I see other threads for this, but don't want to pollute individual support instances unique to an individual..

However, I've noticed recently

1. The WEB UI seems 'laggy'.. a gross measure using Chromes dev tools shows LCP's for 1.5-2.5s most of the time, going between Unraid UI sections (e.g. Settings and Tools) far from snappy. (It's OK in safe mode with the array stopped)

2. Externally, I'm using a cloudflare tunnel (using a cloudflared container) to access internal services, mainly as work (bless) won't allow tailscale (naturally)... but I have (for about 12 months) had the issue that access the server externally has websocket issues with the streaming stuff always hanging, so I can't see CPU usage graphs on the dashboard, or the array information.

Now, having moved to entirely new hardware (but retaining all disks), I've noticed that logs do say

"php-fpm[9710]: [WARNING] [pool www] server reached max_children setting (50), consider raising it"

And sure enough, there are 50 or so php-fpm processes showing in Tools->Processes..

I also have entries saying "monitor_nchan: Stop running nchan processes"

Now, I rebooted yesterday as I've still been messing around with the new server so wanted a clean session..

I noted on reboot I only had about 4 or so php_fpm processes..

Today I went shopping and whilst having some breakfast, checked in on things via the cloudflare tunnel.. I also checked in from a friends house about 30 mins ago..

On getting home and checking the logs, I notice the same 50+ php_fpm processes and a couple of nchan stopping entries in the log that do seem to coincide with me accessing the server via the cloudflare tunnel services

I am running 30+ containers, many with their own WEB UIs that are super snappy and responsive, and a Home Assistant VM, and everything otherwise is absolutely fine, the processor is largely just ticking over with 1-3% usage shown, and the main array spends most of its time spun down since I have 5 x NVME and 1 x SSD which I pretty much run all the run time stuff from.. When accessing the server externally, by and large the array stays spun down.

My gut feeling from other threads is the websockets not working correctly via the cloudflare tunnel may be causing issues and causing the many php-fpm processes, but clearly I could do with someone more knowledgeable to just cast an eye on things and make sure I'm not being daft.

I've attached diagnostics...

nas-diagnostics-20260207-1221.zip

Just experienced that today:

Feb  9 16:00:19 NASty php-fpm[11243]: [WARNING] [pool www] server reached max_children setting (50), consider raising it
Feb  9 16:05:07 NASty php-fpm[11243]: [WARNING] [pool www] child 1123945 exited on signal 9 (SIGKILL) after 288.839758 seconds from start
Feb  9 16:06:04 NASty php-fpm[11243]: [WARNING] [pool www] child 1123946 exited on signal 9 (SIGKILL) after 312.984302 seconds from start
Feb  9 16:08:57 NASty php-fpm[11243]: [WARNING] [pool www] child 1123947 exited on signal 9 (SIGKILL) after 391.020086 seconds from start

Dont know if it might be a bug or issues with the hardware, because I struggle in another topic with strange behaving unraid.

I'm having this same issue and can't figure it out for the life of me...

Forgot this topic existed.

It turned out that I ran a hytale server and that it can cause random memory leaks on high load, causing unraid to hiccup.

I had to adjust the max rendering distance in the config.json from hytale. No more issues

So potentially for you guys maybe could also have some service running that randomly increases ram usage and exhaust the server.

I’m pretty sure for myself it is something to do with hanging web sockets when accessing services via cloudflare tunnels, as every time I start accessing services via that method, the message starts appearing fairly quickly if I access using Tailscale, it never appears.