Linux daemons and services

Services and Daemons — runit

Void uses the runit(8) supervision suite to run system services and daemons.

Some advantages of using runit include:

• a small code base, making it easier to audit for bugs and security issues.
• each service is given a clean process state, regardless of how the service was started or restarted: it will be started with the same environment, resource limits, open file descriptors, and controlling terminals.
• a reliable logging facility for services, where the log service stays up as long as the relevant service is running and possibly writing to the log.

If you don’t need a program to be running constantly, but would like it to run at regular intervals, you might like to consider using a cron daemon.

Section Contents

Service Directories

Each service managed by runit has an associated service directory.

A service directory requires only one file: an executable named run , which is expected to exec a process in the foreground.

Optionally, a service directory may contain:

• an executable named check , which will be run to check whether the service is up and available; it’s considered available if check exits with 0.
• an executable named finish , which will be run on shutdown/process stop.
• a conf file; this can contain environment variables to be sourced and referenced in run .
• a directory named log ; a pipe will be opened from the output of the run process in the service directory to the input of the run process in the log directory.

When a new service is created, a supervise folder will be automatically created on the first run.

Configuring Services

Most services can take configuration options set by a conf file in the service directory. This allows service customization without modifying the service directory provided by the relevant package.

Check the service file for how to pass configuration parameters. A few services have a field like OPTS=»—value . » in their conf file.

To make more complex customizations, you should edit the service.

Editing Services

To edit a service, first copy its service directory to a different directory name. Otherwise, xbps-install(1) can overwrite the service directory. Then, edit the new service file as needed. Finally, the old service should be stopped and disabled, and the new one should be started.

Managing Services

Runsvdirs

A runsvdir is a directory in /etc/runit/runsvdir containing enabled services in the form of symlinks to service directories. On a running system, the current runsvdir is accessible via the /var/service symlink.

The runit-void package comes with two runsvdirs, single and default :

• single just runs sulogin(8) and the necessary steps to rescue your system.
• default is the default runsvdir on a running system, unless specified otherwise by the kernel command line.

Additional runsvdirs can be created in /etc/runit/runsvdir/ .

See runsvdir(8) and runsvchdir(8) for further information.

Booting A Different runsvdir

To boot a runsvdir other than default , the name of the desired runsvdir can be added to the kernel command-line. As an example, adding single to the kernel command line will boot the single runsvdir.

Basic Usage

To start, stop, restart or get the status of a service:

The placeholder can be:

• Service names (service directory names) inside the /var/service/ directory.
• The full paths to the services.

For example, the following commands show the status of a specific service and of all enabled services:

See sv(8) for further information.

Enabling Services

Void Linux provides service directories for most daemons in /etc/sv/ .

To enable a service on a booted system, create a symlink to the service directory in /var/service/ :

If the system is not currently running, the service can be linked directly into the default runsvdir:

This will automatically start the service. Once a service is linked it will always start on boot and restart if it stops, unless administratively downed.

To prevent a service from starting at boot while allowing runit to manage it, create a file named down in its service directory:

The down file mechanism also makes it possible to disable services that are enabled by default, such as the agetty(8) services for ttys 1 to 6. This way, package updates which affect these services

• in this case, the runit-void package — won’t re-enable them.

Disabling Services

To disable a service, remove the symlink from the running runsvdir:

Or, for example, from the default runsvdir, if either the specific runsvdir, or the system, is not currently running:

Testing Services

To check if a service is working correctly when started by the service supervisor, run it once before fully enabling it:

If everything works, remove the down file to enable the service.

Источник

Daemons and Services

Contents

What are daemons, services and agents?

Unix daemons and Windows services are system-wide programs running without user interaction; macOS agents are per user programs (cf system-wide daemons) that may or may not run without user interaction. Although the nomenclature differs, their function is similar: for example www or ftp servers are called daemons under Linux and services under Windows. Because they do not interact with the user directly, they close their stdin, stdout, stderr descriptors at start.

With Free Pascal, Lazarus it is possible to write these daemons/services platform-independent via the Lazarus lazdaemon package. To avoid name conflicts with the Delphi components these classes are called ‘daemons’.

Install LazDaemon

Before you can start, install the lazdaemon package. Either via Components / Configure installed packages or by opening/installing the lpk file directly: lazarus/components/daemon/lazdaemon.lpk. This package installs some new components and menu items in the IDE:

Under File — New: 3 items appear in the dialog, under the heading: «Daemon(service) applications»:

Classes

TCustomDaemon

This is a TDataModule descendant that does all the work. There can be several TCustomDaemons classes and/or instances running at the same time and in the same process (multi threaded).

TCustomDaemonApplication

This is a TCustomApplication descendant which creates the TCustomDaemons. This does not need any change. It runs under Windows until it receives the Stop call or under Linux until the TERM signal.

TDaemonMapper

This component handles the service registration. Each instance needs one entry in the property DaemonDefs.

Daemon — Step by Step

• When the daemon is started the command line parameters are parsed. The following are predefined:

• • -i —install: register the daemon. This has no effect under unix.
  • -u —uninstall: unregister the daemon. This has no effect under unix.
  • -r —run: start the daemon. Windows does this normally itself.

