Updated the twitter_runtime.spec so that it works and includes geospatial indexs

This commit is contained in:
Andrew Serff 2015-03-25 16:21:43 -06:00
parent 9fd14cad4f
commit 518b42c900
1 changed files with 58 additions and 52 deletions

View File

@ -1,12 +1,44 @@
[
{ {
"description": "Ingestion spec for Twitter spritzer. Dimension values taken from io.druid.examples.twitter.TwitterSpritzerFirehoseFactory",
"spec": {
"dataSchema": { "dataSchema": {
"dataSource": "twitterstream", "dataSource": "twitterstream",
"granularitySpec": { "parser": {
"queryGranularity": "all", "parseSpec": {
"segmentGranularity": "hour", "format": "json",
"type": "uniform" "timestampSpec": {
"column": "utcdt",
"format": "iso"
},
"dimensionsSpec": {
"dimensions": [
"text",
"htags",
"contributors",
"lat",
"lon",
"retweet_count",
"follower_count",
"friendscount",
"lang",
"utc_offset",
"statuses_count",
"user_id",
"ts"
],
"dimensionExclusions": [
],
"spatialDimensions": [
{
"dimName": "geo",
"dims": [
"lat",
"lon"
]
}
]
}
}
}, },
"metricsSpec": [ "metricsSpec": [
{ {
@ -94,58 +126,32 @@
"type": "max" "type": "max"
} }
], ],
"parser": { "granularitySpec": {
"parseSpec": { "type": "uniform",
"dimensionsSpec": { "segmentGranularity": "DAY",
"dimensions": [ "queryGranularity": "NONE"
"text",
"htags",
"contributors",
"lat",
"lon",
"retweet_count",
"follower_count",
"friendscount",
"lang",
"utc_offset",
"statuses_count",
"user_id",
"ts"
],
"dimensionExclusions": [
],
"spatialDimensions": [
{
"dimName": "geo",
"dims": [
"lat",
"lon"
]
}
]
},
"format": "json",
"timestampSpec": {
"column": "ts",
"format": "millis"
}
}
} }
}, },
"ioConfig": { "ioConfig": {
"type": "realtime",
"firehose": { "firehose": {
"type": "twitzer",
"maxEventCount": 500000, "maxEventCount": 500000,
"maxRunMinutes": 120, "maxRunMinutes": 120
"type": "twitzer"
}, },
"plumber": {
"type": "realtime" "type": "realtime"
}
}, },
"tuningConfig": { "tuningConfig": {
"intermediatePersistPeriod": "PT10m",
"maxRowsInMemory": 500000,
"type": "realtime", "type": "realtime",
"windowPeriod": "PT10m" "maxRowsInMemory": 500000,
"intermediatePersistPeriod": "PT2m",
"windowPeriod": "PT3m",
"basePersistDirectory": "\/tmp\/realtime\/basePersist",
"rejectionPolicy": {
"type": "messageTime"
} }
},
"type": "index_realtime"
} }
}
]