Supervisor: A System for Allowing the Control of Process State on UNIX
7/3/2006: updated for version 2.0
8/30/2006: updated for version 2.1
3/31/2007: updated for version 2.2
8/15/2007: updated for version 3.0
If you are upgrading from supervisor version 2.X to version 3.X, and you wish to preserve your supervisor configuration file, you will need to read the file named 'UPGRADING.txt' within the same directory as this file. Some configuration file options have changed and new ones have been added.
The supervisor is a client/server system that allows its users to control a number of processes on UNIX-like operating systems. It was inspired by the following:
- It is often inconvenient to need to write "rc.d" scripts for every single process instance. rc.d scripts are a great lowest-common-denominator form of process initialization/autostart/management, but they can be painful to write and maintain. Additionally, rc.d scripts cannot automatically restart a crashed process and many programs do not restart themselves properly on a crash. Supervisord starts processes as its subprocesses, and can be configured to automatically restart them on a crash. It can also automatically be configured to start processes on its own invocation.
- It's often difficult to get accurate up/down status on processes on UNIX. Pidfiles often lie. Supervisord starts processes as subprocesses, so it always knows the true up/down status of its children and can be queried conveniently for this data.
- Users who need to control process state often need only to do that. They don't want or need full-blown shell access to the machine on which the processes are running. Supervisorctl allows a very limited form of access to the machine, essentially allowing users to see process status and control supervisord-controlled subprocesses by emitting "stop", "start", and "restart" commands from a simple shell or web UI.
- Users often need to control processes on many machines. Supervisor provides a simple, secure, and uniform mechanism for interactively and automatically controlling processes on groups of machines.
- Processes which listen on "low" TCP ports often need to be started and restarted as the root user (a UNIX misfeature). It's usually the case that it's perfectly fine to allow "normal" people to stop or restart such a process, but providing them with shell access is often impractical, and providing them with root access or sudo access is often impossible. It's also (rightly) difficult to explain to them why this problem exists. If supervisord is started as root, it is possible to allow "normal" users to control such processes without needing to explain the intricacies of the problem to them.
- Processes often need to be started and stopped in groups, sometimes even in a "priority order". It's often difficult to explain to people how to do this. Supervisor allows you to assign priorities to processes, and allows user to emit commands via the supervisorctl client like "start all", and "restart all", which starts them in the preassigned priority order. Additionally, processes can be grouped into "process groups" and a set of logically related processes can be stopped and started as a unit.
Supervisor has been tested and is known to run on Linux (Fedora Core 5, Ubuntu 6), Mac OS X (10.4), and Solaris (10 for Intel) and FreeBSD 6.1. It will likely work fine on most UNIX systems.
Supervisor will not run at all under any version of Windows.
Supervisor requires Python 2.3 or better.
Run "python setup.py install". This will download and install all distributions depended upon by supervisor and finally install supervisor itself. Once that's done, copy the "sample.conf" file you'll find in the same directory as this file to /etc/supervisord.conf and modify to your liking. If you'd rather not put the supervisord.conf file in /etc, you can place it anywhere and start supervisord and point it at the configuration file via the -c flag, e.g. "python supervisord.py -c /path/to/sample/conf" or, if you use the shell script named "supervisord", "supervisord -c /path/to/sample.conf".
I make reference below to a "$BINDIR" when explaining how to run supervisord and supervisorctl. This is the "bindir" directory that your Python installation has been configured with. For example, for an installation of Python installed via "./configure --prefix=/usr/local/python; make; make install", $BINDIR would be "/usr/local/python/bin". Python interpreters on different platforms use different $BINDIRs. Look at the output of "setup.py install" if you can't figure out where yours is.
Installing Without Internet Access
Since "setup.py install" performs downloads of dependent software, it will not work on machines without internet access. To install to a machine which is not internet connected, obtain the following dependencies on a machine which is internet-connected:
- setuptools (latest) from http://pypi.python.org/pypi/setuptools
- meld3 (0.6) from http://www.plope.com/software/meld3/
- medusa (0.5.4) from http://www.amk.ca/python/code/medusa.html
- elementtree (1.2.6) from http://effbot.org/downloads#elementtree
And then copy these files to removable media and put them on the target machine. Install each onto the target machine as per its instructions.
Note -- if the machine you're installing on does not have a C compiler, meld3's "setup.py install" probably won't work because meld3 uses C extensions, but you can either copy the meld3/meld3 directory into your Python's site-packages directory, or you can build a binary distribution for your platform on a similar machine that does have a C compiler before shipping it over by doing "python setup.py bdist".
Finally, run supervisor's "python setup.py install".
To start supervisord, run $BINDIR/supervisord. The resulting process will daemonize itself and detach from the terminal. It keeps an operations log at "/tmp/supervisor.log" by default.
You can start supervisord in the foreground by passing the "-n" flag on its command line. This is useful to debug startup problems.
To change the set of programs controlled by supervisord, edit the supervisord.conf file and kill -HUP or otherwise restart the supervisord process. This file has several example program definitions.
Supervisord accepts a number of command-line overrides. Type 'supervisord -h' for an overview.
To start supervisorctl, run $BINDIR/supervisorctl. A shell will be presented that will allow you to control the processes that are currently managed by supervisord. Type "help" at the prompt to get information about the supported commands.
supervisorctl may be invoked with "one time" commands when invoked with arguments from a command line. An example: "supervisorctl stop all". If arguments are present on the supervisorctl command-line, it will prevent the interactive shell from being invoked. Instead, the command will be executed and supervisorctl will exit.
