Control health checks and toggle upstream node status in load balancers with ease.
Demo at poolse.vacovsky.us.
Exposes a simple interface to “up” and “down” upstream nodes.
Monitoring of targets can be as simple or complex as needed. HTTP Status code checking, reponse strings parsing, and fail/success thresholds are supported for determining health of upstream nodes.
Multiple status formats are provided, including: 503, no repsonse, as well as partial and full JSON blobs describing the state and parameters per target.
Reconfigure/reload of settings on the fly, without “downing” the node (this is handy when using automation platforms like Chef).
Provides a common format/interface for managing applications which are behind some form of reverse proxy or load balancer that checks at interval for server health status.
go get -u github.com/davecgh/go-spew/spewgit clone https://github.com/vacoj/poolse.gitcd poolse/src/poolsego build
./poolse /path/to/config.json
{"state": {"startup_state": true, // if true, when server starts, if first check passes server state is marked OK"administrative_state": "AdminOff", // If persistent state isn't on, this is the default statup state for the STATUS. If will only be OK if all Targets are also OK on first check"persist_state": true // indicates whether or not STATUS.State.AdministrativeState should be sticky between settings/application restarts and reloads.},"targets": [{"endpoint": "http://localhost:5704/fakehealth", // url to your application's health endpoint"polling_interval": 15, // polling interval for target endpoint, in seconds"expected_status_code": 200, // *required* HTTP status code to look for. If this isn't returned when the check happens, we mark OK as false."up_count_threshold": 10, // this many failed checks will mark target as online"down_count_threshold": 10 // this many failed checks will mark target as offline},{"name": "Expected Example","endpoint": "http://localhost:5704/fakeexpected", // url to your application's health endpoint"polling_interval": 20, // polling interval for target endpoint, in seconds"expected_status_code": 200, // HTTP status code to look for. If this isn't returned when the check happens, we mark OK as false."expected_response_strings": ["{\"is_working\": true}"],"up_count_threshold": 10, // this many failed checks will mark target as online"down_count_threshold": 10 // this many failed checks will mark target as offline},{"name": "Unexpected Example","endpoint": "http://localhost:5704/fakeexpected", // url to your application's health endpoint"polling_interval": 10, // polling interval for target endpoint, in seconds"expected_status_code": 200, // HTTP status code to look for. If this isn't returned when the check happens, we mark OK as false."unexpected_response_strings": ["{\"is_working\": false}"], // response is parsed for this string. If unexpected_response_string is blank, check is ignored. If found, OK is false (an example would be searching repsonse text for {"thisthing": false}, and if found, causes OK to be set to false)"up_count_threshold": 10, // this many failed checks will mark target as online"down_count_threshold": 10 // this many failed checks will mark target as offline},{"name": "Fake Smoke" // Arbitrary - use for your own reasons, or leave it blank."endpoint": "http://localhost:5704/fakesmoke", // url to your application's health endpoint"polling_interval": 300, // polling interval for target endpoint, in seconds"expected_status_code": 200, // *required* HTTP status code to look for. If this isn't returned when the check happens, we mark OK as false.}],"service": {"http_port": "5704", // *string not int; port to listen on for incoming web requests"debug": false, // displays certain pieces of data in console if true"show_http_log": true, // shows log in console of calls being made if true"state_file_name": "state.dat", // name of state file. defaults to state.dat if not preset"follow_redirects": false // whether or not to follow redirects when querying targets}}
cd poolse/src/poolse/staticbower install
id (example: /status/simple?id=1) to get just the target object at that index; id should correspond to index of target in config file.
{"State": {"ok": true,"startup_state": true,"persist_state": true,"administrative_state": ""},"Targets": [{"id": 0,"name": "","endpoint": "http://localhost","polling_interval": 5,"expected_status_code": 200,"expected_response_strings": null,"unexpected_response_strings": null,"last_ok": "2017-03-23T08:30:51.1019184-07:00","last_checked": "2017-03-23T08:30:51.1019184-07:00","ok": true,"up_count": 6987,"up_count_threshold": 10,"down_count": 0,"down_count_threshold": 1}],"Version": "0.3.10"}