am using Kapacitor and InfluxDB on the same box. I am running around 12 tick scripts for alert generation with 12 subscriptions across 5 databases. I have total 297 measurements but only 5-6 are used in TICk script for alert generation. Average points written are 4.5K per second.
I am using n1standard8 box from GCP to run this with 8 core and 30G RAM. InfluxDB takes around 10G of RAM average, but Kapacitor goes crazy with almost 20-22G and then crash out.
Following is the one of TICK script -
var data = stream
| from()
.measurement ( measurememt1)
.groupby( 'tag1','tag2','tag3')
|where( lambda: cond1 )
|where(lambda: cond2)
|eval (lambda : “field1”)
.as( 'value')
var alert1 = data
|where( lambda: “tag1” == “someval1” )
|alert()
.crit(lambda: "value" == "somefailed" )
.id("xxxxx")
.details("xxx")
.email("[someemail1@company.com](mailto:someemail1@company.com)")
.log()
var alert2 = data
|where( lambda: “tag1” == “someval2” )
|alert()
.crit(lambda: "value" == "somefailed" )
.id("xxxxx")
.details("xxx")
.email("[someemail2@company.com](mailto:someemail2@company.com)")
.log()
var alert13= data
|where( lambda: “tag1” == “someval3” )
|alert()
.crit(lambda: "value" == "somefailed" )
.id("xxxxx")
.details("xxx")
.email("[someemail3@company.com](mailto:someemail3@company.com)")
.log()
Am writing bad tickscript to make kapacitor crazy to eat memory ? please guide . Will making use of windows node bring down this consumption. I am afraid because it might miss some real points which needs to be alerted.
