Yesterday I was working on several things, but needed to interrupt my stuff. So I came up with this little piece of code I put in my .xinitrc. It checks if the file remindme.txt exists and than checks, if it has any content. To display my notes I use xmessage. The content of remindme.txt is plain text.

if [[ -e remindme.txt ]]; then
    test `wc -l remindme.txt | cut -d' ' -f1` -gt 0 && \
        xmessage -geometry 320x400+1600+17 -file remindme.txt &
fi

When I now login and my window manager starts, it also displays my notes.

Short and quick!

Update #1: Forgot to enable comments…

Update #2: Made a screenshot earlier, but forgot to upload it…

It's the time of the your were I need to study for my upcoming exams. After I've noticed, that I am doing complete bullshit like 20 - 5 = 5, I decided to do something to relax. I'm pretty good in relaxing in front of my computer, and here I am and also something useful - oh my pizza arrived, brb. During the last week, my IP address changed and I where unable to access my NAS from outside. This was part one on my todo list, the second and final part was to figure out why Ubuntu doesn't ask for two passwords for my two encrypted partitions.

When you call a server your own, where you have full control of the Bind configuration and wanted to have your own dynamic DNS for your home IP address, this is might for your interest.

The easiest way is:

# dnssec-keygen -a <alg> <name>
# e.g.:
$ dnssec-keygen -a HMAC-MD5 home.foo-bar.com

Two file will be genereted, a

K<name>+<alg>+<id>.key

and

K<name>+<alg>+<id>.private

Open one of these files or use cat to get the output. If you opened the file .key, you take the last column, the string often ends with ==, if you opened the file .private you copy the string for Key:.

EXAMPLE0SEcr3tString00==

or your named.conf.local, or where ever you keep track of your different zones. I'll demonstrate with a small example what changes you should make. At first define the key, your client uses to authenticate:

key "home.foo-bar.com" {
    algorithm hmac-md5;
    secret "EXAMPLE0SEcr3tString00==";
};

After that you need to update the zone configuration for foo-bar.com. For example the section looks something like this:

zone "foo-bar.com" {
    type master;
    file "db.foo-bar.com";
    allow-transfer {
        10.0.1.1;
        common-allow-transfer;
    };
};

You need to insert an update-policy¹⁾.

zone "foo-bar.com" {
    type master;
    file "db.foo-bar.com";
    allow-transfer {
        10.0.1.1;
        common-allow-transfer;
    };
    update-policy {
        grant home.foo-bar.com name home.foo-bar.com. A;
    };

};

As a short conclusion for the update-policy syntax, the first parameter grant allows use to update, if the rest of the rule matches. The second parameter is our key we defined above, the third is a matching rule. I'm using the full domain name to check, You should have a look at the Bind documentation to see the other options. name is followed by the matching name and the last one is the type, 'A', 'CNAME', or 'TXT'²⁾.

After everything is setup you need to reload your Bind DNS server.

$ rndc reload

Your Bind is now configured.

I came up with the idea using my DNS server to manage my dynamic address, while scrolling through the OpenWRT Kamikaze package list, and so I gave it a try.

Edit your /etc/ipupdate.conf:

server "ns.foo-bar-com"
{
    zone "foo-bar.com"
    {
        hosts "home"
        keyname "home.foo-bar.com"
        keydata "EXAMPLE0SEcr3tString00=="
    }
}

Now, execute ipupdate

$ ipudate
getconfig: loading '/etc/ipupdate.conf'
Detected IP: 10.00.100.200

You also should see a success message. Sorry but I haven't copied it, but you'll recognize, when it is successful. Otherwise you'll see some errors. If you check in your Bind directory, (e.g.: /var/cache/bind) you should see a file named home.foo-bar.com.jnl. After 15 minutes the changes will be merged with your db.foo-bar.com zone file.

You can start ipupdate as daemon.

$ ipupdate start

If also installed a cronjob to check every 24h, if an update is needed.

59 23 * * * /usr/sbin/ipupdate

I've updated to Xubuntu 9.04 recently and using encrypted LVM to secure my stuff. After I've set up a second partition encrypted with cryptsetup and hooked it up in my LVM. Ubuntu didn't ask for two passwords during boot. I can remember using Debian using with two encrypted partitions and putting the entries into your /etc/crypttab and updating the initrd's was enough to get asked twice during boot. Ubuntu seems to fail at this point. In this chapter I'll want to show you how I've solved this problem, for now. This solution isn't automated, yet, and using update-initramfs overwrites the changes.

With Ubuntu it unlocks my root partition and trying to bring up all volumes of my volume group. At this point boot will fail, because one of the volumes is encrypted separately. After a minute,Ubuntu will drop you to a fail-over console. At the (initramfs) prompt I needed to unlock my second partition and hit Ctrl+D to resume to normal boot.

Extract the initrd of your current kernel:

$ mkdir /tmp/initrd-$(uname -r)
$ cd /tmp/initrd-$(uname -r)
$ gzip -dc /boot/initrd.img-$(uname -r) | cpio -id

Open the file conf/conf.d/cryptroot and add a separate line with your second device. You should know your UUIDs for this. Here is mine, after I've edited it.

target=sdc1_crypt,source=/dev/disk/by-uuid/56fc9490-4afd-484f-9574-640bec210fe1,key=none,lvm=goat-root
target=sdd1_crypt,source=/dev/disk/by-uuid/f6b68c07-ad87-46a6-8602-94889c1233b8,key=none,lvm=goat-home
target=sdc1_crypt,source=/dev/disk/by-uuid/56fc9490-4afd-484f-9574-640bec210fe1,key=none,lvm=goat-swap_1

I've added the line starting with sdd1_crypt.

To complete the setup we need to pack the initrd back together. Make a backup of your old initrd.img first!

