If a Linux system has been running for a long time, this is definitely proof of its stability, but – depending on the distribution – some updates might be waiting to install. Conversely, if the system reboots very frequently, there may be a configuration error – or maybe a hardware component is slowly deteriorating. On a workstation computer, such restarts are quickly noticed, but not necessarily on a remote server that is running quietly and well away from the action.

How long a system has been running continuously can be determined at the command line by a call to uptime. But you might also want to try tuptime [1], a similar tool whose name is based on a contraction of "total uptime." It outputs far more information, including valuable information in the form of the number of reboots and the kernel version used.

Installation

While uptime is available on almost every system, you need to install tuptime separately. Some distributions already have the tool in their repositories, such as Debian and Ubuntu. Arch Linux users will find tuptime in the AUR and CentOS users in the EPEL repository. On any Linux system, you can install the tool with the following:

$ curl -Ls https://git.io/tuptime-install.sh | sudo bash

Tuptime itself consists of a Python script and requires Python 3 with the modules sys, os, optparse, sqlite3, locale, platform, datetime, and logging. These modules should be preinstalled on most distributions. For more information about installation, see the box "In the Background."

Time Measurement

After setting up the program, query the current status via tuptime in the terminal. The tool already provides some basic information (Figure 1): At the very top; it shows the number of system starts. Next to it you can see the time and date since the software started counting the starts. Tuptime doesn't capture events before this date. To get a complete picture, set up the program as soon as possible after system installation.

Figure 1: Shortly after the install, tuptime cannot give you very much information, but over time, the data starts to give you a better impression of the computer's behavior.

Next, tuptime tells you how often the system has been shut down (System shutdowns). Pay special attention to the number of uncontrolled shutdowns labeled bad. If this figure increases rapidly, there is a major problem, for example an unstable power supply. System life indicates how long tuptime has been monitoring the system. If you installed tuptime directly after the installation of the distribution, this value corresponds to the time since the first boot.

This is followed by the total system uptime (System uptime) and the total time the computer was powered off (System downtime). If the percentage of downtime exceeds 50 percent, the computer was off longer than in operation. How long the system was in operation on average is indicated by the value following Average uptime. Similarly, Average downtime indicates how long the computer was switched off on average. The final value is the time since the last system start (Current uptime).

Display Options

The tuptime -l command returns a list with all startup operations. In the list, you can see in detail when the system was in operation and for how long. If the list seems too confusing, tuptime -t converts the data to tabular form like shown in Figure 2. The additional -r parameter reverses the order; the last system start appears at the top.

Figure 2: Tuptime offers different display formats. There has been a reboot since the tool was installed, so two system starts have been counted.

If you still feel overwhelmed by the volume of information, the display can be limited to a certain period of time. For example, if you want to know when and for how long the system was active between July 27 and July 28, 2020, just call tuptime with the command:

$ tuptime --tsince=$(date --date="2020-07-27" +%s) --tuntil=$(date --date="2020-07-28" +%s)

The --tsince and --tuntil parameters each expect a timestamp that is calculated by date. Tuptime also outputs these timestamps with the -s parameter.

Alternatively, you can limit the output to very specific startup operations. For example, if you are interested in how long the system ran from the very first to the fifth logged startup, use the following:

$ tuptime --since 1 --until 5