I'm using jstat to collect jvm metrics and push them to graphite through collectd. A few days ago i noticed that i had "spikes" in the graphs regarding Old Space, but those were not triggering any full GCs, which lead me to thinking that the metric was off. So i started looking at jstat's output, searching for such a metric to pop and here it is :
jstat -gccause 123456 1s
S0 S1 E O M CCS YGC YGCT FGC FGCT GCT LGCC GCC
27,81 0,00 49,52 63,18 65,05 48,04 7146 703,497 20 10,564 714,061 Allocation Failure No GC
27,81 0,00 63,65 63,18 65,05 48,04 7146 703,497 20 10,564 714,061 Allocation Failure No GC
27,81 16,89 100,00 99,93 65,05 48,04 7147 703,497 20 10,564 714,061 Allocation Failure Allocation Failure
0,00 23,40 10,34 63,19 65,05 48,04 7147 703,621 20 10,564 714,185 Allocation Failure No GC
0,00 23,40 35,39 63,19 65,05 48,04 7147 703,621 20 10,564 714,185 Allocation Failure No GC
As you can see, my Old space is increasing normally and just "peaks" to 99.93%, not triggering any full GC and returns to a normal metric right after that.
So i'm wondering if something's off or if it's just an incorrect iteration.
Thanks !