Tuning a collector to work on a t2.micro EC2 instance?
I know that it is not exactly recommended/reliable to use a 1GB/1CPU Core machine to monitor...but it seems that installing a "nano" sized collector on a t2.micro AWS instance and having it just monitor itself brings the AWS instance to a screeching halt.
I am seeing that when the collector is running, top shows that CPU pegs to 100% almost nonstop. Memory is not hit quite as bad..but it does get up there to use 500mb+
But the CPU load average is 5+ cores and it makes the system unusable. Sometimes this causes the instance to throw status alerts & even crash.
Question: Has anybody been able to tweak the wrapper.conf etc files to make the collector CPU load less demanding?
Update- I found the solution:
Edit agent.conf:
1. collector.xxxx.enable=false on every unused/unneeded datasource collection mechanism.
2. eventcollector.xxxx.enable=false on every unused/unneeded eventsource collection mechanism.
3. sbproxy.pool.connections=25
4. sbproxy.linux.threadPoolSize=60
5. collector.snmp.threadpool=25
6. discover.workers=5In my case, I pretty much am monitoring the collector machine only, and I've only left snmp, script & webpage-related collectors enabled and this seems to have done the trick.
The last few configurations might not be necessary..but everything seems to be working fine with these configs..!