dp-frontend-filter-dataset-controller

README

dp-frontend-filter-dataset-controller

An HTTP service for the controlling of data relevant to the filtering of a particular dataset.

Configuration

Environment variable	Default	Description
BIND_ADDR	:20001	The host and port to bind to.
RENDERER_URL	http://localhost:20010	The URL of dp-frontend-renderer.
CODE_LIST_API	http://localhost:22400	The URL of the code list api
FILTER_API_URL	http://localhost:22100	The URL of the filter api
DATASET_API_URL	http://localhost:22000	The URL of the dataset api
HIERARCHY_API_URL	http://localhost:22600	The URL of the hierarchy api
SEARCH_API_URL	http://localhost:23100	The URL of the search api
DOWNLOAD_SERVICE_URL	http://localhost:23600	The URL of the download service
ZEBEDEE_URL	http://localhost:8082	The url of the zebedee service. Only required if profiler is enabled
DATASET_API_AUTH_TOKEN	n/a	The token used to access the Dataset API
FILTER_API_AUTH_TOKEN	n/a	The token used to access the Filter API
SEARCH_API_AUTH_TOKEN	n/a	The token used to access the Search API
ENABLE_DATASET_PREVIEW	false	Flag to add preview of dataset to output page
ENABLE_PROFILER	false	Flag to enable go profiler
PPROF_TOKEN	""	The profiling token to access service profiling
GRACEFUL_SHUTDOWN_TIMEOUT	5s	The graceful shutdown timeout in seconds
HEALTHCHECK_INTERVAL	30s	The time between calling healthcheck endpoints for check subsystems
HEALTHCHECK_CRITICAL_TIMEOUT	90s	The time taken for the health changes from warning state to critical due to subsystem check failures

HEALTHCHECK_INTERVAL and HEALTHCHECK_CRITICAL_TIMEOUT can use the following formats to represent duration of time:

"300ms" = 300 milliseconds
"1.5h" =  1.5 hours
"2h45m" = 2 hours 45 minutes

Valid time units are 
"ns" = nanosecond
"us" (or "µs") = microsecond
"ms" = millisecond
"s" = second
"m" = minute
"h" = hour

Profiling

An optional /debug endpoint has been added, in order to profile this service via pprof go library. In order to use this endpoint, you will need to enable profiler flag and set a PPROF_TOKEN:

export ENABLE_PROFILER=true
export PPROF_TOKEN={generated uuid}

Then you can us the profiler as follows:

1- Start service, load test or if on environment wait for a number of requests to be made.

2- Send authenticated request and store response in a file (this can be best done in command line like so: curl <host>:<port>/debug/pprof/heap -H "Authorization: Bearer {generated uuid} > heap.out - see pprof documentation on other endpoints

3- View profile either using a web ui to navigate data (a) or using pprof on command line to navigate data (b) a) go tool pprof -http=:8080 heap.out b) go tool pprof heap.out, -o flag to see various options

Contributing

See CONTRIBUTING for details

Licence

Copyright ©‎ 2017, Office for National Statistics (https://www.ons.gov.uk)

Released under MIT license, see LICENSE for details.