supine.com- home
This site best viewed with a high blood caffeine level and your monitor upside down.

Tue, 24 Jun 2008

Links of the Day: June 24, 2008
CT3 - Disable optional IOS features on high CPU load
"The applet monitors the average one minute CPU load on the router (using the cpmCPUTotal1min SNMP variable) and disables optional features when the CPU load exceeds predefined value."

[2008/06/24 / daily links permanent link]

Wed, 18 Jun 2008

Links of the Day: June 18, 2008
Eight Black - Going. Going. Nearly Gone.
"To put this into some sort of basic perspective, if you had purchased $1000 worth of BlueFreeway stock exactly 1 year ago, that would be worth the princely sum of $42 as of close of play today."

Virtual Geek - So, how **EXACTLY** does VM HA's admittance algorithm work?
"What's the symptom - you can't create new VMs without violating the availability constraints - even when you think you have plenty left, or you upgraded from Virtual Center 2.0.x to 2.5 and can't shake the "Insufficient Resources to Satisfy" HA error."

xAnalisys - VMware Infrastructure 3 in a Cisco Network Environment
"VMware has collaborated with Cisco to produce a guide for deploying VMware Infrastucture 3 with Cisco switches. This guide covers the physical and virtual data network and storage network deployment ... with suggested topologies and designs."

Pawprints of the Mind - Getting Xen up and running: part II
"This post is intended for people who are trying to set up a private network with Xen virtual machines." Virtual networks are unsupported in base Xen but this is a Debian centric how-to which worked first time.

[2008/06/18 / daily links permanent link]

Tue, 17 Jun 2008

Firefox 3 Download Day
Download Day

"Set a Guinness World Record, Enjoy a Better Web. Sounds like a good deal, right? All you have to do is get Firefox 3 during Download Day to help set the record for most software downloads in 24 hours - it's that easy."

[2008/06/17 / tech / software permanent link]

Virtualisation Buzzwords
So I was thinking a bit more about virtualisation buzzwords after my discovery the other day.

"Virtualisation 1.0" appears to be the concept of virtualising physical hosts onto a defined set of infrastructure. Whether the VM farm is in-house or outsourced is largely irrelevent. What is important for buzzword compliance is that there is still a farm with limits, location etc.etc.

"Virtualisation 2.0" appears to be the concept of virtualising physical hosts onto "clouds". The VM farm is provided by infrastructure-as-a-service and it's implied that limits, location etc.etc. are non-issues because they are "someone else's problem".

The problem is that the possibilities in this space are not easily pigeon-holed into only two catergories. It just does a disservice to whatever you are trying to describe to fall back on these kinds of buzzwords.

[2008/06/17 / tech permanent link]

Links of the Day: June 17, 2008
smh.com.au - Cut fuel price, say voters
Why do people think the government should intervene in the supply/demand determined pricing of a scarce resource? Did everyone fail economics 101?

Acer Aspire One notebook review
"if Acer wants it to be seen as anything other than a mini-notebook it needs to quickly launch its connected version as this edition is only a success on the price point, lacking the usability and build quality we've come to expect from this market."

TrustedReviews - Asus Eee PC 901 20G Linux Edition
"Brimming with features and with a new CPU and better battery, the Eee PC 901 personifies exactly the spirit in which the original was created. ...bit more expensive ... lack a more comfortable keyboard, but ... the 901 has once again set the standard."

TrustedReviews - First Look: MSI Wind
"...though the Wind lacks the classiness of the Mini-Note... it's a good looking machine that shares a visual style with the Eee PC - albeit in a slightly larger chassis. ...it looks as though MSI has put together a very commendable effort..."

VM /ETC - Can you Vmotion between different physical data centers?
"Chad goes on to point out that he feels stretching ESX Clusters is a bad idea in general and lists 4 solid reasons to support why. Check out all the whole post at the link above."

Building a Low Cost (Cheap) VMWare ESX Test Server.
"My job now involves more and more virtualised (or virtualized if you're one of our American cousins from across the pond) server implementations so this is a good excuse to go and build my own VMWare ESX test environment."

Virtual Geek: Building a Home VMware Infrastructure Lab
"This is something I consider mandatory if you're going to take VMware as seriously as I think everyone should :-)"

Connect VMware ESX Server to a free iSCSI SAN using Openfiler
"In this article, we take a look at how you can download a free open-source iSCSI server and use it as your SAN storage for VMware ESX and its advanced features."

Openfiler (The link in the previous article is out of date)
"Openfiler converts an industry standard x86/64 architecture system into a full-fledged NAS/SAN appliance or IP storage gateway and provides storage administrators with a powerful tool to cope with burgeoning storage needs."

[2008/06/17 / daily links permanent link]

Sun, 15 Jun 2008

Links of the Day: June 15, 2008
Virtualization 2.0
So I only came across this buzzword for the first time today. A quick Google search and it became clear that it's been thrown about for almost two years. Why not call a spade a spade and bang on about "infrastructure virtualisation" instead?

Rolling Stone - The Battle For Facebook
Short on technical detail, long on melodrama.

[2008/06/15 / daily links permanent link]

Sat, 14 Jun 2008

Links of the Day: June 14, 2008
smh.com.au - ACCC thwarts eBay PayPal plan
A win for common sense! "The [ACCC] has flagged its intention to scuttle a plan by online auctioneer eBay to force its Australian users on to a PayPal-only payments system. Citing concerns about the 'anti-competitive effect' of the proposal..."

Coding Horror: ASCII Pronunciation Rules for Programmers
Be sure to read the comments too... "What the heck is an octothorpe? I know this as the pound key, but that turns out to be a US-centric word; most other cultures know it as the hash key."

PostCarbon Information and Communication Technology (ICT)
"... ICT is ... completely dependent on fossil fuels. IF this ... is not changed ... there will be no PostCarbon ICT, period." Original has horrible formatting.

[2008/06/14 / daily links permanent link]

Fri, 13 Jun 2008

Links of the Day: June 13, 2008
37signals - Git smart: How we're using Git to track our source code
I've been using bzr a lot lately but every man and his dog and their source code seems to be switching to git. Must take a deeper look one of these days.

Changing the IP-address of an ESX host and HA - Yellow Bricks
"...changed the ip-address of three VMware ESX hosts. ...standard VMware procedure, which usually works... In this case after the ip-address was changed HA did not work..." See also http://www.booches.nl/index.php/2008/06/05/change-esx-host-ip-address/

NAT caveats in IOS release 12.4T - CT3
"The Network Address Translation (NAT) in IOS release 12.4T works significantly differently from the previous implementations (including mainstream IOS release 12.4)."

Cloud Computing vs Grid Computing
"...the term 'Cloud Computing' is relatively new to the Technology buzz. But just how new is it? ...analyzing search trends of different computing keywords to try to put everything in perspective."

Audi TP52 crash gybes to avoid broaching Cristabella - 1 of 3
Crazy!

Audi TP52 crash gybes to avoid broaching Cristabella - 2 of 3


Audi TP52 crash gybes to avoid broaching Cristabella - 3 of 3


[2008/06/13 / daily links permanent link]

Thu, 12 Jun 2008

They made this bed, they should sleep in it.
After reading yet another story on a "credit crisis" casualty, this time Bear Stearns, I was affronted enough to write the following letter to the editor:

To: letters@smh.com.au
Subject: Financial companies shouldn't have their cake and eat it too.

Another day, another debt laden company goes to the wall. The US Federal
Reserve steps in "to promote the orderly functioning of the financial system".
Why couldn't they step in during the good times, billion dollar profits and
million dollar bonuses "to promote the orderly functioning of the financial
system"? The industry decries regulation that would "prevent innovation" during
the good times but then they want bail outs when things turn sour. Natural
selection should be allowed to take care of the "innovators". They made this
"bed", they should "sleep" in it.

