Flood container huge RAM requirements?

[adyanth]

I am running Flood on Kubernetes (k3s) with a memory limit of 512MiB. I recently saw that the pod had crashed with code 139 (SIGSEGV signal 11). Upon checking the previous pod logs, I see the following:

<--- Last few GCs --->

[1:0x7f0446ccd320] 104550474 ms: Mark-sweep 252.2 (258.9) -> 252.2 (258.9) MB, 699.1 / 0.0 ms  (average mu = 0.878, current mu = 0.001) allocation failure GC in old space requested
[1:0x7f0446ccd320] 104553063 ms: Mark-sweep 252.8 (258.9) -> 252.3 (259.2) MB, 576.4 / 0.0 ms  (average mu = 0.845, current mu = 0.777) allocation failure GC in old space requested


<--- JS stacktrace --->

FATAL ERROR: Reached heap limit Allocation failed - JavaScript heap out of memory

Looks like it exceeded the memory limit upon which it crashed? Usually, when Kubernetes kills a pod for exceeding the memory limits, the error code will be 137 (signal 9 OOMKilled) and no 139.

Not sure what is causing Flood to consume more than 512 MiB of RAM.

[adyanth]

Does this look like a memory leak? @jesec can you confirm?

[jesec]

#516