Is there anyway we can query how many queries piling up in Druid?

[Noppanit Charassinvichai]

We have some long running queries and sometime that would slow down the cluster. We would like to start killing some of the slow running queries. I've just seen a way to cancel the query but is there anyway we can list how many queries running in Druid? http://druid.io/docs/0.8.3/querying/querying.html 

Thanks. You guys are awesome.  

[charles.allen]

On historicals, you'll want to look at the segment/scan/pending metric http://druid.io/docs/latest/operations/metrics.html 

[Noppanit Charassinvichai]

Thanks for your reply. I tried the query with our historical node and with GET http://historical-ip:8083/segment/scan/pending but I get 404 but I can go to http://historical-ip:8083/status and get result back. Not sure if my query is right?

[charles.allen]

Please double-check the metrics docs:

Metrics are emitted as JSON objects to a runtime log file or over HTTP (to a service such as Apache Kafka). Metric emission is disabled by default.

The way you handle them is by setting up an emitter as per http://druid.io/docs/latest/configuration/#emitting-metrics

[Noppanit Charassinvichai]

Hi,

Currently I have the emitter running already and it looks great. I was wondering if the log will be able to capture any pending query? Correctly me if I'm wrong, currently I'm seeing this 

{"feed":"metrics","timestamp":"2016-03-02T17:19:56.628Z","service":"druid/realtime","host":"address:8084","metric":"query/time","value":4,"context":"{\"finalize\":false,\"queryId\":\"1e2f29cf-b454-4e20-9a25-ede69cdc1217\",\"timeout\":30000}","dataSource":"firehose-web","duration":"PT3600S","hasFilters":"false","id":"1e2f29cf-b454-4e20-9a25-ede69cdc1217","interval":["2016-03-02T17:00:00.000Z/2016-03-02T18:00:00.000Z"],"remoteAddress":"address","type":"timeBoundary"}

Does this mean that it only shows the finished query? 

[Fangjin Yang]

segment/scan/pending is closest to what you're looking for