Apps / Jails Crashing

[georgelza]

Hi all,

So I've had for a while now some of my jails at random times crashing... I mean I go to their url and they're dead, to get them working again I have to restart the jail.
I've always had 3 jails,
Plex
Unifi Controller
qbitorrent

As Plex has been the most important I simply put a scheduled restart of the jail in place that restarts early morning. The other 2 I mostly use so have been managing that manually, but at the moment I'm spinning up another jail and I need it to be stable so guess it's time to resolve this problem.

I originally had 16GB ram, everyone said ye it's because of RAM, so I upgraded to 32GB, problem stayed.
Re disk, have 3 pools, Apps: 1 x 128GB M.2 (single disk, no protection) , Tank; 3 x 4TB (Z1) , Bunker: 3 x 4TB (Z1)
Version: TrueNAS-12.0-U8.1

dmesg | tail -200
...
...

swap_pager_getswapspace(20): failed
swap_pager_getswapspace(18): failed
swap_pager_getswapspace(9): failed
swap_pager_getswapspace(3): failed
swap_pager_getswapspace(20): failed
swap_pager_getswapspace(18): failed
swap_pager_getswapspace(9): failed
swap_pager_getswapspace(3): failed
pid 80835 (Plex Media Server), jid 48, uid 972, was killed: out of swap space
pid 87506 (qbittorrent-nox), jid 49, uid 0, was killed: out of swap space
vnet0.48: link state changed to DOWN
epair0b: link state changed to DOWN
epair0a: Ethernet address: 02:ad:8d:37:9d:0a
epair0b: Ethernet address: 02:ad:8d:37:9d:0b
epair0a: link state changed to UP
epair0b: link state changed to UP
epair0a: changing name to 'vnet0.50'
vnet0.50: promiscuous mode enabled
lo0: link state changed to UP
in6_purgeaddr: err=65, destination address delete failed

so it's showing problem with swap... but with only 3 jails running, and 32GB RAM, do I maybe have something else configured wrong thats eating up RAM.

 

[Patrick M. Hausen]

Somehow your Plex server is eating memory until the operationg system decides to kill the process. You need to check with the Plex guys what could be a possible cause of this. Software bugs, specifically memory hungry options - there are probably a ton of configuration options for caching and the like ... definitely not a TrueNAS problem. A memory leak in your application software.

I don't run any of this, sorry.

Brute force "solution": create a cron job that does iocage restart plex every night at 3:00 am or so.

 

[georgelza]

so atm I have that brute force in place to restart plex. interesting that it's causing impact lower down, impacting other jails, luckily it seems TrueNAS is protecting it self to keep itself stable, thanks for that.
will look at maybe putting brute force in place for the other jails also, as a stop gap in the mean time.
G