PORTSCOUT=limit:^1.4

In a ports Makefile to prevent portscout looking for newer versions, very useful if you need to put a legacy version in the tree.

Prepare the target drive

cd /tmp/
df -h > liveFSsizes
bsdlabel da0s1 > liveSPsizes
dd if=/dev/zero of=/dev/da1 count=2
fdisk -BI /dev/da1
bsdlabel -B -w da1s1
bsdlabel -R da1s1 liveSPsizes
cat liveSPsizes
newfs -U /dev/da1s1a
newfs -U /dev/da1s1d
newfs -U /dev/da1s1e
newfs -U /dev/da1s1f


Mount each partition and dump the source partition (da0) to our backup (-)

mount /dev/da1s1a /mnt
cd /mnt
dump -0Lauf - /dev/da0s1a | restore -rf -
mount /dev/da1s1d /mnt/var
cd /mnt/var
dump -0Lauf - /dev/da0s1d | restore -rf -
mount /dev/da1s1e /mnt/tmp
cd /mnt/tmp
dump -0Lauf - /dev/da0s1e | restore -rf -
mount /dev/da1s1f /mnt/usr
cd /mnt/usr
dump -0Lauf - /dev/da0s1f | restore -rf -
sync
cd /tmp
rm liveSPsizes
rm liveFSsizes
umount /mnt/usr
umount /mnt/tmp
umount /mnt/var
umount /mnt


Note when you fdisk you should expec the following errors:

******* Working on device /dev/da1 *******
fdisk: Class not found


The original version of this can be found on the FreeBSD fora in this thread: http://forums.freebsd.org/showthread.php?t=11680.

I wrote something like:

require 'net/http'
res = Net::HTTP.start('http://www.example.com') do |http|
  http.open_timeout = 4
  http.read_timeout = 4
  http.get('/index.html')
end
puts res.body

Pah, why does Ruby seem to totally ignore the timeout settings? It just ignores it… humm. Okay some digging required. It would appear the instance method invokes the connect then yields so by the time the block is actually executed the connection has already been attempted, with the default timeouts – rather than those that appears to be set.

So we need to use code akin to:

require 'net/http'
http = Net::HTTP.new("http://www.example.com")
http.open_timeout = 4
http.read_timeout = 4
res = http.get('/index.html')
puts res.body

For it to respect our timeouts!

case something.class
  when String, Symbol then 1
  when Fixnum then 2
  when Time then 3
end

Only to find it didn’t work, always returning nil. A bit of trawling turned up the following blog post: http://www.postal-code.com/mrhappy/blog/2007/02/01/ruby-comparing-an-objects-class-in-a-case-statement/, which explains that you don’t need the .class. So my above code would become:

case something
  when String, Symbol then 1
  when Fixnum then 2
  when Time then 3
end

Which works.

• No OS, with 2 Seagate 2.5″ drives, 2x4GB RAM, DVD, Drac Express: 60.75W
• No OS, with 1 Seagate 2.5″ drive, 2x4GB RAM, DVD, Drac Express: 58.32W
• No OS, with 2 Seagate 2.5″ drives, 2x4GB RAM, Drac Express: 60.75W
• No OS, with 2 Seagate 2.5″ drives, 2x2GB RAM, Drac Express: 61.25W
• No OS, with 2 Seagate 2.5″ drives, 1x4GB RAM, Drac Express: 60.75W
• No OS, with 1 Western Digital RE2, 2x4GB RAM, DVD, Drac Express: 57.6W
• No OS, 2x4GB RAM, DVD, Drac Express: 55.0W

Now putting an OS on the machine and actually optimising so the OS can power down the CPU, etc.

• FreeBSD 8 (base), with 2 Seagate 2.5″ drives (RAID1), 2x4GB RAM, DVD, Drac Express: 60.75W
• FreeBSD 8 (optimised), with 2 Seagate 2.5″ drives (RAID1), 2x4GB RAM, DVD, Drac Express: 55.0W
• FreeBSD 8 (heavily optimised), with 2 Seagate 2.5″ drives, 2x4GB RAM, DVD, Drac Express: 51.0W

On a more expensive calibrated meter we tickled 47VA with that last configuration.

SSH into the iDRAC and then type the racadm command as indicated:


/admin1-> racadm config -g cfgRacSecurity -o cfgRacSecCsrKeySize 2048
Object value modified successfully
/admin1->


Credit to: http://christoph.ruegg.name/blog/2010/7/20/how-to-create-2048bit-certificate-csrs-for-dells-idrac6.html

/System/Library/Frameworks/Ruby.framework/Versions/1.8/usr/lib/ruby/1.8/rexml/element.rb:685:in `write': undefined local variable or method `output' for <xml_node> ... </>:REXML::Element (NameError)

A bit of digging lead me to this bug report, which lead me to look properly at the code and sure enough correcting the code (line 674) corrects the bug and I can generate XML. i.e. changing

def write(writer=$stdout, indent=-1, transitive=false, ie_hack=false)

to

def write(output=$stdout, indent=-1, transitive=false, ie_hack=false)

Note writer has now become output. Now since the that function in REXML is depreciated anyway I should probably modify the XML-Mapping code to use the more modern REXML::Formatters

Anyhow if I write something like this:

File.open('foo:bar.txt', 'w') {|f| f.write("hello world") }

Rather than get a file in the current directory named foo:bar.txt I actually get one named foo/bar.txt. I’ve tried various potential escape sequences (\: \c etc.).

The colon name totally works as you’d imagine on my FreeBSD machines, so it must be something either to do with the Ruby version or OSX – I guess it must be half identifying the colon as a path seperator, but the file created with the ‘/’ in it is a proper file with that name.

Chalk one up to not enough time to debug, but enough time to note.

/Library/Ruby/Site/1.8/rubygems/custom_require.rb:31:in `gem_original_require': no such file to load -- xml/mapping (LoadError)

Everything seemed okay, until eventually I noticed that the Gem files themselves were rather restrictive on permissions – owner read and write only. So I had to go into the directory (/Library/Ruby/Gems/1.8/gems/xml-mapping-0.8.1/ on my Mac) and chmod the files to 644 and the directories to 755 and then the Gem loads okay.

cd /Library/Ruby/Gems/1.8/gems/xml-mapping-0.8.1/
find . -type d -exec chmod 755 {} ";"
find . -type f -exec chmod 644 {} ";"