Plex Server stops, jail still show as up

[jbrown705]

Hi @HoneyBadger, I do already have a thread pertaining to this. It started from a Plex issues and has continued to the swap issue because it seems to be that it is suspected it is part of the cause of this.

 

[HoneyBadger]

Hi @HoneyBadger, I do already have a thread pertaining to this. It started from a Plex issues and has continued to the swap issue because it seems to be that it is suspected it is part of the cause of this.

Thanks for tagging me in here.

Based on this:

I am running a few jails, Plex, Sonarr, Radarr, Sabnzbd, Homebridge, Tautulli, Organizr2, Nginx. I also have nzbget and jackett, in jails, but not running.

And this:

I turned off my Ubuntu VM and it had a pretty dramatic effect on the swap memory. I suspect this is the main culprit, but don't know how to fix it.

You're running a lot of additional, non-storage load, and FreeNAS out of the box isn't expecting to have to share memory with any plugins, let alone eight. And since you're on an older Xeon E3, you're at your memory max of 32GB - so you'll have to start picking and choosing what to prioritize or offload to another host.

Your arc_max on a 32GB machine is probably in the area of 30GB - which is fighting for memory with eight jails/plugins and a bhyve VM. That ends in first "swap" when you're out of memory, and then "dead process" when the swap runs out.

I would go straight to a shell prompt and fire up the line:

Code:

sysctl vfs.zfs.arc_max

to see what your current arc_max is - and then after writing it down, cut it down to 16GB with

Code:

sysctl vfs.zfs.arc_max=17179869184

 

[jbrown705]

I would go straight to a shell prompt and fire up the line:

Code:

sysctl vfs.zfs.arc_max

to see what your current arc_max is - and then after writing it down, cut it down to 16GB with

Code:

sysctl vfs.zfs.arc_max=17179869184

Ok, thank you so much for the help. This is all new to me and I was not aware that was too much. I thought with 32GB RAM I would be ok. Still have a lot to learn I guess.

Here is the output from the arc_max:

Code:

vfs.zfs.arc_max: 23748816896

I ran the command you said and now the output matches what you listed. Do I need to reboot or anything? I should also note that the VM is only there to run Ombi and I don't do anything else with it.

I'm not sure how to pick and choose what to offload. If i had to list the jails in order of priority, I would say Plex, Jotta, Sabnzbd, Sonarr/Radarr, Tautulli, Homebridge, Nginx, Ombi/VM, Organizr. (If that helps)

 

[HoneyBadger]

You shouldn't need to reboot - the memory change will take effect immediately. Check your reporting pane in the GUI, or swapinfo or top from a shell to see if it has reduced the used portion.

If Ombi can run in a jail, that will be more efficient memory-wise than a bhyve VM.

As far as the amount of jails to pick and choose - you no longer have the option of "add more RAM" and have to go with "use less" - perhaps each individual plugin can be tuned, for example by reducing concurrent downloads or adding other limits in the newsgroup/torrent scrapers, lowering Plex transcode quality.

But if all else fails, you're going to have to either move some of them to another physical host, or do your best Thanos impression and snap a few of them out of existence.

 

[jbrown705]

You shouldn't need to reboot - the memory change will take effect immediately. Check your reporting pane in the GUI, or swapinfo or top from a shell to see if it has reduced the used portion.

If Ombi can run in a jail, that will be more efficient memory-wise than a bhyve VM.

As far as the amount of jails to pick and choose - you no longer have the option of "add more RAM" and have to go with "use less" - perhaps each individual plugin can be tuned, for example by reducing concurrent downloads or adding other limits in the newsgroup/torrent scrapers, lowering Plex transcode quality.

But if all else fails, you're going to have to either move some of them to another physical host, or do your best Thanos impression and snap a few of them out of existence.

I did see it drop a bit, but not much. However, I should note that Jotta is doing an initial cloud upload of my entire library, so I have to guess that is a a big draw on resources.

Right now, it doesn't look like there is support for Ombi v3 on FreeNAS, which is why i am running in a VM. I powered off the VM for now to help with the situation.

It sounds like the best solution is to upgrade my mobo and process so it can handle what i want to do. Does that make more sense? If so, do you have a recommendation on what I should look for? I can start to monitor eBay and Amazon for used stuff and see if I can get a better setup?



Thanks!

 

[jbrown705]

 

[file_haver]

This is happening to me too, plex server actually works but i'm trying to move the jail from a hard drive based volume to a SSD based volume and i can't get the plex jail to stop long enough to actually move the jail mount point. 11.1-u6

 

[jbrown705]

to either move some of them to another physical host, or do your best Thanos impression and snap a few of them out of existence.

@HoneyBadger, someone else suggested to me that I may be able to add a SSD (32-64gb) and have it dedicated to swap. They said there is a dip switch or jumper pin? Does that sound right? If that is possible I’ll gladly buy another SSD for that. I have room.