They published it but they made some small changes, mostly fixing my grammar and spelling:

Subject: Lopsided intervention

Another day, another debt-laden company goes to the wall. The US Federal
Reserve steps in "to promote the orderly functioning of the financial system"
("Nervous traders fear worse to come", March 17). Why couldn't they step in
during the good times of billion-dollar profits and million-dollar bonuses to
promote the orderly functioning of the financial system? The industry decries
regulation that would "prevent innovation" during the good times but then they
want bail-outs when things turn sour. Natural selection should be allowed to
take care of the "innovators". They made this bed, they should sleep in it.

And the fallout from this one? Bear has been bought by JP Morgan Chase for around $2/share after trading near $160/share as little as a year ago. Ka-ching!

Bear Stearns plunge
Source: Yahoo! Finance

[2008/06/12 / social / finance permanent link]

Links of the Day: June 12, 2008
del.icio.us/help/api/posts
Blosxom doesn't have XML-RPC support by default so I'll need to script an API call to download daily bookmarks.

Aaron Straup Cope / Net-Delicious - search.cpan.org
Someone may have already done the heavy lifting for me!

Publishing Links With Perl, by Jeffrey Veen
Just in case I get lazy and need a cheat sheet...

http://edward.de.leau.net/code/wpds.txt
Actually this one turned out more useful.

net::delicious bug #30310: Error messages when used with warnings on
I was getting an error ("Use of uninitialized value in -f at /usr/share/perl5/Net/Delicious.pm line 228.") from my script, seems there is already a bug filed for it.

Royal Pingdom : Javascript framework usage among top websites
Caught my attention because the name MooTools sounds like something JohnF would nave named.

[2008/06/12 / daily links permanent link]

Obscure compliment...
...in a ticket at $work[0]:

MB's mutt-fu is strong.

[2008/06/12 / tech permanent link]

Ubuntu Hardy unlikely to have working Ralink drivers...
So I've been persisting with using the rt2x00 drivers with 2.6.24 on Ubuntu Hardy but it wasn't only slow (interface was quite often at 1Mb/s rather than 54Mb/s) but unreliable, dropping out every 15-20 minutes. A quick bounce of the interface brought it back but it's just un-acceptable to do that indefinately.

I'd been booting into an older kernel in order to get working drivers, in the hope the new ones would be fixed and using them would be a quick reboot away, but after following the bug in launchpad it seemed quite clear that I was dreaming.

I bit the bullet last night and installed the old drivers. I had to pull them from CVS rather than using the packaged ones. But a quick 'make && sudo make install' and everything is working perfectly again.

I left a little note in the bug:

I'm not sure why this is "Medium". Most computers these days are doorstops if
they don't have reliable networking.

You have a LTS release due next month and you are persisting with releasing
broken drivers for Ralink chipsets. There was a point about a month ago where
someone needed to make a call and either revert to the old drivers or commit to
getting the new ones working in time for Hardy.

I for one am using the old drivers.

[2008/06/12 / tech / linux permanent link]

Wireless interface renaming fun on Hardy...
As I mentioned in in my last post I was having problems with wireless.

As the drivers for the rt2500 chipset have been a problem in the past, I assumed it was the same issue rearing it's ugly head again. I was wrong.

My system had a wlan0_rename interface as well as the expected ra0 interface.

$ sudo iwconfig 
lo        no wireless extensions.

eth0      no wireless extensions.

ra0       no wireless extensions.

wlan0_rename  IEEE 802.11g  ESSID:"XXXXXXXX"  

Hmmmmm... looks like some udev fun! I'd come across the udev replacement for /etc/iftab after we swapped out a network card at $work[0] and the interfaces came up in an odd order. I figured it was something similar here.

$ sudo vim /etc/udev/rules.d/70-persistent-net.rules 
:%s/ra0/wlan0/g
$ sudo vim /etc/network/interfaces
:%s/ra0/wlan0/g
$ sudo rmmod rt2500pci rt2x00pci rt2x00lib 
$ sudo modprobe rt2500pci 
$ iwconfig 
lo        no wireless extensions.

eth0      no wireless extensions.

wmaster0_rename  no wireless extensions.

wlan0     IEEE 802.11g  ESSID:"XXXXXX"  

Bingo!

