Error processing is wrong. No throwable catch should be, but message should be sent to parent about error and parent should relaunch child. Probably it's better to use separate process for it.
Now worker crashing handled in proper way, but now design sucks because backend doesn't restart worker if it's crashed. This job is complex and I return to it later.
Error processing is wrong. No throwable catch should be, but message should be sent to parent about error and parent should relaunch child. Probably it's better to use separate process for it.