• Create the TDaemonMapper
• Create one TCustomDaemon for each entry of DaemonDefs.
• install, uninstall or run every instance.
• if run: start every instance in its own thread and then wait for Stop/TERM signal.

Daemon Methods

Getting Started

Before you are able to create a Service or Daemon application you must first ensure that the Lazarus Daemon package «lazdaemon» is installed.

Example

There is a simple example in examples/cleandir/. Read the README.txt.

Service Installation

Windows

You can install the service by executing the process with the Install parameter. Windows service manager will do the rest for you. You can configure the service and its options from the service manager.

System codepage / UTF-8

A LazDeamon project is working with default, not UTF-8, codepage. The -dDisableUTF8RTL mode has to be activated with Project Options . -> Compiler Options -> Additions and Overrides -> Use system encoding.

Linux (only for older Debian)

Download, configure, and «Save As» — the sample script located at Web Archive: [1] (The original link is dead for a long time).

• SVC_ALIAS is the long description of your application
• SVC_FILENAME is the actual file name of your compiled service application
• SVC_DIR is the place your you copied the service application
• SVC_SERVICE_SCRIPT is the final name of the service.sh when you «Save As» the customized debian-service.sh script.

Place your script in the /etc/init.d/ folder

start the service by running «sudo service Name_Of_Your_Script start»

Note: sudo has some variations, e.g.:
sudo -s #
sudo -H #
sudo -i #
sudo su #
sudo sh #

In order to auto-run the service at startup you can try update-rc.d or else will need a third party tool that will do this.

systemd (Fedora, Debian, SLES12)

Presently, linux flavors are trending away from differing daemon launching and into a unified service model.

Fedora and SuSE Enterprise Linux Server 12 use systemd and with that commands to start/stop services are the same as on debian but there are differences on the configuration files.

• From the command prompt sudo gedit
• Copy and Paste

Источник

Linux Jargon Buster: What are Daemons in Linux?

Last updated June 5, 2021 By Bill Dyer 29 Comments

Daemons work hard so you don’t have to.

Imagine that you are writing an article, Web page, or book, Your intent is to do just that – write. It’s rather nice not having to manually start printer and network services and then monitor them all day to make sure that they are working right.

We can thank daemons for that – they do that kind of work for us.

What is a Daemon in Linux?

A daemon (usually pronounced as: day-mon , but sometimes pronounced as to rhyme with diamond ) is a program with a unique purpose. They are utility programs that run silently in the background to monitor and take care of certain subsystems to ensure that the operating system runs properly. A printer daemon monitors and takes care of printing services. A network daemon monitors and maintains network communications, and so on.

Having gone over the pronunciation of daemon, I’ll add that, if you want to pronounce it as demon, I won’t complain.

For those people coming to Linux from the Windows world, daemons are known as services. For Mac users, the term, services, has a different use. The Mac’s operating system is really UNIX, so it uses daemons. The term, services is used, but only to label software found under the Services menu.

Daemons perform certain actions at predefined times or in response to certain events. There are many daemons that run on a Linux system, each specifically designed to watch over its own little piece of the system, and because they are not under the direct control of a user, they are effectively invisible, but essential. Because daemons do the bulk of their work in the background, they can appear a little mysterious and so, perhaps difficult to identify them and what they actually do.

What Daemons are Running on Your Machine?

To identify a daemon, look for a process that ends with the letter d. It’s a general Linux rule that the names of daemons end this way.

There are many ways to catch a glimpse of a running daemon. They can be seen in process listings through ps , top , or htop . These are useful programs in their own right – they have a specific purpose, but to see all of the daemons running on your machine, the pstree command will suit our discussion better.

The pstree command is a handy little utility that shows the processes currently running on your system and it show them in a tree diagram. Open up a terminal and type in this command:

You will see a complete listing of all of the processes that are running. You may not know what some of them are, or what they do, they are listed. The pstree output is a pretty good illustration as to what is going on with your machine. There’s a lot going on!

Looking at the screen shot, a few daemons can be seen here: udisksd, gvfsd, systemd, logind and some others.

Our process list was long enough to where the listing couldn’t fit in a single terminal window, but we can scroll up using the mouse or cursor keys:

Spawning Daemons

Again, a daemon is a process that runs in the background and is usually out of the control of the user. It is said that a daemon has no controlling terminal.

A process is a running program. At a particular instant of time, it can be either running, sleeping, or zombie (a process that completed its task, but waiting for its parent process to accept the return value).

In Linux, there are three types of processes: interactive, batch and daemon.

Interactive processes are those which are run by a user at the command line are called interactive processes.

Batch processes are processes that are not associated with the command line and are presented from a list of processes. Think of these as “groups of tasks”. These are best at times when the system usage is low. System backups, for example, are usually run at night since the daytime workers aren’t using the system. When I was a full-time system administrator, I often ran disk usage inventories, system behavior analysis scripts, and so on, at night.

Interactive processes and batch jobs are not daemons even though they can be run in the background and can do some monitoring work. They key is that these two types of processes involve human input through some sort of terminal control. Daemons do not need a person to start them up.