If supervisorctl is invoked in interactive mode against a supervisord that requires authentication, you will be asked for authentication credentials.
The server piece of the supervisor is named "supervisord". It is responsible for responding to commands from the client process as well as restarting crashed or exited processes. It is meant to be run as the root user in most production setups. NOTE: see "Security Notes" at the end of this document for caveats!
The server process uses a configuration file. This is typically located in "/etc/supervisord.conf". This configuration file is an "Windows-INI" style config file. It is important to keep this file secure via proper filesystem permissions because it may contain unencrypted usernames and passwords.
The command-line client piece of the supervisor is named "supervisorctl". It provides a shell-like interface to the features provided by supervisord. From supervisorctl, a user can connect to different supervisord processes, get status on the subprocesses controlled by a supervisord, stop and start subprocesses of a supervisord, and get lists of running processes of a supervisord.
The command-line client talks to the server across a UNIX domain socket or an Internet socket. The server can assert that the user of a client should present authentication credentials before it allows him to perform commands. The client process may use the same configuration file as the server; any configuration file with a [supervisorctl] section in it will work.
Web ServerA (sparse) web user interface with functionality comparable to supervisorctl may be accessed via a browser if you start supervisord against an internet socket. Visit the server URL (e.g. http://localhost:9001/) to view and control process status through the web interface after changing the configuration file's 'http_port' parameter appropriately.
The same HTTP server which serves the web UI serves up an XML-RPC interface that can be used to interrogate and control supervisor and the programs it runs. To use the XML-RPC interface, connect to supervisor's http port with any XML-RPC client library and run commands against it. An example of doing this using Python's xmlrpclib client library:import xmlrpclib server = xmlrpclib.Server('http://localhost:9001')
Call methods against the supervisor and its subprocesses by using the 'supervisor' namespace:server.supervisor.getState()
You can get a list of methods supported by supervisor's XML-RPC interface by using the XML-RPC 'system.listMethods' API:server.system.listMethods()
You can see help on a method by using the 'system.methodHelp' API against the method:print server.system.methodHelp('supervisor.shutdown')
Supervisor's XML-RPC interface also supports the nascent XML-RPC multicall API described at http://www.xmlrpc.com/discuss/msgReader$1208.
You can extend supervisor functionality with new XML-RPC API methods by adding new top-level RPC interfaces as necessary. See "Configuration File ['rpcinterface:x] Section Settings" in this file.
Configuration File '[supervisord]' Section Settings
The supervisord.conf log file contains a section named '[supervisord]' in which global settings for the supervisord process should be inserted. These are:
'http_port' -- Either a TCP host:port value or (e.g. 127.0.0.1:9001) or a path to a UNIX domain socket (e.g. /tmp/supervisord.sock) on which supervisor will listen for HTTP/XML-RPC requests. Supervisorctl itself uses XML-RPC to communicate with supervisord over this port.
'sockchmod' -- Change the UNIX permission mode bits of the http_port UNIX domain socket to this value (ignored if using a TCP socket). Default: 0700.
'sockchown' -- Change the user and group of the socket file to this value. May be a username (e.g. chrism) or a username and group separated by a dot (e.g. chrism.wheel) Default: do not change.
'umask' -- The umask of the supervisord process. Default: 022.
'logfile' -- The path to the activity log of the supervisord process.
'logfile_maxbytes' -- The maximum number of bytes that may be consumed by the activity log file before it is rotated (suffix multipliers like "KB", "MB", and "GB" can be used in the value). Set this value to 0 to indicate an unlimited log size. Default: 50MB.
'logfile_backups' -- The number of backups to keep around resulting from activity log file rotation. Set this to 0 to indicate an unlimited number of backups. Default: 10.
'loglevel' -- The logging level, dictating what is written to the activity log. One of 'critical', 'error', 'warn', 'info', 'debug' or 'trace'. Note that at log level 'trace', the supervisord log file will record the stderr/stdout output of its child processes, which is useful for debugging. Default: info.
'pidfile' -- The location in which supervisord keeps its pid file.
'nodaemon' -- If true, supervisord will start in the foreground instead of daemonizing. Default: false.
'minfds' -- The minimum number of file descriptors that must be available before supervisord will start successfully. Default: 1024.
'minprocs' -- The minimum nymber of process descriptors that must be available before supervisord will start successfully. Default: 200.
'nocleanup' -- prevent supervisord from clearing any existing "AUTO" log files at startup time. Default: false.
'http_username' -- the username required for authentication to our HTTP server. Default: none.
'http_password' -- the password required for authentication to our HTTP server. Default: none.
'childlogdir' -- the directory used for AUTO log files. Default: value of Python's tempfile.get_tempdir().
'user' -- if supervisord is run as root, switch users to this UNIX user account before doing any meaningful processing. This value has no effect if supervisord is not run as root. Default: do not switch users.
'directory' -- When supervisord daemonizes, switch to this directory. Default: do not cd.
'strip_ansi' -- Strip all ANSI escape sequences from process log files.
'environment' -- A list of key/value pairs in the form "KEY=val,KEY2=val2" that will be placed in the supervisord process' environment (and as a result in all of its child process' environments). Default: none. Note that subprocesses will inherit the environment variables of the shell used to start "supervisord" except for the ones overridden here and within the program's "environment" configuration stanza. See "Subprocess Environment" below.
'identifier' -- The identifier for this supervisor server, used by the RPC interface. Default: 'supervisor'.
Configuration File '[supervisorctl]' Section Settings
The configuration file may contain settings for the supervisorctl interactive shell program. These options are listed below.
