1STARMAN(1) User Contributed Perl Documentation STARMAN(1)
2
6 starman - Starman launcher
7
9 starman --listen :5001 --listen /tmp/starman.sock
10 starman --workers 32 --port 8080
11
13 -l, --listen
14 --listen HOST:PORT --listen :PORT --listen UNIX_SOCKET
15 --listen HOST:PORT:ssl
16 Specifies the TCP address, ports and UNIX domain sockets to bind to
18 wait for requests. You can repeat as many times as you want and mix
19 TCP and UNIX domain sockets.
20 For TCP sockets you can append ":ssl" after the port to specify
22 that connections on that port should use SSL. Note that the SSL
23 support is experimental and hasn't been widely tested.
24 Defaults to any IP address and port 5000.
26 --host
28 --host 127.0.0.1
29 Specifies the address to bind.
31 This option is for a compatibility with plackup and you're
33 recommended to use "--listen" instead.
34 --port
36 --port 8080
37 Specifies the port to bind.
39 This option is for a compatibility with plackup and you're
41 recommended to use "--listen" instead.
42 -S, --socket
44 -S /tmp/starman.sock
45 Specifies the path to UNIX domain socket to bind.
47 This option is for a compatibility with plackup and you're
49 recommended to use "--listen" instead.
50 --workers
52 Specifies the number of worker pool. Defaults to 5.
53 Starman by default sets up other spare server configuration based
55 on this workers value, making sure there are always only "N" worker
56 processes running. So even if there're no idle workers, Starman
57 won't spawn off spare processes since that's mostly what you want
58 to do by fine tuning the memory usage etc. in the production
59 environment.
60 --backlog
62 Specifies the number of backlog (listen queue size) of listener
63 sockets. Defaults to 1024.
64 On production systems, setting a very low value can allow failover
66 on frontend proxy (like nginx) to happen more quickly, if you have
67 multiple Starman clusters.
68 If you're doing simple benchmarks and getting connection errors,
70 increasing this parameter can help avoid them. You should also
71 consider increasing "net.core.somaxconn". Note that this is not
72 recommended for real production system if you have another cluster
73 to failover (see above).
74 --max-requests
76 Number of the requests to process per one worker process. Defaults
77 to 1000.
78 --preload-app
80 This option lets Starman preload the specified PSGI application in
81 the master parent process before preforking children. This allows
82 memory savings with copy-on-write memory management. When not set
83 (default), forked children loads the application in the
84 initialization hook.
85 Enabling this option can cause bad things happen when resources
87 like sockets or database connections are opened at load time by the
88 master process and shared by multiple children.
89 Since Starman 0.2000, this option defaults to false, and you should
91 explicitly set this option to preload the application in the master
92 process.
93 Alternatively, you can use -M command line option (plackup's common
95 option) to preload the modules rather than the <application>
96 itself.
97 starman -MCatalyst -MDBIx::Class myapp.psgi
99 will load the modules in the master process for memory savings with
101 CoW, but the actual loading of "myapp.psgi" is done per children,
102 allowing resource managements such as database connection safer.
103 If you enable this option, sending "HUP" signal to the master
105 process will not pick up any code changes you make. See "SIGNALS"
106 for details.
107 --disable-keepalive
109 Disable Keep-alive persistent connections. It is an useful
110 workaround if you run Starman behind a broken frontend proxy that
111 tries to pool connections more than a number of backend workers
112 (i.e. Apache mpm_prefork + mod_proxy).
113 --keepalive-timeout
115 The number of seconds Starman will wait for a subsequent request
116 before closing the connection if Keep-alive persistent connections
117 are enabled. Setting this to a high value may cause performance
118 problems in heavily loaded servers. The higher the timeout, the
119 more backend workers will be kept occupied waiting on connections
120 with idle clients.
