watchd is a process watch dog to keep the watched processes alive.
the sample config file please see src/sample.conf.
config fragment in src/sample.conf:
# one section for a group sub processes
# you can config more than one sections
[test1]
# the type: daemon or cron (for crontab)
# default: daemon
type = daemon
# the mode: all or failover
## all: run all subprocess_command
## failover: run the first command,
## run the second command when the first exit,
## run the first command again when the last exit,
## and so on.
# default: all
mode = failover
# if run by shell as: sh -c command
## true or 1: run with sh -c command
## false or 0: run command directly
## auto: auto detect if need run by sh -c
# default auto
run_by_sh = auto
# force restart interval in seconds
# this parameter only for daemon
# default 0 for never force restart
force_restart_interval = 86400
# set environment variable
# can ocur more than once
set_env = OMP_NUM_THREADS=4
# the command line of sub process
# can be a simple command line or a shell command as following four formats:
## 1. (command)
## 2. command > output_filename or command >> output_filename
## 3. command1 | command2 ...
## 4. command &
## the shell command line while be exec as: sh -c command_line
subprocess_command = /bin/echo OMP_NUM_THREADS=$OMP_NUM_THREADS $host_index >> /tmp/echo.log
# check sub process alive interval in seconds
# 0 for never check
check_alive_interval = 10
# retry threshold of check alive
# kill the sub process when the check fail count exceeds this parameter
# default: 3
check_alive_retry_threshold = 3
# check_alive_command can be a command or a library whose filename ends with .so
## the check command output OK for check passed, others for fail
#
## the library must export c function:
## int check_alive(int argc, char **argv);
## argv[0] is the library filename, argv[1] is the first parameter and so on.
## return 0 for success, != 0 for fail.
##
## for example:
## #@function REPLACE_VARS
## check_alive_command = /usr/local/lib/libdfscheckalive.so %{encoder_port} 2 30
#@function REPLACE_VARS
check_alive_command = echo OK
[test2]
type = cron
subprocess_command = ls -l / >> /tmp/ls.log
# the time base to schedule
# this parameter only for cron
time_base = 00:00
# repeat interval in seconds
# this parameter only for cron
repeat_interval = 60