'serverurl' -- The URL that should be used to access the supervisord server, e.g. "http://localhost:9001". For UNIX domain sockets, use "unix:///absolute/path/to/file.sock".
'username' -- The username to pass to the supervisord server for use in authentication (should be same as 'http_username' in supervisord config). Optional.
'password' -- The password to pass to the supervisord server for use in authentication (should be the same as 'http_password' in supervisord config). Optional.
'prompt' -- String used as supervisorctl prompt. Default: supervisor.
Configuration File '[program:x]' Section Settings
The .INI file must contain one or more 'program' sections in order for supervisord to know which programs it should start and control. A sample program section has the following structure, the options of which are described below it:[program:foo] command=/path/to/foo process_name = %(program_name)s numprocs=1 priority=1 autostart=true autorestart=true startsecs=1 startretries=3 exitcodes=0,2 stopsignal=TERM stopwaitsecs=10 user=nobody redirect_stderr=false stdout_logfile=AUTO stdout_logfile_maxbytes=50MB stdout_logfile_backups=10 stdout_capturefile=NONE stderr_logfile=AUTO stderr_logfile_maxbytes=50MB stderr_logfile_backups=10 stderr_capturefile=NONE environment=A=1,B=2
'[program:foo]' -- the section header, required for each program. 'programname' is a descriptive name (arbitrary) used to describe the program being run. It must not include a colon character or a bracket character.
'command' -- the command that will be run when this program is started. The command can be either absolute, e.g. ('/path/to/programname') or relative ('programname'). If it is relative, the PATH will be searched for the executable. Programs can accept arguments, e.g. ('/path/to/program foo bar'). The command line can used double quotes to group arguments with spaces in them to pass to the program, e.g. ('/path/to/program/name -p "foo bar"'). Note that the value of 'command' may include Python string expressions, e.g. "/path/to/programname --port=80%(process_num)02d" might expand to "/path/to/programname --port=8000" at runtime. String expressions are evaluated against a dictionary containing the keys "group_name", "process_num" and "program_name". Controlled programs should themselves not be daemons, as supervisord assumes it is responsible for daemonizing its subprocesses (see "Nondaemonizing of Subprocesses" later in this document).
'process_name' -- a Python string expression that is used to compose the supervisor process name for this process. You usually don't need to worry about setting this unless you change 'numprocs'. The string expression is evaluated against a dictionary that includes "group_name", "process_num" and "program_name". Default: %(program_name)s. (New in 3.0)
'numprocs' -- Supervisor will start as many instances of this program as named by numprocs. Note that if numprocs > 1, the 'process_name' expression must include '%(process_num)s' (or any other valid Python string expression that includes 'process_num') within it. Default: 1. (New in 3.0)
'priority' -- the relative priority of the program in the start and shutdown ordering. Lower priorities indicate programs that start first and shut down last at startup and when aggregate commands are used in various clients (e.g. "start all"/"stop all"). Higher priorities indicate programs that start last and shut down first. Default: 999.
'autostart' -- If true, this program will start automatically when supervisord is started. Default: true.
'autorestart' -- If true, when the program exits (either expectedly or unexpectedly), supervisor will restart it automatically. Default: true.
'startsecs' -- The total number of seconds which the program needs to stay running after a startup to consider the start successful. If the program does not stay up for this many seconds after it is started, even if it exits with an "expected" exit code (see "exitcodes"), the startup will be considered a failure. Set to 0 to indicate that the program needn't stay running for any particular amount of time. Default: 1
'startretries' -- The number of serial failure attempts that supervisord will allow when attempting to start the program before giving up and puting the process into an ERROR state. Default: 3.
'exitcodes' -- The list of 'expected' exit codes for this program. Supervisor log messages will note if the program exits with an exit code which is not in this list and a stop of the program has not been explicitly requested. Default: 0,2.
'stopsignal' -- The signal used to kill the program when a stop is requested. This can be any of TERM, HUP, INT, QUIT, KILL, USR1, or USR2. Default: TERM.
'stopwaitsecs' -- The number of seconds to wait for the program to return a SIGCHILD to supervisord after the program has been sent a stopsignal. If this number of seconds elapses before supervisord receives a SIGCHILD from the process, supervisord will attempt to kill it with a final SIGKILL. Default: 10.
'user' -- If supervisord is running as root, this UNIX user account will be used as the account which runs the program. If supervisord is not running as root, this option has no effect. Default: do not switch users.
'redirect_stderr' -- If true, cause the process' stderr output to be sent back to supervisor on it's stdout file descriptor (in UNIX shell terms, this is the equivalent of executing "/the/program 2>&1". Default: false. (New in 3.0, replaces 2.0's "log_stdout" and "log_stderr")
'stdout_logfile' -- Put process stdout output in this file (and if redirect_stderr is true, also place stderr output in this file). If 'stdout_logfile' is unset or set to 'AUTO', supervisor will automatically choose a file location. If this is set to 'NONE', supervisord will create no log file. AUTO log files and their backups will be deleted when supervisord restarts. The stdout_logfile value can contain Python string expressions that will evaluated against a dictionary that contains the keys "process_num", "program_name" and "group_name". Default: AUTO. (New in 3.0, replaces 2.0's "logfile")
'stdout_logfile_maxbytes' -- The maximum number of bytes that may be consumed by stdout_logfile before it is rotated (suffix multipliers like "KB", "MB", and "GB" can be used in the value). Set this value to 0 to indicate an unlimited log size. Default: 50MB. (New in 3.0, replaces 2.0's "logfile_maxbytes")
'stdout_logfile_backups' -- The number of stdout_logfile backups to keep around resulting from process stdout log file rotation. Set this to 0 to indicate an unlimited number of backups. Default: 10. (New in 3.0, replaces "logfile_backups")
'stdout_capturefile' -- file written to when process is in "stdout capture mode" (see "Capture Mode and Process Communication Events" later in this document). May be a file path, NONE, or AUTO. The stdout_capturefile value can contain Python string expressions that will evaluated against a dictionary that contains the keys "process_num", "program_name" and "group_name". Default: NONE. (New in 3.0)
'stderr_logfile' -- Put process stderr output in this file unless redirect_stderr is true. Accepts the same value types as "stdout_logfile" and may contain the same Python string expressions. Default: AUTO. (New in 3.0)
'stderr_logfile_maxbytes' -- The maximum number of bytes before logfile rotation for stderr_logfile. Accepts the same value types as "stdout_logfile_maxbytes". Default: 50MB. (New in 3.0)
'stderr_logfile_backups' -- The number of backups to keep around resulting from process stderr log file rotation. Default: 10. (New in 3.0)
'stderr_capturefile' -- file written to when process is in "stderr capture mode" (see "Capture Mode and Process Communication Events" later in this document). May contain the same Python string expressions as "stdout_capturefile". May be a file path, NONE, or AUTO. Default: NONE. (New in 3.0)
'environment' -- A list of key/value pairs in the form "KEY=val,KEY2=val2" that will be placed in the child process' environment. The environment string may contain Python string expressions that will be evaluated against a dictionary containing "process_num", "program_name" and "group_name". Default: none. Note that the subprocess will inherit the environment variables of the shell used to start "supervisord" except for the ones overridden here. See "Subprocess Environment" below.
Note that a '[program:x]' section actually represents a "homogeneous process group" to supervisor (new in 3.0). The members of the group are defined by the combination of the 'numprocs and 'process_name' parameters in the configuration. By default, if numprocs and process_name are left unchanged from their defaults, the group represented by '[program:x]' will be named 'x' and will have a single process named 'x' in it. This provides a modicum of backwards compatibility with older supervisor releases, which did not treat program sections as homogeneous process group defnitions.
But for instance, if you have a '[program:foo]' section with a 'numprocs' of 3 and a 'process_name' expression of '%(program_name)s_%(process_num)02d', the "foo" group will contain three processes, named 'foo_00', 'foo_01', and 'foo_02'. This makes it possible to start a number of very similar processes using a single '[program:x]' section. All logfile names, all environment strings, and the command of programs can also contain similar Python string expressions, to pass slightly different parameters to each process.
Configuration File '[group:x]' Section Settings (New in 3.0)
It is often useful to group "homogeneous" processes groups (aka "programs") together into a "heterogeneous" process group so they can be controlled as a unit from supervisor's various controller interfaces.
To place programs into a group so you can treat them as a unit, define a '[group:x]' section in your configuration file, e.g.:[group:foo] programs=bar,baz priority=999
For the example above to work, there must be two 'program' sections elsewhere in your configuration file: '[program:bar]' and '[program:baz]'. If "homogeneous" program groups" (represented by program sections) are placed into a "heterogeneous" group via '[group:x]' section's "programs=" line, the homogeneous groups that are implied by the program section will not exist at runtime in supervisor. Instead, all processes belonging to each of the homogeneous groups will be placed into the heterogeneous group. In the above example, it means that the 'bar' and 'baz' homogeneous groups will not exist, and the processes that would have been under them will now be moved into the 'foo' group.
Configuration File '[eventlistener:x]' Section Settings (New in 3.0)
Supervisor allows specialized homogeneous process groups ("event listener pools") to be defined within the configuration file. These pools contain processes that are meant to receive and respond to event notifications from supervisor's event system. See "Supervisor Events" elsewhere in this document for an explanation of how events work and how to implement event listener programs.
An example of an eventlistener section defined within a supervisor configuration file, which creates a pool:[eventlistener:theeventlistenername] command=/bin/eventlistener process_name=%(program_name)s_%(process_num)02d numprocs=5 events=PROCESS_STATE_CHANGE buffer_size=10 priority=-1 autostart=true autorestart=true startsecs=1 startretries=3 exitcodes=0,2 stopsignal=QUIT stopwaitsecs=10 user=chrism redirect_stderr=true stdout_logfile=/a/path stdout_logfile_maxbytes=1MB stdout_logfile_backups=10 stderr_logfile=/a/path stderr_logfile_maxbytes=1MB stderr_logfile_backups environment=A=1,B=2
Note that all the options available to '[program:x]' sections are respected by eventlistener sections except for "stdout_capturefile" and "stderr_capturefile" (event listeners cannot emit process communication events, see "Capture Mode and Process Communication Events" elsewhere in this document).
'[eventlistener:x]' sections have two keys which '[program:x]' sections do not have:
'buffer_size' -- The event listener pool's event queue buffer size. When a listener pool's event buffer is overflowed (as can happen when an event listener pool cannot keep up with all of the events sent to it), the oldest event in the buffer is discarded.
'events' -- A comma-separated list of event type names that this listener is "interested" in receiving notifications for (see "Supervisor Events" elsewhere in this document for a list of valid event type names).
Configuration File '[rpcinterface:x]' Section Settings (ADVANCED, New in 3.0)
Changing "rpcinterface:x" settings in the configuration file is only useful for people who wish to extend supervisor with additional behavior.
In the sample config file, there is a section which is named "rpcinterface:supervisor". By default it looks like this:[rpcinterface:supervisor] supervisor.rpcinterface_factory = supervisor.xmlrpc:make_main_rpcinterface
This section must remain in the configuration for the standard setup of supervisor to work properly. If you don't want supervisor to do anything it doesn't already do out of the box, this is all you need to know about this type of section.