[2008/06/12 / tech / linux permanent link]

Reading something into your bash history...
An interesting idea from jimmac via jdub.

At home:

(marty@merboo)-(03:42:46 Fri Apr 11)-(~)
$ history|awk '{a[$2]++ } END{for(i in a){print a[i] " " i}}'|sort -rn|head
106 screen
96 ssh
48 sudo
43 vim
25 ifconfig
25 cd
17 make
12 ls
12 cat
10 iwconfig

...and at work:

marty@boober:~ $ history|awk '{a[$2]++ } END{for(i in a){print a[i] " " i}}'|sort -rn|head
142 ssh
82 screen
45 sudo
42 dig
24 pstree
18 offlineimap
14 vim
12 telnet
8 ooffice
8 mutt

[2008/06/12 / tech / linux permanent link]

Corner cases make me a sad panda
At $work[0] we've had issues with a server that has a Nvidia SATA controller. The errors look a little something like:

 ata2.00: cmd 60/08:78:3f:8f:3b/00:00:00:00:00/40 tag 15 cdb 0x0 data 4096 in
          res 40/00:00:00:00:00/00:00:00:00:00/00 Emask 0x4 (timeout)
 ata2.00: cmd 61/00:80:66:32:77/01:00:00:00:00/40 tag 16 cdb 0x0 data 131072 out
          res 40/00:00:00:00:00/00:00:00:00:00/00 Emask 0x4 (timeout)
 ata2: soft resetting port
 ata2: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
 ata2.00: configured for UDMA/133
 ata2: EH complete
 ata2: timeout waiting for ADMA LEGACY clear and IDLE, stat=0x400
 sd 1:0:0:0: [sdb] 488390625 512-byte hardware sectors (250056 MB)
 sd 1:0:0:0: [sdb] Write Protect is off
 sd 1:0:0:0: [sdb] Mode Sense: 00 3a 00 00
 sd 1:0:0:0: [sdb] Write cache: disabled, read cache: enabled, doesn't support DPO or FUA
 ata2: EH in ADMA mode, notifier 0x0 notifier_error 0x0 gen_ctl 0x1501000 status 0x0 next cpb count 0x10 next cpb idx 0x0

It appears some sata_nv "features" (NCQ) were added to Linux 2.6.22 but some Hitachi hard drives don't support it. Redhat has errata on this.

Of course we're running that combination. Sigh.

[2008/06/12 / tech / linux permanent link]

Google doesn't seem to care about your privacy...
SMH is reporting about Google's imminent launch of "Street View" in Australia. Richard Chirgwin points out the spineless journalism that let quotes like the following through to the keeper:

Still reeling from the backlash by privacy activists after Street View began in
the US last year, Ms Mayer said Google was developing technology to blur faces
and number plates but she did not know whether it would be ready in time for
the Australian launch.

My response to that would be: if the technology is coming you should wait for it and avoid the snafus in the first place.

[2008/06/12 / tech / internet permanent link]

Bankwest feedback falls on deaf ears...
$better_half was trying to login to Bankwest the other day and it didn't work from the desktop. Tried a few different things and eventually tested it from the laptop which worked. The only difference between the two is that one is amd64 and the other is x86. I had a look at the source but there didn't appear to be any wacky javascript to mess things up.

So, I thought I'd be helpful and let their web developers know:

...our home computer which used to work fine. It just returns her to the login
prompt without any error message.

The system is AMD64 Linux (Ubuntu 8.04) and I had her test with both Firefox2
(2.0.0.7) and Firefox3 (beta4).

I had her test from my laptop and that worked fine.

The laptop is x86 Linux (Ubuntu 8.04) and Firefox3 (beta4).

I wasn't expecting a great response but I got a "talk to the hand"!

Thank you for your message.
Please have your wife contact our 24 hour Customer Help Centre on 13 17 
18 to enable one of our operators to assist with the enquiry.

Due to security restrictions, we are unable to access or discuss 
information specific to your account via this unsecure message service 
as it is not possible to complete the identification requirements.
Please accept our apologies for this inconvenience however the strict 
guidelines of the Privacy Act commit us to this action.

