoing ping requests for some reason (I didn’t design it), so no worries there either.

After about an hour of fiddling (it’s getting late on this cold Friday evening), I give up and decide to try again on Monday.

I go back on Monday morning with a pair of fresh CAT5 patch cables, I thought that since the old ones were a tad short anyhow I might as well replace them, one less thing to worry about.

I get there, and the building with the computer is locked, so I switch the one by the switch and hand the other one to the enterprising teacher, saying that “this will certainly not fix it, but I’ll be back in a couple of days when I’ve had some time to think this through”.

During the next one and a half week I spend a lot of time googling this problem and consulting guys with more Vista experience than me. This led to all kinds of weird theories, Group Policies that weren’t right due to the fact that the machine was on the same VLAN as the domain controller but not in the domain due to the home license. I didn’t buy this so I never looked into it.

Now, fast forward until today, and I went back there armed with a Knoppix CD to make sure it wasn’t a hardware related issue. The teacher had changed the network cable, but he wasn’t there so I couldn’t talk to him. I booted into shell, everything looked alright, I started Lynx, and google loaded just fine. This really started to annoy me, so I took a copy of the ip-settings from Knoppix and booted into Vista to do the same there so I could post them on the web in case I was going blind or something.

And wouldn’t you know it, the damned thing worked. And the only thing I had done to fix it was changing to patch cables.

Ladies and gentlemen, I give you the straight CAT5 cables that passes DHCP requests, pings and DNS look-ups, but not HTTP traffic.

Ain’t life grand! 🙂

Peculiar Network Problem

This Monday I was called out to fix a computer that couldn’t get on-line. I quickly discovered that the problem wasn’t with the computer itself, but with it’s wired connection. The computer was wired as follows;

Control Room: Computer <-> Jack (unlabeled)

Electrical Room: Jack <-> Jack A1

Switch Room in another building: Jack A1 <-> Port 13 on the main switch

As you can see, it wasn’t very clear what was going on as the first connection wasn’t labeled. Also, the port number on the switch wasn’t that nice either.

Well, I started looking for clues. The weird thing was that both the switch and the computers port lights were on, so the connection was obviously good. Therefore I assumed that the NIC was faulty in the computer, and took the computer to my office to troubleshoot it. This turned out to be the exactly opposite to what I should have done, but more on that later.

So, I took the computer back to my office, and guess what? It worked perfectly. So now I didn’t really know what to think, but anyhow I needed to return it, so I went back there.

This time I checked every connection, and the all matched up, everything seemed to be working properly, but it didn’t. Then I noticed the give-away sign. The activity LED for port 13 was blinking like crazy. Like it was working at max capacity, which I knew that it didn’t since the computer didn’t send or receive any packages. So I figured that the switch might be faulty, and since half of it’s ports was unoccupied, I swapped the cable over to the next empty port, Port 15.

Then the weirdest thing happened. The LED for Port 15 went on, but the LED for Port 13 didn’t go out, it just kept going, blinking like crazy. Later I learned that there had been some tunderstorms in the area during the weekend, which had nocked out the climate control as well. The switched later failed compleatly on Tuesday morning, so I guess a voltage-spike must have fried it.