However, if you wish to add rpc interface namespaces to a custom version of supervisor, you may add additional [rpcinterface:foo] sections, where "foo" represents the namespace of the interface (from the web root), and the value named by "supervisor.rpcinterface_factory" is a factory callable which should have a function signature that accepts a single positional argument "supervisord" and as many keyword arguments as required to perform configuration. Any key/value pairs defined within the rpcinterface:foo section will be passed as keyword arguments to the factory. Here's an example of a factory function, created in the package "my.package":def make_another_rpcinterface(supervisord, **config): retries = int(config.get('retries', 0)) another_rpc_interface = AnotherRPCInterface(supervisord, retries) return another_rpc_interface And a section in the config file meant to configure it:: [rpcinterface:another] supervisor.rpcinterface_factory = my.package:make_another_rpcinterface retries = 1
Nondaemonizing of Subprocesses
Programs run under supervisor should not daemonize themselves. Instead, they should run in the foreground and not detach from the "terminal" that starts them. The easiest way to tell if a command will run in the foreground is to run the command from a shell prompt. If it gives you control of the terminal back, it's daemonizing itself and that will be the wrong way to run it under supervisor. You want to run a command that essentially requires you to press Ctrl-C to get control of the terminal back. If it gives you a shell prompt back after running it without needing to press Ctrl-C, it's not useful under supervisor. All programs have options to be run in the foreground but there's no standard way to do it; you'll need to read the documentation for each program you want to do this with.
Subprocesses will inherit the environment of the shell used to start the supervisord program. Several environment variables will be set by supervisor itself in the child's environment also, including "SUPERVISOR_ENABLED" (a flag indicating the process is under supervisor control), "SUPERVISOR_PROCESS_NAME" (the config-file-specified process name for this process) and "SUPERVISOR_GROUP_NAME" (the config-file-specified process group name for the child process).
These environment variables may be overridden within the "environment" global config option (applies to all subprocesses) or within the per-program "environment" config option (applies only to the subprocess specified within the "program" section). These "environment" settings are additive. In other words, each subprocess' environment will consist of:The environment variables set within the shell used to start supervisord... ... added-to/overridden-by ... ... the environment variables set within the "environment" global config option ... ... added-to/overridden-by ... ... supervisor-specific environment variables ("SUPERVISOR_ENABLED", "SUPERVISOR_PROCESS_NAME", "SUPERVISOR_GROUP_NAME") .. (New in 3.0) ... added-to/overridden-by ... .. the environment variables set within the per-process "environment" config option.
No shell is executed by supervisord when it runs a subprocess, so settings such as USER, PATH, HOME, SHELL, LOGNAME, etc. are not changed from their defaults or otherwise reassigned. This is particularly important to note when you are running a program from a supervisord run as root with a "user=" stanza in the configuration. Unlike cron, supervisord does not attempt to divine and override "fundamental" environment variables like USER, PATH, HOME, and LOGNAME when it performs a setuid to the user defined within the "user=" program config option. If you need to set environment variables for a particular program that might otherwise be set by a shell invocation for a particular user, you must do it explicitly within the "environment=" program config option. For example:[program:apache] command=/home/chrism/bin/httpd -DNO_DETACH user=chrism environment=HOME=/home/chrism,USER=chrism
Examples of Program Configurations
Apache 2.0.54:[program:apache] command=/usr/sbin/httpd -DNO_DETACH
Postgres 8.14:[program:postgres] command=/path/to/postmaster ; we use the "fast" shutdown signal SIGINT stopsignal=INT redirect_stderr=true
Zope 2.8 instances and ZEO:[program:zeo] command=/path/to/runzeo priority=1 [program:zope1] command=/path/to/instance/home/bin/runzope priority=2 redirect_stderr=true [program:zope2] command=/path/to/another/instance/home/bin/runzope priority=2 redirect_stderr=true
OpenLDAP slapd:[program:slapd] command=/path/to/slapd -f /path/to/slapd.conf -h ldap://0.0.0.0:8888
A process controlled by supervisord will be in one of the below states at any given time. You may see these state names in various user interface elements.
- STOPPED (0) -- The process has been stopped due to a stop request or
- has never been started.
STARTING (10) -- The process is starting due to a start request.
RUNNING (20) -- The process is running.
- BACKOFF (30) -- The process entered the STARTING state but
- subsequently exited too quickly to move to the RUNNING state.
STOPPING (40) -- The process is stopping due to a stop request.
- EXITED (100) -- The process exited from the RUNNING state (expectedly
- or unexpectedly).
FATAL (200) -- The process could not be started successfully.
UNKNOWN (1000) -- The process is in an unknown state (programming error).
Process progress through these states as per the following directed graph:--> STOPPED / | | | | | STOPPING | ^ ^ V | \--- STARTING <-----> BACKOFF | / ^ | | V | | \-- RUNNING / \ | | / \ V V / \ ----- FATAL EXITED
A process is in the STOPPED state if it has been stopped adminstratively or if it has never been started.
When an autorestarting process is in the BACKOFF state, it will be automatically restarted by supervisord. It will switch between STARTING and BACKOFF states until it becomes evident that it cannot be started because the number of startretries has exceeded the maximum, at which point it will transition to the FATAL state. Each start retry will take progressively more time.
An autorestarted process will never be automtatically restarted if it ends up in the FATAL state (it must be manually restarted from this state).
A process transitions into the STOPPING state via an administrative stop request, and will then end up in the STOPPED state.
