Conversation
…hange default behavior, but if user has specified a value, in seconds, for RPTDaemonTTL, then any run-periodic-tasks daemons will exit after this limit has been reached. Primary usage is to avoid memory leak issues, works great when used in conjunction with daemontools supervise to auto-restartthe rpt daemons.
|
Thank for the response, Tim. I thought about documentation but wasn't sure where/how to do so -- sorry I am not up to speed on the melody process. Thought about adding a couple of line to the pod docs for this module, but wasn't sure if there was a better place for it. The documentation for the MT config directive would be fairly short, something like: "Specifying a value for RPTDaemonTTL in your config file, in seconds, will effectively create a time limit on all run-periodic-tasks process running with the -daemon flag. For example 'RPTDaemonTTL 21600' would result in run-periodic-tasks daemons exiting after 6 hours. Used in conjunction with other tools to restart the daemons, this can help ensure that these process do not consume increasingly large amounts of server RAM." WRT the daemontools usage, that it just one such 3rd-party tool that can be used in conjunction with this directive. I am sure there are others. Does it make sense to include this in melody docs? Here is a quick stab at it, FWIW: Assuming that daemontools is already installed and running on the server, follow these steps for each run-periodic-tasks daemon process:
The combined usage of RPTDaemonTTL and daemontools can ensure that rpt daemons are restarted periodically, without the need for killing processes in the middle of some task or job. |
Implemented new RPTDaemonTTL config directive. This commit does not change default behavior, but if user has specified a value, in seconds, for RPTDaemonTTL, then any run-periodic-tasks daemons will exit after this limit has been reached.
Primary usage is to avoid memory leak issues, works great when used in conjunction with daemontools supervise to auto-restart the rpt daemons.