Need help with Rutorrent crashing unraid webui

[RIDGID]

Background: I have a pretty basic unraid server for radarr/sonarr >> Rutorrent (linuxserver docker) >> Plex running on an i7-6700k / 16gb. Pretty much everything I know about unraid is thanks to Spaceinvaderone.

 

I made the critical mistake of adding some very large lists to my radarr library in monitored status. About 1000 torrents got queued up in rutorrent (which was already seeding around 3k torrents). Things seemed to be okay for a while, around 200 files successfully downloaded and process via radarr. Rutorrent goes unresponsive, 502 gateway, whole server running like molasses, CPU/RAM both at 100% according to the dashboard. I restart the container but after a few hours it become inaccessible again. System log shows:

Nov 15 20:42:01 DefineR5 kernel: Out of memory: Kill process 5960 (rtorrent main) score 781 or sacrifice child
Nov 15 20:42:01 DefineR5 kernel: Killed process 5960 (rtorrent main) total-vm:14805204kB, anon-rss:12587032kB, file-rss:0kB, shmem-rss:0kB

After it kills the process, the server is still hung up on something that only stopping the container will fix. I check logs in appdata, the most relevant one at 
appdata\rutorrent\log\php\error.txt:

[17-Nov-2019 07:00:00] NOTICE: fpm is running, pid 11005
[17-Nov-2019 07:00:00] NOTICE: ready to handle connections
[17-Nov-2019 19:06:51] WARNING: [pool www] server reached pm.max_children setting (5), consider raising it
[17-Nov-2019 19:15:43] WARNING: [pool www] server reached pm.max_children setting (5), consider raising it
[17-Nov-2019 19:25:04] WARNING: [pool www] server reached pm.max_children setting (5), consider raising it
[17-Nov-2019 19:40:17] WARNING: [pool www] server reached pm.max_children setting (5), consider raising it
[17-Nov-2019 20:02:24] NOTICE: Terminating ...
[17-Nov-2019 20:02:27] NOTICE: exiting, bye-bye!
[18-Nov-2019 04:07:56] NOTICE: fpm is running, pid 313
[18-Nov-2019 04:07:56] NOTICE: ready to handle connections
[18-Nov-2019 15:19:47] WARNING: [pool www] server reached pm.max_children setting (5), consider raising it

I tried to raise the pm.max_children setting to 20 by editing the www2.conf file but that failed with the following error:

[22-Nov-2019 02:30:47] ERROR: [pool www] the chdir path '/usr/local/emhttp' does not exist or is not a directory

[22-Nov-2019 02:30:47] ERROR: failed to post process the configuration

[22-Nov-2019 02:30:47] ERROR: FPM initialization failed

I am bit out of my depth here. As it stands I have 800 partially downloaded torrents (most 20-99% complete), many of which throw 

 

hash check on download completion found bad chunks, consider using \"safe_sync\".

 

when checked, despite having pieces.sync.always_safe.set=1 at the bottom of my otherwise stock config.

 

Even with most of the torrents stopped it keeps crashing. Twice now it has also taken the entire unraid webui with it, and I have had to ssh in to reboot which then requires a 20+hr parity check. 

 

I feel like I am probably missing something simple, but it increasingly seems like an uphill battle and I am officially lost.