A process that cannot be stopped successfully will stay in the STOPPING state forever. This situation should never be reached during normal operations as it implies that the process did not respond to a final SIGKILL, which is "impossible" under UNIX.
State transitions which always require user action to invoke are these:
FATAL -> STARTING
RUNNING -> STOPPING
State transitions which typically, but not always, require user action to invoke are these, with exceptions noted:
- STOPPED -> STARTING (except at supervisord startup if process is
- configured to autostart)
EXITED -> STARTING (except if process is configured to autorestart)
All other state transitions are managed by supervisord automatically.
Supervisor Events (New in 3.0)
At certain predefined points during supervisord's operation, "event notifications" are emitted. An event notification implies that something potentially interesting happened. Event listeners (see the "Event Listeners" section below) can be configured to subscribe to event notifications selectively, and may perform arbitrary actions based on an event notification (send email, make an HTTP request, etc).
Event types that may be subscribed to by event listeners are predefined by supervisor and fall into several major categories, including "process state change", "process communication", "supervisor state change", and "event system meta" events. These are described in detail below.
EVENT -- The base event type. This event type is abstract. It will never be sent directly. Subscribing to this event type will cause a subscriber to receive all event notifications emitted by supervisor.
Subtypes of EVENT:
PROCESS_STATE_CHANGE -- The value of this event type will be the process name. This event type is abstract, it will never be sent directly. Subscribing to this event type will cause a subscriber to receive event notifications of all the types listed below in "Subtypes of PROCESS_STATE_CHANGE".
The serialized body of a PROCESS_STATE_CHANGE event (and all subtypes) is in the form:process_name: <name> group_name: <name>
Subtypes of PROCESS_STATE_CHANGE:
PROCESS_STATE_CHANGE_STARTING -- indicates a process has moved from a state to the STARTING state. Subscribing to this event type will cause a subscriber to receive event notifications of all the types listed below in "Subtypes of PROCESS_STATE_CHANGE_STARTING".
Subtypes of PROCESS_STATE_CHANGE_STARTING:
PROCESS_STATE_CHANGE_STARTING_FROM_STOPPED -- subtype of PROCESS_STATE_CHANGE_STARTING, indicates a process has moved from the STOPPED state from the STARTING state.
PROCESS_STATE_CHANGE_STARTING_FROM_BACKOFF -- subtype of PROCESS_STATE_CHANGE_STARTING, indicates a process has moved from BACKOFF state to the STARTING state.
PROCESS_STATE_CHANGE_STARTING_FROM_EXITED -- subtype of PROCESS_STATE_CHANGE_STARTING, indicates a process has moved from the EXITED state to the STARTING state.
PROCESS_STATE_CHANGE_STARTING_FROM_FATAL -- subtype of PROCESS_STATE_CHANGE_STARTING, indicates a process has moved to the FATAL state to the STARTING state.
PROCESS_STATE_CHANGE_RUNNING_FROM_STARTING -- inidicates a process has moved from the STARTING state to the RUNNING state.
PROCESS_STATE_CHANGE_BACKOFF_FROM_STARTING -- indicates a process has moved from the STARTING state to the BACKOFF state.
PROCESS_STATE_CHANGE_STOPPING_FROM_RUNNING -- indicates a process has moved from the RUNNING state to the STOPPING state.
PROCESS_STATE_CHANGE_STOPPING_FROM_STARTING -- indicates a process has moved from the RUNNING state to the STARTING state.
PROCESS_STATE_CHANGE_EXITED_OR_STOPPED -- indicates a process has undergone a state change which caused it to move to the EXITED or STOPPED state.
Subtypes of PROCESS_STATE_CHANGE_EXITED_OR_STOPPED:
PROCESS_STATE_CHANGE_EXITED_FROM_RUNNNING -- indicates a process has moved from the RUNNING state to the EXITED state.
PROCESS_STATE_CHANGE_STOPPED_FROM_STOPPING -- indicates a process has moved from the STOPPING state to the STOPPED state.
PROCESS_STATE_CHANGE_FATAL_FROM_BACKOFF -- indicates a process has moved from the BACKOFF state to the FATAL state.
PROCESS_STATE_CHANGE_TO_UNKNOWN -- indicates a process has moved from a state to the UNKNOWN state (indicates an error in supervisord).
PROCESS_COMMUNICATION -- an event type raised when any process attempts to send information between <!--XSUPERVISOR:BEGIN--> and <!--XSUPERVISOR:END--> tags in its output. This event type is abstract, it will never be sent directly. Subscribing to this event type will cause a subscriber to receive event notifications of all the types listed below in "Subtypes of PROCESS_COMMUNICATION".
The serialized body of a PROCESS_COMMUNICATION event (and all subtypes) is:process_name: <name> group_name: <name> <data>
Subtypes of PROCESS_COMMUNICATION:
PROCESS_COMMUNICATION_STDOUT -- indicates a process has sent a message to supervisor on its stdout file descriptor.
PROCESS_COMMUNICATION_STDERR -- indicates a process has sent a message to supervisor on its stderr file descriptor.
SUPERVISOR_STATE_CHANGE -- an event type raised when supervisor's state changes. There is no value. Subscribing to this event type will cause a subscriber to receive event notifications of all the types listed below in "Subtypes of SUPERVISOR_STATE_CHANGE".
The serialization of a SUPERVISOR_STATE_CHANGE event is the empty string.
Subtypes of SUPERVISOR_STATE_CHANGE:
SUPERVISOR_STATE_CHANGE_RUNNING -- indicates that supervisor has started.
SUPERVISOR_STATE_CHANGE_STOPPING -- indicates that supervisor is stopping or restarting.