We know that a daemon is a computer program that runs as a background process, rather than being under the direct control of an interactive user. When the system boot is complete, the system initialization process starts spawning (creating) daemons through a method called forking, eliminating the need for a terminal (this is what is meant by no controlling terminal).

I will not go into the full details of process forking, but hopefully, I can be just brief enough to show a little background information to describe what is done. While there are other methods to create processes, traditionally, in Linux, the way to create a process is through making a copy of an existing process in order to create a child process. An exec system call to start another program in then performed.

The term, fork isn’t arbitrary, by the way. It gets its name from the C programming language. One of the libraries that C uses, is called the standard library, containing methods to perform operating services. One of these methods, called fork, is dedicated to creating new processes. The process that initiates a fork is considered to be the parent process of the newly created child process.

The process that creates daemons is the initialization (called init ) process by forking its own process to create new ones. Done this way, the init process is the outright parent process.

There is another way to spawn a daemon and that is for another process to fork a child process and then die (a term often used in place of exit). When the parent dies, the child process becomes an orphan. When a child process is orphaned, it is adopted by the init process.

If you overhear discussions, or read online material, about daemons having “a parent process ID of 1,” this is why. Some daemons aren’t spawned at boot time, but are created later by another process which died, and init adopted it.

It is important that you do not confuse this with a zombie. Remember, a zombie is a child process that has finished its task and is waiting on the parent to accept the exit status.

Examples of Linux Daemons

Again, the most common way to identify a Linux daemon is to look for a service that ends with the letter d. Here are some examples of daemons that may be running on your system. You will be able to see that daemons are created to perform a specific set of tasks:

systemd – the main purpose of this daemon is to unify service configuration and behavior across Linux distributions.

rsyslogd – used to log system messages. This is a newer version of syslogd having several additional features. It supports logging on local systems as well as on remote systems.

udisksd – handles operations such as querying, mounting, unmounting, formatting, or detaching storage devices such as hard disks or USB thumb drives

logind – a tiny daemon that manages user logins and seats in various ways

httpd – the HTTP service manager. This is normally run with Web server software such as Apache.

sshd – Daemon responsible for managing the SSH service. This is used on virtually any server that accepts SSH connections.

ftpd – manages the FTP service – FTP or File Transfer Protocol is a commonly-used protocol for transferring files between computers; one act as a client, the other act as a server.

crond – the scheduler daemon for time-based actions such as software updates or system checks.

What is the origin of the word, daemon?

When I first started writing this article, I planned to only cover what a daemon is and leave it at that. I worked with UNIX before Linux appeared. Back then, I thought of a daemon as it was: a background process that performed system tasks. I really didn’t care how it got its name. With additional talk of other things, like zombies and orphans, I just figured that the creators of the operating system had a warped sense of humor (a lot like my own).

I always perform some research on every piece that I write and I was surprised to learn that apparently, a lot of other people did want to know how the word came to be and why.

The word has certainly generated a bit of curiosity and, after reading through several lively exchanges, I admit that I got curious too. Perform a search on the word’s meaning or etymology (the origin of words) and you’ll find several answers.

In the interest of contributing to the discussion, here’s my take on it.

The earliest form of the word, daemon, was spelled as daimon, a form of guardian angel – attendant spirits that helped form the character of people they assisted. Socrates claimed to have one that served him in a limited way, but correctly. Socrates’ daimon only told him when to keep his mouth shut. Socrates described his daimon during his trial in 399 BC, so the belief in daimons has been around for quite some time. Sometimes, the spelling of daimon is shown as daemon. Daimon and daemon, here, mean the same thing.

While a daemon is an attendant, a demon is an evil character from the Bible. The differences in spelling is intentional and was apparently decided upon in the 16th century. Daemons are the good guys, and demons are the bad ones.

The use of the word, daemon, in computing came about in 1963. Project MAC is shorthand for Project on Mathematics and Computation, and was created at the Massachusetts Institute of Technology. It was here that the word, daemon, came into common use to mean any system process that monitors other tasks and performs predetermined actions depending on their behavior, The word, daemon was named for Maxwell’s daemon.

Maxwell’s daemon is the result of a thought experiment. In 1871, James Clerk Maxwell imagined an intelligent and resourceful being that was able to observe and direct the travel of individual molecules in a specific direction. The purpose of the thought exercise was to show the possibility of contradicting the second law of thermodynamics.

I did see some comments that the word, daemon, was an acronym for Disk And Executive MONitor . The original users of the word, daemon, never used it for that purpose, so the acronym idea, I believe, is incorrect.

Lastly – to end this on a light note – there is the BSD mascot: a daemon that has the appearance of a demon. The BSD daemon was named after the software daemons, but gets is appearance from playing around with the word.

The daemon’s name is Beastie. I haven’t researched this fully (yet), but I did find one comment that states that Beastie comes from slurring the letters, BSD. Try it; I did. Say the letters as fast as you can and out comes a sound very much like beastie.

Beastie is often seen with a trident which is symbolic of a daemon’s forking of processes.

Like what you read? Please share it with others.

Источник