Ubuntu VM Resource leakage

[jonathan-m-hamilton]

Set up the simple-assert-transfer example with all the containers running on the same ubuntu 1804 vm. One of the JVM processes took 100% on all 8 CPU cores and filled up the ram in a couple hours of operation.

[jonathan-m-hamilton]

Reproduction:

• Examples/Simple Asset Quorum
• npm quorum
• observe resource spiking

Is it "default" behavior with Quorum? Can this be validated by running less nodes e.g. 4 as default v0.1?

@denis-yu-glotov will check resources pre-post run to validate.

[denis-yu-glotov]

I am running our current Quorum network (v2.1.1, 4 nodes) for 16 hours at Ubuntu 16.04 vm with 4 cpu and 6 Gb RAM. Now htop shows that CPU cores are uniformly 10-40% busy (with spikes to 70%). Memory consumption is 1.87 - 1.92 Gb and stays constantly in this range.
Going to try the latest Quorum (v2.3.0) sample-network to see if it is less resource-greedy.

[denis-yu-glotov]

For Quorum v2.3.0, 4 nodes, same other factors, after 45 minutes of running I see memory consumption 1.55 - 1.7 Gb, CPU cores are busy for 20-70% uniformly.

[petermetz]

Thank you for the investigation @denis-yu-glotov
Issue must be on my side. Best guess is that it's specific to how Hyper-V manages RAM for the VM.
For me RAM usage shoots up to 20+GB in a matter of minutes just by having npm run corda and the CPUs are pegged at a solid 100% usage constantly. Will try to look for other culprits in my own setup, but it's clear that Corda is not at fault.

[denis-yu-glotov]

Final note about Quorum 2.3.0 after 20 hours, memory consumption is 1.67 - 1.72 Gb. So from this side, it is not significantly better, then our current Quorum 2.1.1. Although I noticed that tessera (v0.10.0) has started significantly faster then ours (v0.7.3). So it is good to upgrade although not a first priority.