I had just done apt-get update && apt-get dist-upgrade when I saw something that made me cringe:

No apport report written because MaxReports is reached already
Errors were encountered while processing:
initramfs-tools
plymouth
linux-image-3.2.0-38-generic-pae
ubuntu-minimal
plymouth-theme-ubuntu-text
linux-image-generic-pae
linux-generic-pae
mountall
upstart
E: Sub-process /usr/bin/dpkg returned an error code (1)

Well that’s not the exact error – I ended up seeing “disk out of space” among other issues. Basically the /root filled up and it broke my apt-get upgrade right in the middle of updates. No good.
So I go to /boot and clean up the older kernel files – just make sure you delete the “abi”, the “config”, the “initrd”, the “System.map”, the “vmcoreinfo”, the “vmlinuz” etc all with the same version number.

Then I went about reinstalling the packages. Unfortunately since some of the packages were already installed prior to running out of space, the dependencies were a little off.

apt-get install -f was no help:

You might want to run ‘apt-get -f install’ to correct these.
The following packages have unmet dependencies:
initramfs-tools : Depends: initramfs-tools-bin (< 0.99ubuntu13.1~) but 0.99ubuntu13.1 is installed E: Unmet dependencies. Try using -f.

And I didn’t want to have to restore this beast since the last full backup was over 2 months ago. So I had to carry on.

First I noticed that initramfs-tools needed the bin installed, but the bin was too new for initramfs-tools to utilize and therefore could not update even via –configure. Chicken and egg?

So then I looked up what versions were available so that I could downgrade:
apt-cache showpkg initramfs-tools-bin

Versions:
0.99ubuntu13.1 (/var/lib/apt/lists/us.archive.ubuntu.com_ubuntu_dists_precise-updates_main_binary-i386_Packages) (/var/lib/dpkg/status)
Description Language:
File: /var/lib/apt/lists/us.archive.ubuntu.com_ubuntu_dists_precise_main_binary-i386_Packages
MD5: 14e601bd8c0a0905d238d89be3036fa8
Description Language: en
File: /var/lib/apt/lists/us.archive.ubuntu.com_ubuntu_dists_precise_main_i18n_Translation-en
MD5: 14e601bd8c0a0905d238d89be3036fa8

0.99ubuntu13 (/var/lib/apt/lists/us.archive.ubuntu.com_ubuntu_dists_precise_main_binary-i386_Packages)
Description Language:
File: /var/lib/apt/lists/us.archive.ubuntu.com_ubuntu_dists_precise_main_binary-i386_Packages
MD5: 14e601bd8c0a0905d238d89be3036fa8
Description Language: en
File: /var/lib/apt/lists/us.archive.ubuntu.com_ubuntu_dists_precise_main_i18n_Translation-en
MD5: 14e601bd8c0a0905d238d89be3036fa8

Ok, so this was getting easier. All I have to do is install the older version. I had to look it up since it generally doesn’t happen too often:
apt-get install initramfs-tools-bin=0.99ubuntu13

The following packages will be DOWNGRADED:
initramfs-tools-bin
0 upgraded, 0 newly installed, 1 downgraded, 0 to remove and 1 not upgraded.

From there I run the following:
apt-get dist-upgrade
apt-get autoremove
apt-get autoclean
reboot

Profit.

Generally when I give a debian/ubuntu server to developers or other end users, I install webmin and teach them how to use it. Webmin is a great tool for those who either don’t like command line or are just not up to speed with the latest commands to use. Unfortunately a developer installed jira confluence and it broke the SSL settings on apache. Instead of “fixing” the problem, I just went and did it the lazy and easy way: disable ssl for webmin.

SSH to the server as root (or use sudo in front)
nano /etc/webmin/miniserv.conf
Change ssl=1 to ssl=0, save and quit nano
/etc/init.d/webmin restart
Profit!

I had an ubuntu server with 8.04LTS upgraded to 10.04LTS (not quite making it to 12.04, but that’s in the works) with webmin, LAMP, postfix, etc. I used webmin to install dovecot after I already configured postfix to receive mail without requiring SSL with plaintext. It’s an internal only server used to testing purposes, so I don’t mind if someone is sniffing the test passwords that only apply to this box.

Anyway, I was receiving the following in my /var/log/mail.err

Oct 17 08:35:33 SERVERNAME dovecot: POP3(USERNAME): mail_location not set and autodetection failed: Mail storage autodetection failed with home=/home/USERNAME
Oct 17 08:35:33 SERVERNAME dovecot: POP3(USERNAME): Fatal: Namespace initialization failed

Great. Brief search tells me that dovecot’s automatic detection failed (duh) and the fix is to enable the mail location in the configuration file
nano /etc/dovecot/dovecot.conf
Find #mail_location
Assuming it’s commented out, I just added the following below the line:
mail_location = maildir:~/Maildir
Save and close
/etc/init.d/dovecot restart

Recently had to start monitoring a confluence installation on a virtual server using Nagios. Nagios’s built-in check_http looked like it could solve my woes. Unfortunately this was a slight pain due to the fact that confluence uses a non-standard port. So I had to relearn the commands.

I edited the service check first
nano /usr/local/nagios/etc/objects/commands.cfg
I copied the check_http command into a new one:

define command {
command_name check_http_port
command_line $USER1$/check_http -I $ARG1$ -u $ARG2$ -p $ARG3$ -s $ARG4$
}

So the command would be ./check_http -I (IP/Hostname) -u (URL Full Address) -p (Port) -s (String to look for). Technically this will already warn if there is a 400 or 500 error, but I also wanted to verify that the string could be found.

Now we edit the configuration of the system itself
nano /usr/local/nagios/etc/objects/confluence.cfg

define host{
use linux-server
host_name confluence
alias confluence
address 10.555.555.555
parents parent1, switch1, mfer1
hostgroups linux-production-servers, datacenter2
}

define service{
use generic-service
host_name confluence
service_description HTTP
check_command check_http_port!confluence!http://confluence:8090/login.action?os_destination=%2Fhomepage.action!8090!”Remember me”
}

Using the above check_command, you’re able to connect to the server and port listed above and check for the familiar login “remember me” string.