[ https://issues.apache.org/jira/browse/PIO-95?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16052337#comment-16052337 ]
ASF GitHub Bot commented on PIO-95: ----------------------------------- Github user dszeto commented on a diff in the pull request: https://github.com/apache/incubator-predictionio/pull/394#discussion_r122524115 --- Diff: common/src/main/resources/application.conf --- @@ -9,3 +9,7 @@ spray.can { verbose-error-messages = "on" } } + +spray.can.server { + request-timeout = 35s +} --- End diff -- @mars You can put `request-timeout` inside the block above. We don't need to repeat `spray.can.server`. The configuration library will flatten the hierarchical configuration key. > Configurable request timeout for REST API > ----------------------------------------- > > Key: PIO-95 > URL: https://issues.apache.org/jira/browse/PIO-95 > Project: PredictionIO > Issue Type: Improvement > Components: Core > Affects Versions: 0.11.0-incubating > Reporter: Mars Hall > > We've found the default 20-second REST API request timeout is too short for > our batch-prediction use cases. We're running PredictionIO on Heroku which > has its own [timeout starting at > 30-seconds|https://devcenter.heroku.com/articles/limits#http-timeouts]. So > we'd prefer a more generous or easily configurable timeout to allow Heroku's > routing layer to impose & track this limit in the platform layer. > I investigated how to configure this and found [Spray > `application.conf`|http://spray.io/documentation/1.2.4/spray-can/configuration/]. > This PR simply increases the timeout. > I would love guidance on how we might extract this config into an environment > variable or a value in `pio-env.sh`. > Investigation / implementation PR: > https://github.com/apache/incubator-predictionio/pull/394 -- This message was sent by Atlassian JIRA (v6.4.14#64029)