121 Defaults to 1.
123 --read-timeout
125 The number of seconds Starman will wait for a request on a new
126 connection before closing it. Setting this to a high value may
127 cause performance problems in heavily loaded servers. The higher
128 the timeout, the more backend workers will be kept occupied waiting
129 on connections with idle clients. You may need this if your proxy /
130 load balancer likes to keep a pool of open connections while
131 waiting for clients (eg. Amazon ELB).
132 Defaults to 5.
134 --user
136 To listen on a low-numbered (<1024) port, it will be necessary to
137 start the server as root. Use the "--user" option to specify a
138 userid or username that the server process should switch to after
139 binding to the port.
140 Defaults to the current userid.
142 --group
144 Specify the group id or group name that the server should switch to
145 after binding to the port. This option is usually used with
146 "--user".
147 Defaults to the current group id.
149 --pid
151 Specify the pid file path. Use it with "-D|--daemonize" option,
152 described in "plackup -h".
153 --error-log
155 Specify the pathname of a file where the error log should be
156 written. This enables you to still have access to the errors when
157 using "--daemonize".
158 --ssl-cert
160 Specify the path to SSL certificate file.
161 --ssl-key
163 Specify the path to SSL key file.
164 --enable-ssl
166 Enable SSL on all TCP sockets. This is an experimental feature.
167 --disable-proctitle
169 Disable the behavior to set proctitle to "starman (master)" and
170 "starman (worker)" respectively on master and workers.
171 Starman passes through other options given to Plack::Runner, the common
173 backend that plackup uses, so the most options explained in "plackup
174 -h" such as "--access-log" or "--daemonize" works fine in starman too.
175 Setting the environment variable "STARMAN_DEBUG" to 1 makes the Starman
177 server running in the debug mode.
178
180 HUP Sending "HUP" signal to the master process will restart all the
181 workers gracefully (meaning the currently running requests will
182 shut down once the request is complete), and by default, the
183 workers will pick up the code changes you make by reloading the
184 application.
185 If you enable "--preload-app" option, however, the code will be
187 only loaded in the startup process and will not pick up the code
188 changes you made. If you want to preload the app and do graceful
189 restarts by reloading the code changes, you're recommended to use
190 Server::Starter, configured to send "QUIT" signal when superdaemon
191 received "HUP", i.e:
192 start_server --interval 5 --port 8080 --signal-on-hup=QUIT -- \
194 starman --preload-app myapp.psgi
195 You will then send the HUP signal to "start_server" process to
197 gracefully reload the starman cluster (master and workers).
198 With Server::Starter 0.12 or later, you should also be able to set
200 "--signal-on-term" to QUIT so that you can safely shutdown Starman
201 first and then stop the "start_server" daemon process as well.
202 TTIN, TTOU
204 Sending "TTIN" signal to the master process will dynamically
205 increase the number of workers, and "TTOU" signal will decrease it.
206 INT, TERM
208 Sending "INT" or "TERM" signal to the master process will kill all
209 the workers immediately and shut down the server.
210 QUIT
212 Sending "QUIT" signal to the master process will gracefully
213 shutdown the workers (meaning the currently running requests will
214 shut down once the request is complete).
215
217 You're recommended to use signals (see above) to reload the
218 application, and are strongly discouraged to use "-r" or "-R"
219 (reloading flag) from plackup. These options will make a separate
220 directory watcher process, and makes your life difficult if you want to
221 combine with other process daemon tools such as Server::Starter.
222
224 "starman" executable is basically the equivalent of using "plackup"
225 with "Starman" server handler i.e. "plackup -s Starman", except that
226 "starman" delay loads the application with the Delayed loader by
227 default, which can be disabled with "--preload-app".
228 "starman" command also automatically sets the environment ("-E") to the
230 value of deployment.
231 You're recommended to use "starman" unless there's a reason to stick to
233 "plackup" for compatibility.
234
236 Starman
237perl v5.30.0 2019-07-26 STARMAN(1)