Interpret GC log messages - which value is pre and which is post?


(Nikita Tovstoles) #1

I see messages like below in ES server 1.7.1 log:

[2015-10-27 12:20:07,669][INFO ][monitor.jvm ] [my_host] [gc][young][67][20] duration [721ms], collections [1]/[1.7s], total [721ms]/[4s], memory [285.6mb]->[291.8mb]/[1.9gb], all_pools {[young] [27mb]->[280.4kb]/[66.5mb]}{[survivor] [8.3mb]->[8.3mb]/[8.3mb]}{[old] [250.2mb]->[283.2mb]/[1.9gb]}

Let's take the old gen: {[old] [250.2mb]->[283.2mb]/[1.9gb]} or $A -> $B / $C

is $A = value post-gc, $B = value pre-gc and $C = capacity? Or did heap allocation really grow as result of GC?

Thanks,

-nikita


(Nikita Tovstoles) #2

answering own question (after getting a cup of coffee): format is $pre -> $post / $capacity. In the above example some objects were likely promoted from young to tenured (old) gen, thus the latter did indeed increase post-GC


(system) #3