A web based monitor for PM2
A web based monitor for PM2.
With the release of 0.11 pm2 no longer uses TCP sockets for the event bus, instead using unix sockets. TCP sockets make monitoring hosts remotely possible so that functionality is currently broken.
Hopefully the pm2 team have a solution for this.
For debugging to work, node-inspector must be installed and running on the same machine as pm2 (not necessarily the same as pm2-web).
$ npm install -g pm2-web
All configuration options are documented in the default configuration file.
pm2-web will load one of the following files if they exist (in order of preference)
The default configuration file is always loaded and other config files will be applied to the default configuration.
The configuration file(s) loaded and the final configuration object will both be recorded in the logs.
Configuration files are loaded using cjson so comments are ok.
All options can be passed as command line arguments and will override any settings found in configuration files.
To use HTTP basic auth, set
www:authentication:enabled to true in your configuration file. See the default configuration file for more information.
N.b. Your password will be sent in plain text. If you enable HTTP auth, you should probably enable SSL as well.
pm2-web can start a https server if so desired. To do so, set
www:ssl:enabled to true in your configuration file and supply your certificate details. If you do not have a SSL certificate, the
generate_certificate.sh script in the
/certs directory will create a self-signed certificate for you.
To debug a running process, node-inspector must be installed and running on the same host as the process.
Specify the port it's running on as the
inspector property of pm2:host. E.g.:
"pm2":"host": "foo.baz.com""inspector": 8080
You should then see a debug icon appear next to the stop/restart/reload icons when the process is running.
Clicking this icon will send a
SIGUSR1 signal to the process to put it into debug mode and open node-inspector in a new window.
N.b. you may need to change which source file you are looking at in node-inspector to see anything useful.
By default node will listen for debugger connections on port 5858. If you attempt to debug multiple processes you must specify different debug ports for them:
$ pm2 start --node-args="--debug=7000" foo.js$ pm2 start --node-args="--debug=7001" bar.js
This is not possible because:
$ pm2 start --node-args="--debug=7000" -i 4 foo.js
will start four separate processes all listening on port 7000.
If you are expecting to debug your process, please only start one of them.
Restarting a process stops the current process and starts a new one, dropping connections in the process.
Reloading starts a new process in the background, killing the old process after the new one has initialised which reduces downtime.
N.b. your process must exit cleanly (e.g. close sockets, database connections) otherwise the old process will never be killed.
Soft reloading (the default) will cause pm2 to send your process a
shutdown message and kill it shortly afterwards. Hard reloading will kill it immediately.
To control this behaviour, specify the following in your config file:
To listen for the
shutdown event, add the following to your program:
You can tweak the resource usage graph to be more or less specific. The thinking here is that lots of processes with lots of process usage data will make your browser a bit crashey.
By default it will retain 1000 resource usage measurements of a process (CPU and memory) over a maximum of five days with 40% of the measurements taken in the last 24 hours, 25% from the day before, 10% each from the two days before that and 5% from the day before that.
The update frequency of the graph is controlled by
--updateFrequency as detailed above.
The number of data points that will be plotted on the graph in total. If you've got a lot of processes, you may wish to set this to a lowish number to minimise memory consumption in your browser and the pm2-web process itself.
--graph.distribution 40 --graph.distribution 25 --graph.distribution 10 --graph.distribution 10 --graph.distribution 5
The number of
--graph.distribution arguments is the number of days worth of data to graph (default 5) and the value is what percentage of
--graph.datapoints should be plotted on a given day (the first
--graph.distribution argument is today, the second is yesterday, etc).
What this means is that any recent resource usage data will have a more accurate representation in the graph at the cost of consuming more memory and older data will be less accurate but also less likely to crash your browser.
pm2-web will display log lines emitted by monitored processes after pm2-web itself was started. In order to keep resource usage reasonable by default it will show 1000 log lines.
You can alter this behaviour by specifying
--logs:max, so for example to lower this to 500 lines: