My PMS jail died over night - trying to figure out why

[gribbler]

I moved my freenas plex server into a dmz and I thought I was struggling with firewall rules when it turns out my plex install is having some issues. I went to bed last night and everything was working - except I can't get the library to update - I looked at logs a bit last night but it was late and I wanted to get some sleep. I woke up this morning and I went to the plex web interface and made a couple minor changes - I'm pretty certain those won't be the issue - I noticed it seemed slow to respond so I went to the server settings and it just spun. I then started looking at the firewall thinking that was the issue, but through the logs I could see it was passing traffic. I then ssh'd out to my friends server to use telnet to connect into the PMS port, where it connected fine. I could telnet from my LAN to port 80 but to 32400 it was getting refused. Digging around further, I could see my NFS mounted media from the LAN to the DMZ was fine, so I thought I'd restart the service via inside the jail. Then I started getting an error that I'd seen a few times before in the last couple days that I thought was related to the firewall rules again, I'm not sure if it's an issue with the network stack or something else. Here's what I see:

root@plexmediaserver_1:/ # /usr/local/etc/rc.d/plexmediaserver stop
ps: /boot/kernel/kernel: No such file or directory
plexmediaserver not running? (check /var/run/plex/plex.pid).
root@plexmediaserver_1:/ # ping google.com
ping: cannot resolve google.com: Host name lookup failure

I can ping localhost, but nothing on the network, I get a network is unreachable error. Sorry I probably spent too much time explaining how I got there, when I want to find out why the jail is in the state it is. I've tried to stop and start the jail but it errors.

service jail onestop plexmediaserver_1
jexec 4 /usr/pbi/plexmediaserver-amd64/control stop
Stoppingthejail....Unmounting user-supplied file-systems
umount: /mnt/storage/jails/plexmediaserver_1//mnt/Movies: not a file system root directory
umount: /mnt/storage/jails/plexmediaserver_1//mnt/TV_Shows: not a file system root directory
Stopping jail with: /etc/rc.shutdown
.Terminating on signal 15
..Doneroot@xfreenas:/mnt
Yet the jail is still up.

The umounts above are NFS mounts from the lan to the dmz, which is mounted read-only to the freenas host via a init script, then mounted using the Jails -> storage

The system is a few months old HPE Proliant gen9, it's only got 4gb of RAM for now, I will pick up either 8 or 16 more, it's just pricey here in Canada, running FreeNAS-11.1-U4. I have FreeNAS running on the LAN where I have sonarr and all the other tools for media acquiring, hence the not so worried right now about the 4gb of RAM

Currently networking system is dead though when I first got up hours ago (PST) it was pingable, and ports were forwarding from the wan - I can't recall now for sure if they were from the lan -

I've since restarted, everything has come ok - if this happens again what further info can I get to you here?

 

[Jailer]

it's only got 4gb of RAM for now

That's half the required minimum RAM to run just FreeNAS, let alone plex on top of that. You'll need at least 8GB more of you want to run Plex on FreeNAS. Until then expect unexplained issues such as you are having as it's one of the side effects of not having enough RAM.

 

[gribbler]

OK fine, but it does seem like a lot of Unix's can run with the less than - I've ordered 8gb should be here soon, but:

Mem: 387M Active, 767M Inact, 758M Wired, 1900M Free

I think it seems a odd that my memory usage is extremely light, I'm direct playing not transcoding. I have thousands of systems at work that run at 95-99% memory usage for the whole time they render and they keep chugging. I get that if I was transcoding streams, or had a lot of disk activity - I've got a 64gb ssd for the system and a 1tb spinning disk that the only thing I've got on it, is the 1 jail. It's just a system in the dmz for this 1 single purpose. Hence the 16tb, 32gb of RAM dual xeon for the system on the lan. I can install a number lighter OS's that can handle this spec and not "fall apart". I've run thousands of FreeBSD/Linux/Unix systems without seeing this sort of failure - I will get the 12gb in there soon and see how it is, I can NFS mount over the lan to my playback machine so it's not a show stopper.

Thanks for the input - appreciated!