We should try an existing solution instead of rolling our own.
terminate doesn't work, but tree-kill or similar might.
Desired behavior/Acceptance criteria (returned value, expected error, performance expectations, etc.)
root@3ed8d7dddc2f:/srv/service# ps aux USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND root 1 0.6 0.2 1105088 74724 ? Ssl 16:15 0:01 node server.js root 17 0.1 0.0 4608 3712 pts/0 Ss 16:18 0:00 /bin/bash root 25 0.1 0.1 724296 41088 ? Ssl 16:18 0:00 node /srv/service/executor-classes/javaScriptWasmSubprocess.js root 36 0.0 0.0 2576 1536 ? S 16:18 0:00 /bin/sh -c . /srv/service/programming-languages/wasmedge/wasmedge-binary/env && wasmedge --enable-all-statistics --dir .:. --dir /executo root 37 99.9 0.1 12746860 42556 ? Sl 16:18 0:14 wasmedge --enable-all-statistics --dir .:. --dir /executors/javascript-wasmedge:./executors/javascript-wasmedge /srv/service/programming- root 45 0.0 0.0 8480 4352 pts/0 R+ 16:18 0:00 ps aux root@3ed8d7dddc2f:/srv/service# ps aux USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND root 1 0.7 0.2 1172928 77528 ? Ssl 16:15 0:01 node server.js root 17 0.1 0.0 4608 3712 pts/0 Ss 16:18 0:00 /bin/bash root 36 0.0 0.0 0 0 ? Z 16:18 0:00 [sh] <defunct> root 37 97.0 0.0 0 0 ? Z 16:18 0:15 [wasmedge] <defunct> root 50 0.0 0.0 8480 4480 pts/0 R+ 16:18 0:00 ps aux
tree-kill was not powerful enough to handle zombies
ps-node also fails, in an even worse way: it kills the two parent processes but leaves the wasmedge process itself running (in a non-zombie state, even).
nexpect might be useful to clean up the current solution, but I think the current strategy (kill the bottom-most process) might be the best we can do (without turning Docker into a full-on VM)
Is this being over-taken by the Rust work? I assume so, but let's leave this open until confirmed.
It is, indeed. Also, I've tried tree-kill, terminate, and a couple of other libraries. It appears that our solution is more stable (for our use case) than anything available in npm world. Closing.