Using drip with logstash and systemd

[ruskotron]

Hi, not so much a bug, as an enquiry.

I've been trying to optimise my startup times, and I was trying out drip, and I was noticing a slight increase in startup time which I thought was weird.

Logstash makes using drip easy since all you need specify is the USE_DRIP environment variable.

Upon closer investigation I noticed that a restart was killing all the processes, not just the active one basically meaning that drip has to reinitialise once more, and spin up the extra jvm, which explains the extra delay.

I think this is to do with systemd killing all processes associated with the logstash services as well as the logstash runtime. Curiously this also happens when I kill the active process itself, so presumably systemd is monitoring that somehow.

I suppose this is strictly speaking a "logstash" issue, since it's they who supply the USE_DRIP feature, but I can imagine them saying "it's not for that" (running under systemd), which is fair enough I suppose.

Just wondering if you have any thoughts or suggestions as to how I might address this issue?

I might see if I can figure it out myself.

Thanks!
Rob