EVENT_BUFFER_OVERFLOW -- an event type raised when a listener pool's event buffer is overflowed (as can happen when an event listener pool cannot keep up with all of the events sent to it). When the pool's event buffer is overflowed, the oldest event in the buffer is thrown out.
The serialization of an EVENT_BUFFER_OVERFLOW body is:group_name: <name> event_type: <type of discarded event>
Event Listeners (New in 3.0)
Supervisor event listeners are subprocesses which are treated almost exactly like supervisor "programs" with the following differences:
- They are defined using an [eventlistener:x] section in the config file instead of a [program:x] section in the configuration file.
- Supervisor sends specially-formatted input to an event listener's stdin and expects specially-formatted output from an event listener's stdout in a request-response cycle. A protocol agreed upon between supervisor and the listener's implementer allows listeners to process event notifications.
- Supervisor does not respect "capture mode" output from event listener processes (see "Capture Mode and Process Communication Events" elsewhere in this document).
When an [eventlistener:x] section is defined, it actually defines a "pool", where the number of event listeners in the pool is determined by the "numprocs" value within the section. Every process in the event listener pool is treated equally by supervisor, and supervisor will choose one process from the pool to receive event notifications (filtered by the "events=" key in the eventlistener section).
An event listener can send arbitrary output to its stderr, which will be logged or ignored by supervisord depending on the stderr-related configuration files in its [eventlistener:x] section.
When an event notification is sent by the supervisor, all event listener pools which are subscribed to receive events for the event's type will be found. One of the listeners in each listener pool will receive the event notification (any "available" listener).
If the event cannot be sent because all listener in a pool are "busy", the event will be buffered and notification will be retried later. "Later" is defined as "the next time that supervisord's select loop executes".
A listener pool has an event buffer queue. The queue is sized via the listener pool's "buffer_size" config file option. If the queue is full and supervisor attempts to buffer an event, supervisor will throw away the oldest event in the buffer, log an error, and send an EVENT_BUFFER_OVERFLOW event. EVENT_BUFFER_OVERFLOW events are never themselves buffered.
Event listeners can be implemented in any language. Event listeners can be long-running or may exit after a single request (depending on the implementation and the "autorestart" parameter in the eventlistener's configuration).
An event listener implementation should operate in "unbuffered" mode or should flush its stdout every time it needs to communicate back to the supervisord process.
Event Listener States
An event listener process has three possible states that are maintained by supervisord:
ACKNOWLEDGED -- The event listener has acknowledged (accepted or rejected) an event send.
READY -- Event notifications may be sent to this event listener.
BUSY -- Event notifications may not be sent to this event listener.
When an event listener process first starts, supervisor automatically places it into the ACKNOWLEDGED state to allow for startup activities or guard against startup failures (hangs). Until the listener sends a READY token to its stdout, it will stay in this state.
When supervisord sends an event notification to a listener in the READY state, the listener will be placed into the BUSY state until it receives an OK or FAILED response from the listener, at which time, the listener will be transitioned back into the ACKNOWLEDGED state.
Event Listener Notification Protocol
Supervisord will notify an event listener in the READY state of an event by sending data to the stdin of the process. Supervisord will never send anything to the stdin of an event listener process while that process is in the BUSY or ACKNOWLEDGED state.
When supervisord sends a notification to an event listener process, the listener will first be sent a single "header" line on its stdin. The composition of the line is a set of four tokens separated by single spaces. The line is terminated with a 'n' (linefeed) character. The tokens on the line are:
<PROTOCOL_VERSION> <EVENT_TYPE_NAME> <EVENT_SERIAL_NUM> <PAYLOAD_LENGTH>
The PROTOCOL_VERSION always consists of "SUPERVISORD" followed immediately by numeric characters indicating the protocol version, with no whitespace in between. An example: "SUPERVISOR3.0"
The EVENT_TYPE_NAME is the specific event type name (see "Supervisor Events" elsewhere in this document). An example: "PROCESS_COMMUNICATION_STDOUT".
The EVENT_SERIAL_NUM is an integer assigned to each event. It is useful for functional testing. An example: "30".
The PAYLOAD_LENGTH is an integer indicating the number of bytes in the event payload. An example: "22".
An example of a complete header line:
SUPERVISOR3.0 PROCESS_COMMUNICATION_STDOUT 30 22n
Directly following the linefeed character in the header is the event payload. It consists of PAYLOAD_LENGTH bytes representing a serialization of the event data. See "Supervisor Events" for the specific event data serialization definitions. An example payload for a PROCESS_COMMUNICATION_STDOUT event notification is:process_name: foo group_name: bar This is the data that was sent between the tags
Once it has processed the header, the event listener implementation should read PAYLOAD_LENGTH bytes from its stdin, perform an arbitrary action based on the values in the header and the data parsed out of the serialization. It is free to block for an arbitrary amount of time while doing this. Supervisor will continue processing normally as it waits for a response and it will send other events of the same type to other listener processes in the same pool as necessary.
After the event listener has processed the event serialization, in order to notify supervisord about the result, it should send either an "OK" token or a "FAILED" token immediately followed by a carriage return character to its stdout. If supervisord receives an "OK" token, it will assume that the listener processed the event notification successfully. If it receives a "FAILED" token, it will assume that the listener has failed to process the event, and the event will be rebuffered and sent again at a later time. The event listener may reject the event for any reason by returning a "FAILED" token. This does not indicate a problem with the event data or the event listener. Once an "OK" or "FAILED" token is received by supervisord, the event listener is placed into the ACKNOWLEDGED state.
