The No Hang Daemon

Nohang is a highly configurable daemon for Linux which is able to correctly prevent out of memory conditions.

What is the problem?

OOM killer doesn't prevent OOM conditions.

Solutions

• Use of earlyoom. This is a simple OOM preventer written in C.
• Use of nohang. This is an advanced OOM preventer written in Python.

Some features

• convenient configuration with a well commented config file (there are 35 parameters in the config)
• SIGKILL and SIGTERM as signals that can be sent to the victim
• zram support (mem_used_total as a trigger)
• customizable intensity of monitoring
• desktop notifications: results of preventings OOM and low memory warnings
• prefer and avoid lists via regex matching
• possibility of restarting processes via command like systemctl restart something if the process is selected as a victim
• look at the config to find more

Demo

Video: nohang prevents OOM after the command while true; do tail /dev/zero; done has been executed.

An example of output

...if command while true; do stress -m 9 --vm-bytes 3G; done has been executed with rate_mem = 9:

MemAvail: 2515 M, 42.8 %
MemAvail: 1510 M, 25.7 %
MemAvail:  909 M, 15.5 %
MemAvail:  520 M,  8.9 %
* MemAvailable (520 MiB, 8.9 %) < mem_min_sigterm (529 MiB, 9.0 %)
  SwapFree (0 MiB, 0.0 %) < swap_min_sigterm (0 MiB, - %)
  Preventing OOM: trying to send the SIGTERM signal to stress,
  Pid: 9828, Badness: 82, VmRSS: 485 MiB, VmSwap: 0 MiB
  Success; reaction time: 11 ms
MemAvail: 4114 M, 70.0 %
MemAvail: 2532 M, 43.1 %
MemAvail: 1495 M, 25.4 %
MemAvail:  927 M, 15.8 %
MemAvail:  553 M,  9.4 %
MemAvail:  342 M,  5.8 %
* MemAvailable (342 MiB, 5.8 %) < mem_min_sigkill (353 MiB, 6.0 %)
  SwapFree (0 MiB, 0.0 %) < swap_min_sigkill (0 MiB, - %)
  Preventing OOM: trying to send the SIGKILL signal to stress,
  Pid: 9841, Badness: 87, VmRSS: 513 MiB, VmSwap: 0 MiB
  Success; reaction time: 11 ms
MemAvail: 4084 M, 69.5 %
MemAvail: 2543 M, 43.3 %
MemAvail: 1535 M, 26.1 %

And demo: https://youtu.be/5d6UovJzK8k

Requirements

• Linux 3.14+ (because the MemAvailable parameter appeared in /proc/meminfo since kernel version 3.14) and Python 3.4+ (compatibility with earlier versions was not tested) for basic usage
• libnotify (Fedora, Arch) or libnotify-bin (Debian, Ubuntu) for desktop notifications and sudo for desktop notifications as root

Memory and CPU usage

• VmRSS is 10 — 13.5 MiB depending on the settings
• CPU usage depends on the level of available memory (the frequency of memory status checks increases as the amount of available memory decreases) and monitoring intensity (can be changed by user via config)

Status

The program is unstable and some fixes are required before the first stable version will be released (need documentation, translation, review and some optimisation).

Download

git clone https://github.com/hakavlad/nohang.git
cd nohang

Installation and start for systemd users

sudo ./install.sh

Purge

sudo ./purge.sh

Command line options

./nohang -h
usage: nohang [-h] [-c CONFIG]

optional arguments:
  -h, --help            show this help message and exit
  -c CONFIG, --config CONFIG
                        path to the config file, default values:
                        ./nohang.conf, /etc/nohang/nohang.conf

How to configure nohang

The program can be configured by editing the config file. The configuration includes the following sections:

• Thresholds for sending signals to victims
• Intensity of monitoring (and CPU usage)
• Prevention of killing innocent victims
• Avoid and prefer victim names via regex matching
• Execute the command instead of sending the SIGTERM signal
• GUI notifications: results of preventing OOM and low memory warnings
• Self-defense and preventing slowing down the program
• Output verbosity

Just read the description of the parameters and edit the values. Please restart nohang to apply changes. Default path to the config arter installing via ./install.sh is /etc/nohang/nohang.conf.

Feedback

Please create issues. Use cases, feature requests and any questions are welcome.