The build script in the repo over in bug 1122754 sets the payload size to 8MB (increased from 64KB) -- we can change it there as needed. See: https://github.com/mreid-moz/data-pipeline/blob/master/bin/build_pipeline_heka.sh#L20
Status: NEW → RESOLVED
Last Resolved: 4 years ago
OS: Mac OS X → All
Hardware: x86 → All
Resolution: --- → FIXED
fyi: in Heka's default configuration this could consume over 3.2GiB of memory. If we set it any larger we should reduce the message pool size.
What are the implications of reducing the pool size?
It all depends on the configuration and data profile but the worst cases: 1) reduced throughput 2) reduced number of messages that the decoder can output in a 1 to N case (hangs if exceeded)
You need to log in before you can comment on or make changes to this bug.