25

Let me start by saying I have been forbidden to enable automatic updates on our Ubuntu servers, for both security and regular packages.

When I log into any of my four Ubuntu servers, the welcome message contains this:

39 packages can be updated.
26 updates are security updates.

However, when I run the Nagios plugin that monitors APT, I get:

% /usr/lib/nagios/plugins/check_apt
APT WARNING: 33 packages available for upgrade (0 critical updates). 

I need to know how to properly detect that there are pending security updates, and regular updates. Once I can do that, I plan to write a Nagios script that will return WARNING for pending regular updates, and CRITICAL for pending security updates.

Anyone know how to detect those two conditions?

12

The Nagios plugin /usr/lib/nagios/plugins/check_apt does not detect critical updates in Ubuntu correctly due how it detects critical updates via apt combined with how Ubuntu non-critical updates are published. More details are in the bug here: https://bugs.launchpad.net/bugs/1031680

Using /usr/lib/update-notifier/apt-check instead is a reliable workaround.

31

Turns out that the number of pending regular updates can be found using:

/usr/lib/update-notifier/apt-check 2>&1 | cut -d ';' -f 1

And the number of pending security updates can be found using:

/usr/lib/update-notifier/apt-check 2>&1 | cut -d ';' -f 2

In the end, my Nagios plugin was as follows:

#!/bin/sh
#
# Standard Nagios plugin return codes.
STATUS_OK=0
STATUS_WARNING=1
STATUS_CRITICAL=2
STATUS_UNKNOWN=3

# Query pending updates.
updates=$(/usr/lib/update-notifier/apt-check 2>&1)
if [ $? -ne 0 ]; then
    echo "Querying pending updates failed."
    exit $STATUS_UNKNOWN
fi

# Check for the case where there are no updates.
if [ "$updates" = "0;0" ]; then
    echo "All packages are up-to-date."
    exit $STATUS_OK
fi

# Check for pending security updates.
pending=$(echo "${updates}" | cut -d ";" -f 2)
if [ "$pending" != "0" ]; then
    echo "${pending} security update(s) pending."
    exit $STATUS_CRITICAL
fi

# Check for pending non-security updates.
pending=$(echo "${updates}" | cut -d ";" -f 1)
if [ "$pending" != "0" ]; then
    echo "${pending} non-security update(s) pending."
    exit $STATUS_WARNING
fi

# If we've gotten here, we did something wrong since our "0;0" check should have
# matched at the very least.
echo "Script failed, manual intervention required."
exit $STATUS_UNKNOWN
1

Why not simply using the apt-get command?:

apt-get -s dist-upgrade | grep "^Inst" | grep -i security | wc -l
  • 2
    This hack will not reliably distinguish between security and non-security updates. For example, on Ubuntu, security updates are also published to the updates pocket. If the updates pocket is listed first in sources.list, your suggestion will lead to missing security update notifications. apt will choose to download them from the updates pocket instead, and so your grep will miss them. – Robie Basak Jan 25 '16 at 1:04
  • The problem identified by @RobieBasak can be corrected as per my answer at serverfault.com/a/856769/134053 – mc0e Jun 20 '17 at 13:13
0

Once Nagios has reported you have security updates, this is how you get a list of which ones are needed.

grep security /etc/apt/sources.list > /tmp/security.list
sudo apt-get upgrade -oDir::Etc::Sourcelist=/tmp/security.list -s

You could also use these commands piped into wc -l to give you a count, but the answers above are probably more efficient and appropriate for a Nagios script.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.