Ummmm, hello?

Sigh.

[2008/06/12 / tech / internet permanent link]

If fibre goes down in a forest, does the twisted pair hear it scream?
So at $work[0] we're trying to bring up a new service for a customer. It's a 2meg service from T, which terminates on our side like so:

fibre -> managed media converter -> twisted pair -> router

We think we're ready to turn it up and a T technician is booked. For 01:30am due to customer requirements. Sigh.

Technician reports he can't see either the media converter nor the router. He's on the "basement switch" (a designation which means nothing to yours truly, especially at that hour of the morning). Eventually my sleep deprived haze allows me to parse that he is on a switch in the facility and can't connect to the service.

Except our understanding is that their end of the service doesn't terminate in the facility. Their switch is in one meet-me-room but it had cable tray / riser capacity issues. So this service's fibre went to the meet-me-room at the other end of the building. Where T doesn't have a switch. But they did have some spare fibre already pulled from an exchange. So a fibre tie later and our service terminates elsewhere. And in my blubbering state I can't convince the technician of this. Cutover canceled. Furrfu.

So what does the subject have to do with this saga I hear you ask?

The technician wanted me to pull the twisted pair from the router to check for link protocol going down on an interface of the switch he was logged into to. Even presuming he was on the right switch, my original reaction was that it wouldn't work because the media converters act like bridges, not repeaters, and the link protocol wouldn't change at his end.

I wanted to confirm / deny this ungodly hour of the morning theory, but my Google-fu appears to be suffering 5 month old child process lossage. JohnF did the heavy lifting for me and turned up this question and answer which covered a similar scenario. The pertinent response that address my question is:

Coming in a bit late here, but...

Has the OP looked at how the media converter is set up? Many media converters
have a test mode where a down link status is not passed from one side to
another. Once you take it out of test mode, it will pass status. Usually it's a
small push button, or a switch.

Sean

So the answer to my question is: it depends!

[2008/06/12 / tech / hardware permanent link]

Many to one SNAT on Cisco IOS
So, at $work[0] we needed to do a "many to one" SNAT on Cisco IOS. I'd only ever previously done this by SNAT'ing to an interface, however this required specifying the IP.

My Google-fu failed me, I could turn up how to do the "SNAT to interface" and "SNAT one to one from a network to a pool" but not quite what we needed.

So I thought I'd try having a pool with a single IP in it, ala:

ip access-list extended SNAT_SRC_ACL
 permit ip w.x.y.z 0.0.0.63 any
ip nat pool SNAT_POOL a.b.c.d a.b.c.d netmask 255.255.255.0
ip nat inside source list SNAT_SRC_ACL pool SNAT_POOL overload

Which worked!

And then five seconds later a workmate found a site that described exactly this technique. Argh!

[2008/06/12 / tech / cisco permanent link]

Daily Links Automated
As can be seen from the previous post I'm back!

The best bit is that post was created automatically. I use del.icio.us to bookmark interesting things I read day to day. It has support for daily links posted to your blog but that requires XML-RPC support which Blosxom doesn't have out of the box.

So I wrote a small perl script which downloads links daily and creates a post if there are any for that day. Most of the heavy lifting is done by Net::Delicious (which I turned into a deb package for Ubuntu Hardy) and to save time a lot of inspiration was drawn from Edward de Leau's WordPress script.

Final problem. I want it to run at midnight UTC (del.icio.us' "get by date" API works in UTC) in order to grab all the links for an entire day when they are fresh, fresh, fresh. But how do you run this out of CRON on a system that is in timezone "Australia/Sydney" with all it's daylight savings fun, fun, fun?

# Only want this to run just before midnight GMT
59 9 * * *	/bin/date +\%z | /bin/grep -q 1000 && /path/to/delicious2blosxom.pl
59 10 * * *	/bin/date +\%z | /bin/grep -q 1100 && /path/to/delicious2blosxom.pl

Run it twice a day but check for the current UTC offset first. The first covers normal Sydney time, the second daylight savings. Sweet!

[2008/06/12 / meta permanent link]