$ cd /tmp/initrd-$(uname -r)
$ find ./ | cpio -H newc -o | gzip -c > initrd.img-$(uname -r)
$ cp initrd.img-$(uname -r) /boot/

You now should be asked twice for a password, or more, if you have more devices added.

Hi guys and girls reading this blog entry. Today I've encountered a strange problem with one of my partitions. My debian is unable to mount /boot, but it boots up normally, just /boot isn't avaible.

The strange thing is. It's still there - why I believe this?! It's simple. Grub loads my kernel and initrd image from /dev/sdd1. During boot it gets mounted into /boot. Having a look at /dev shows me:

goat:~% ls /dev/sdd*
/dev/sdd /dev/sdd5

A listing of the partition table shows this:

Partition Table for /dev/sdd

               First       Last
 # Type       Sector      Sector   Offset    Length   Filesystem Type (ID) Flag
-- ------- ----------- ----------- ------ ----------- -------------------- ----
 1 Primary           0      498014     63      498015 Linux (83)           Boot
 2 Primary      498015   976768064      0   976270050 Extended (05)        None
 5 Logical      498015   976768064     63   976270050 Linux (83)           None
 

As you can see, it's real and the files should be there too:

 goat:~% cat /dev/sdd | strings -n 10 head
 
Loading stage1.5
/grub/stage2 /grub/menu.lst
GRUB loading, please wait...
internal error: the second sector of Stage 2 is unknown.
\~t)G;wXc{E)
gD9AM_:VET
lost+found
System.map-2.6.30-rc5g-newwk
vmlinuz-2.6.30-rc5-2.6.28.8-goat.new
initrd.img-2.6.28.8-goat

I really wonder what the problem is. Trying to insert the missing device through kpartx, also fails:

goat# kpartx -v -l /dev/sdd
sdd1 : 0 497952 /dev/sdd 63
sdd2 : 0 976270050 /dev/sdd 498015
sdd5 : 0 976269987 sdd1 63
goat# kpartx -v -a /dev/sdd
device-mapper: reload ioctl failed: Invalid argument
add map sdd1 (0:0): 0 497952 linear /dev/sdd 63
device-mapper: reload ioctl failed: Invalid argument
add map sdd2 (0:0): 0 976270050 linear /dev/sdd 498015
device-mapper: reload ioctl failed: Invalid argument
add map sdd5 (0:0): 0 976269987 0:0 63

Something goes wrong with the device-mapper

goat# dmesg|tail
device-mapper: table: 253:10: mirror: Device lookup failure
device-mapper: ioctl: error adding target to table
device-mapper: table: 253:10: mirror: Device lookup failure
device-mapper: ioctl: error adding target to table
device-mapper: table: 253:10: linear: dm-linear: Device lookup failed
device-mapper: ioctl: error adding target to table
device-mapper: table: 253:10: linear: dm-linear: Device lookup failed
device-mapper: ioctl: error adding target to table
device-mapper: table: 253:10: linear: dm-linear: Device lookup failed
device-mapper: ioctl: error adding target to table

During boot I recognized similar messages, but instead of dm-linear it's called dm-mirror.

If you have any ideas fixing this, please leave me a message. Thank you!

Update: The Problem seems to be resolved. Don't know what the cure was. I've just removed some packages and switched back to sysvinit:

[INSTALL] sysvinit
[REMOVE] upstart
[REMOVE] upstart-compat-sysv
[REMOVE] upstart-logd
[REMOVE] gparted
[REMOVE] libdmraid1.0.0.rc15
[REMOVE] libparted1.8-10
[REMOVE] parted

When you ever tried to loop over a file listing with filenames containing SPACES, you will get into trouble, because for splits its arguments on those SPACES.

for foo in `ls -1`; do
  echo $foo
done

This will print a list of one word per line.

for foo in `ls -bQ1`; do
  echo $foo
done

This looks promising, because of the -b and -Q switch. -b will escape non-visible characters and -Q quotes the results, but even this attempt will fail with the same problem.

Here is my solution to avoid this problem.

ls -b1 | while read filename; do
  echo "$filename"
done

Passing the ls output to a while loop through a pipe will make it possible to process files with SPACES in its name.

Here is my example to use this and where I've encounterd this problem. I've tried to move some audio files into sub-folders named by the album-tag of the file.

ls -b1 *.flac | while read filename; do
    echo ">>> Processing file: $filename"
    album=""
    flac -t "$filename" 2>/dev/null
    [[ $? -eq 0 ]] && album=$(metaflac --show-tag=album "$filename" | cut -d'=' -f2)
    if [ "$album" != "" ]; then
        album=$(echo $album | strings)
        echo "[ ] Album: $album"
        mkdir -p "$album"
        mv "$filename" "$album"
        echo "<<< $filename moved."
    else
        echo "[!] ERROR on file: $filename"
    fi
done

Update: demod came up with a solution for doing it with a for-loop. (Thanks)

for foo in *; do
  echo "$foo"
done

Which is much easier to read. For aboves usage example we replace

ls -b1 *.flac | while read filename; do

with a much shorter

for filename in *.flac; do

If some of you have tried to access my site in the past few days, may have noticed some things not working properly. I have moved to a new server. This time it is a vserver hosted by http://www.hosteurope.de. I liked my old server, but we need it for some important tasks and I don't want to interfere with my stuff. I also moved both goatpr[o0]n.de domains to this new host. In the past they were hosted at http://www.liberty-hosting.de/. I liked it, especially the support! But I wanted to have everything packed together, so I needed to move. A third domain including webspace, is also about to move. But the host isn't very friendly, didn't answer on support mails and so on, but I need to write them a email again. I cannot recommend http://www.xpservices.de. Everything should be back at normal.