Proxmox Server (with debian LXC) Synology NAS for CIFS storage - random disconnects

[syker69]

Hi, I was previously running this config on a Rasp Pi 4b
debian os/docker/portainer with Netxcloud/Jellyfin/Sonarr etc and while it was stable and no issues, i decided to give proxmox a go, and its so much better in every way, the inbuilt backupr/estore option is a killer feature and I've successfully backed up/restored LXC to different hardware in minutes, so love that!

So the hardware
Dell optiplex 7070 MicroPC i7/16gb ram/nvme running proxmox 9.0.3
also tested same issues restoring backup to a Intel NUC / Celeron CPU / 8GB Ram / SSD
Synology DS411+II (ancient nas but healthy and used only for network storage) - DOS protection disabled in security / SMB set to Max 3/Min 2
UDM Pro Router/Switch (IDS bypassed for Proxmox host & Synology NAS)

Container is setup unprivileged
example of a mount (done in /etc/fstab) on PVE
//192.168.x.x/Media /mnt/media cifs _netdev,x-systemd.automount,noatime,uid=100000,gid=110000,dir_mode=0770,file_mode=0770,credentials=/root/.smb,users,rw,iocharset=utf8,noforceuid,noforcegid,nofail,noserververino,vers=3.0 0 0


(the last parts i only added yesterday as a test noserververino,vers=3.0 0 0 - (made no difference)



SO for the whole part its 100% working and stable, but the issue has been presenting itself while watching movies/tv via Jellyfin, some days zero issues, some days things would freeze, then "media cant be played" i then wait around 3 mins or reboot either the NAS or Proxmox and it all works again (during this time no errors or warnings logged on the nas, the nas is available and pinging fine to all other devices.

At first i thought jellyfin issue, then looking closely at the nas and experimented enabling jumbo frames on the networks / nas / switch (no change so back at default)
then i noticed random issues with Sonarr, it grabs/downloads/sorts fine most of the time, but occasionally stuff is stuck pending to import (ive now noticed that its due to the mount randomly disconnecting - and this not while playing jellyfin media
i then added a tdarr container just to verify / test media for corruption (in health check mode) - media tested fine, but again at some random times, the mount would disconnect and health check jobs would stall/need requeuing

so i don't think its a specific playback issue, its as if the NAS just becomes unavailable (when its not dropping off) - example I've left pings running from my desktop pc and file transfers running at the times when Proxmox loses the mount.

last nights test i shutdown sonarr/tdarr and disabled any scheduled tasks in Jellyfin (media library scans etc) and 3 movies played fine without issue, zero drops
this morning ran a media scan and all completed ok, ran a media segment scan (jellyfin) and after 5 mins the mount dropped off.

restarted Proxmox, media segment scan ran again, and all fine..

So its an annoying one thats hard to pin down, it does point to network drop off, but im confident the cables (excel cat5e) short 1m patch cables to each NAS/Proxmox are fine, have tried different cables / ports (no change)

The exact setup on a Rasp Pi 4b (same dockers/network/config) works flawlessly
Sonarr never hangs, Streams never stall, jellyfin Scheduled tasks never fail



Before i give up on proxmox for now and head back to Pi, just wondering if anyone has any other suggestions?
(the nas does have NFS which i was going to try, but with the unpriveleged setup of the container, and the changes needed to mounts etc im not sure its going to be worth the effort.

any know issues with this kind of config? is CIFS just buggy in proxmox? is it related to unpriveleaged?

thanks!

 

[syker69]

just to add I thought I was up to date (but i didn't have the "no subscription" repositry in, i have added & updated to 9.0.11)
same issues

 

[Kumquat]

I've been wrestling with the same exact issue. It also happens the most for me with jellyfin, but I potentially see it with any traffic to my network storage (a Synology NAS).

I've tried SMB and NFS connections, and I've tried bind mounts on LXC containers and directly mounting in VMs. Nothing has corrected this problem yet.

It *used* to work. I can't say when the issue started exactly, but I believe it was after I updated my Proxmox 8 machine recently. I have since done a fresh install of Proxmox 9 on a different machine and get the same behavior.

This thread may be the same problem, but it seems to have gone quiet: https://forum.proxmox.com/threads/s...6-14-8-2-pve-when-mounting-nfs-shares.169571/

 

[syker69]

Hey thanks for the info, yeh ive tried a few things since my last post, found some articles suggesting issues with intel nic's so i tried turning off the" tso off gso off gro off rx off tx off rxvlan off txvlan off sg off" last night but that made no difference

I then tried changing from the onboard intel nic to a usb3 nic, again no difference.

while in normal daily use it is intermittent, i can reproduce it within 30 seconds here
jellyfin media segment scan & keyframe extractor tasks, run at the same time, within 20-30 seconds the mounts are disconnected


but again in normal use, I can watch a movie, a show, no issues, running media scans never causes an issue. is so random, would really like to stick with proxmox, but appears is no fix for this.
thanks for the thread, at least i know im not alone!

 

[syker69]

well some good news i appear to of fixed it (in case helps anyone else in future)

last night i saw the thread about the 6.17 kernel (opt in) - installed / tested same issues...

then realised my LXC container was still last dist, so upgraded the repositories to Trixie and di the dist update, rebooted and not seen the issue.

just to stress test it ive had all
jellyfin tasks set to run every hour
added 10 random shows to sonarr which all downloaded/processed fine
tdarr is cpu health checking for the last 18 hours
had movie playing in 3 different rooms at the same time...

so the mounts appear now stable!, they've not dropped once since i updated the container last night.

Previously tdarr heath checks would crap out after 10mins, sonarr would often stall importing as mounts dropped, and the one killer thing is the keyframe extract task in jellyfin, would disconnect the mount in 10 seconds.... (thats running hourly last 18hours+)

so fingers crossed! looking good