electrode development workflow frequently leaves zombie processes behind

[cdaringe]

Type of issue

bug

My environment

os: mac, many versions
node/npm: many versions

Expected behavior

no zombie processes

Current behavior

• electrode processes crash periodically, or are exited manually, then
• electrode leaves PIDs running for the internal proxy, wds, or application server

Steps to reproduce the behavior

• use electrode in dev mode
• actively develop code that causes processes to crash (non-deterministic)
• re-run electrode, observe EADDRINUSE

Command line output

example:

[proxy] {"level":50,"time":1583866298115,"pid":43308,"hostname":"m-c02x6042jgh7","name":"redbird","code":"EADDRINUSE","errno":-48,"syscall":"listen","address":"::","port":3000,"msg":"Server Error","stack":"Error: listen EADDRINUSE: address already in use :::3000

[jchip]

These two released PRs should've fix the zombies
#1531
#1542

Do you still see them? Do you have a specific steps to repo?

[cdaringe]

@jchip, it happens pretty regularly in our repo ATM. All it takes is starting and editing. Generally a TS compiler error from Babel triggers it somewhat reliably. We can get a DM going and I can hook you up