Once the listener is in the ACKNOWLEDGED state, it may either exit (and subsequently be restarted by supervisor if its "autorestart" config parameter is true), or it may continue running. If it continues to run, in order to be placed back into the READY state by supervisord, it must send a "READY" token followed immediately by a carriage return to its stdout.
Example Event Listener Implementation
A Python implementation of a "long-running" event listener which accepts an event notification, prints the header and a list of event serial numbers it has received to its stderr, and responds with an OK, and then subsequently a READY is as follows:import sys L =  def stdout_write(s): sys.stdout.write(s) sys.stdout.flush() def stderr_write(s): sys.stderr.write(s) sys.stderr.flush() while 1: stdout_write('READY\n') line = sys.stdin.readline() stderr_write(line) ver, event, serial, length = line.split(' ', 3) L.append(serial) data = sys.stdin.read(int(length)) stderr_write(str(L)) stdout_write('OK\n')
Event Listener Error Conditions
If the event listener process dies while the event is being transmitted to its stdin, or if it dies before sending an OK/FAILED response back to supervisord, the event is assumed to not be processed and will be rebuffered by supervisord and sent again later.
If an event listener sends data to its stdout which supervisor does not recognize as an appropriate response based on the state that the event listener is in, the event listener will be placed into the UNKNOWN state, and no further event notifications will be sent to it. If an event was being processed by the listener during this time, it will be rebuffered and sent again later.
Capture Mode and Process Communication Events (New in 3.0)
If a '[program:x]' section in the configuration file defines a "stdout_capturefile" or "stderr_capturefile" parameter, each process represented by the program section may emit special tokens on its stdout or stderr stream (respectively) which will effectively cause supervisor to emit a "PROCESS_COMMUNICATION" event type.
The process communications protocol relies on two tags, one which commands supervisor to enter "capture mode" for the stream and one which commands it to exit. When a process stream enters "capture mode", data sent to the stream will be sent to a separate logfile (the "capturefile"). When a process stream exits capture mode, the data in the capturefile is read into memory (a maximum of 2MB), and a PROCESS_COMMUNICATION event is emitted by supervisor, which may be intercepted by event listeners.
The tag to begin "capture mode" in a process stream is '<!--XSUPERVISOR:BEGIN-->'. The tag to exit capture mode is '<!--XSUPERVISOR:END-->'. The data between these tags may be arbitrary, and forms the payload of the PROCESS_COMMUNICATION event. For example, if a program is set up with a stdout_capturefile, and it emits the following on its stdout stream:<!--XSUPERVISOR:BEGIN-->Hello!<!--XSUPERVISOR:END-->
data in the payload of "Hello!".
Killing supervisord with SIGHUP will stop all processes, reload the configuration from the config file, and restart all processes.
Killing supervisord with SIGUSR2 will close and reopen the supervisord activity log and child log files.
The UNIX permissions on the socket effectively control who may send commands to the server. HTTP basic authentication provides access control for internet and UNIX domain sockets as necessary.
I have done my best to assure that use of a supervisord process running as root cannot lead to unintended privilege escalation, but caveat emptor. Particularly, it is not as paranoid as something like DJ Bernstein's "daemontools", inasmuch as "supervisord" allows for arbitrary path specifications in its configuration file to which data may be written. Allowing arbitrary path selections can create vulnerabilities from symlink attacks. Be careful when specifying paths in your configuration. Ensure that supervisord's configuration file cannot be read from or written to by unprivileged users and that all files installed by the supervisor package have "sane" file permission protection settings. Additionally, ensure that your PYTHONPATH is sane and that all Python standard library files have adequate file permission protections. Then, pray to the deity of your choice.
Some examples of shell scripts to start services under supervisor can be found "here":http://www.thedjbway.org/services.html. These examples are actually for daemontools but the premise is the same for supervisor. Another collection of recipes for starting various programs in the foreground is "here":http://smarden.org/runit/runscripts.html .
Some processes (like mysqld) ignore signals sent to the actual process/thread which is created by supervisord. Instead, a "special" thread/process is created by these kinds of programs which is responsible for handling signals. This is problematic, because supervisord can only kill a pid which it creates itself, not any child thread or process of the program it creates. Fortunately, these programs typically write a pidfile which is meant to be read in order to kill the process. As a workaround for this case, a special "pidproxy" program can handle startup of these kinds of processes. The pidproxy program is a small shim that starts a process, and upon the receipt of a signal, sends the signal to the pid provided in a pidfile. A sample supervisord configuration program entry for a pidproxy-enabled program is provided here:[program:mysql] command=/path/to/pidproxy /path/to/pidfile /path/to/mysqld_safe
The pidproxy program is named 'pidproxy.py' and is in the supervisor distribution.
My program never starts and supervisor doesn't indicate any error: Make sure the "x" bit is set on the executable file you're using in the command= line.
How can I tell if my program is running under supervisor? Supervisor and its subprocesses share an environment variable "SUPERVISOR_ENABLED". When a process is run under supervisor, your program can check for the presence of this variable to determine whether it is running under supervisor (new in 2.0).
My command line works fine when I invoke it by hand from a shell prompt, but when I use the same command line in a supervisor "command=" section, the program fails mysteriously. Why? This may be due to your process' dependence on environment variable settings. See "Subprocess Environment" in this document.
Maillist and Reporting Bugs
You may subscribe to the 'Supervisor-users' "maillist":http://lists.palladion.com/mailman/listinfo/supervisor-users
Please report bugs at "the collector":http://www.plope.com/software/collector .
Chris McDonough (email@example.com) "Agendaless Consulting":http://www.agendaless.com