NTP over IPv6 on a Netscreen

I spent a little time trying to get my Netscreen to talk to an IPv6 NTP server today but it wouldn't work. I could ping the IPv6 server but no NTP. I am running 5.4 code and it does IPv6 flawlessly but NTP didn't want to work. But I found out why. Per Juniper, this isn't supported until 6.3 and I am not sure if there will even be a 6.3 code put out for the EOL 5GTs. Bah!

downforeveryoneorjustme.com

Interesting tool to have in the toolbelt: downforeveryoneorjustme

First router

Obviously this goes back pretty far but I found this recent post on the Cisco Blog site linky. It has some basic ICMP how-tos for IOS but it links to something very nostalgic - The story of the PING program. The reason it is nostalgic for me is it is a time in which I first saw networking of any real sort (other than goofy access to local BBSes using the 300 bps modem on my Apple IIe which doesn't quite count) at the local University (albeit I understood none of it) and I would love to have been able to just go back and look over sholders during those days. The site even has some old TCP/IP Digest archives featuring many names now famous in the history of networking.

I wish I had a time machine.

Ethernet crossover

Yesterday provided a moment of 'wtf' when the test cable-diagnostics on a Cisco switch was showing some really screwy output and it made me wonder if the cabling was incorrectly terminated. To make a long story short, the connection was switch-to-switch over copper gigabit Ethernet and the MDIX was causing the odd results but I was sent on an adventure into answering a question, "If copper Gigabit Ethernet uses all 4 pairs for data, is there something unusual about the straight-through versus crossover pinouts for Gigabit Ethernet?" Turns out there is. First, the 'duh' moment where 10/100 Ethernet pinouts are displayed:

 
10/100 Megabit normal:

Pin Function Color Match Function Pin
1     TD+   White/Orange           1
2     TD-   Orange/White           2
3     RD+   White/Green            3
4           Blue/White             4
5           White/Blue             5
6     RD-   Green/White            6
7           White/Brown            7
8           Brown/White            8

10/100 Megabit crossover:

Pin Function Color/Match Function Pin
1    TD+     White/Orange  RD+     3
2    TD-     Orange/White  RD-     6
3    RD+     White/Green   TD+     1
4            Blue/White   
5            White/Blue   
6    RD-     Green/White   TD-     2
7            White/Brown 
8            Brown/White

Nothing unusual there - standard 1,2 3 and 6 in use. If you use this pinout on most Gigabit Ethernet ports, you end up with 100 Mb/s. This creates a hard problem to troubleshoot since the ports will show up and you will get data over the cable but not at the speed you want. Of course, anyone not terminating Cat 5 (or better) with all 4 pair needs to be beaten in my opinion. Now for the Gigabit Ethernet pinouts:

First, Gigabit Ethernet normal:

Pin Function Color Match Function Pin
1    BI_DA+  White/Orange  BI_DA+  1
2    BI_DA-  Orange/White  BI_DA-  2
3    BI_DB+  White/Green   BI_DB+  3
4    BI_DC+  Blue/White    BI_DC+  4
5    BI_DC-  White/Blue    BI_DC-  5
6    BI_DB-  Green/White   BI_DB-  6
7    BI_DD+  White/Brown   BI_DD+  7
8    BI_DD-  Brown/White   BI_DD-  8

So, this should work fine (assuming the cable wasn't terminated by someone trying to cause you grief). But the crossover is strange:

Gigabit Ethernet crossover:

Pin Function Color Match Function Pin
1    BI_DA+  White/Orange  BI_DB+  3
2    BI_DA-  Orange/White  BI_DB-  6
3    BI_DB+  White/Green   BI_DA+  1
4    BI_DC+  Blue/White    BI_DD+  7
5    BI_DC-  White/Blue    BI_DD-  8
6    BI_DB-  Green/White   BI_DA-  2
7    BI_DD+  White/Brown   BI_DC+  4
8    BI_DD-  Brown/White   BI_DC-  5

This is really odd and isn't exactly easy to remember. What makes matters even more fun is the MDIX auto feature on most (if not all) Gigabit Ethernet ports on Gigabit Ethernet switches (at least good ones) will correct the crossover pinout and work so you will never know the cable is 'wrong'. A cable tester (or the Cisco 'test cable-diagnostics' command) will likely give some odd results - especially on switch-to-switch connections. For more on the Cisco 'test cable-diagnostics' command, look here.

Tracert - two ways to do it

A discussion came up at work about how traceroute works on different platforms. I knew some use UDP and some just use ICMP but I didn't know why they were so different. Here is a description of why and how each works. Rather interesting.

Active vs Passive FTP

Active vs Passive FTP.

SQL Crunch

SQL Crunch. Lots of links to SQL information.