Grobalt

Hi, i cannot mount remote drives any more - i am doing that for 2 years and nothing changed, today after reboots the mount button stays greyed out. the VPN connection is stable, i can ssh to the remote server - but i cannot mount the other servers exported nfs drives. if i disable nfs export and i add the drive i click on load shares - none appear. then i activate nfs again, click on my server on load share and all nfs shares are shown. so i select it and the "drive" appears but mount is greyed out. i uninstalled Unassigned Devices (and plus) and installed again, no changes when i mount the share manually : mount -t nfs 10.253.1.2:/mnt/user/filme /mnt/remotes/alex -o nolock,rw mount.nfs: Connection refused for 10.253.1.2:/mnt/user/filme on /mnt/remotes/alex on remote server: root@Tower:~# exportfs -rav exporting *:/mnt/user/rips exporting *:/mnt/user/filme unraid-diagnostics-20240829-1422.zip

sure, but it is a "central" script and you have like a config file where you exclude the "not wanted" LXCs all good

it is one of the "base" scripts on the proxmox helper scrips side: https://tteck.github.io/Proxmox/ https://tteck.github.io/Proxmox/#proxmox-ve-cron-lxc-updater This script will add/remove a crontab schedule that updates all LXCs every Sunday at midnight. To exclude LXCs from updating, edit crontab (crontab -e) and add CTID as shown in the example (-s 103 111)

ok ... would have been great in combination with the lxc updater etc

Thanks for your collaboration, that would be awesome if all promox helper scripts can be used an easy way soon!

is it possible to create / install lxc containers with the proxmox VE helper scripts ? on promox they run flawless but here on a "naked new lxc container" it seems they are not working: https://tteck.github.io/Proxmox/#unifi-network-server-lxc bash -c "$(wget -qLO - https://github.com/tteck/Proxmox/raw/main/ct/unifi.sh)" __ __ _ _____ / / / /__ (_) __(_) / / / / __ \/ / /_/ / / /_/ / / / / / __/ / \____/_/ /_/_/_/ /_/ Loading... /dev/fd/63: line 516: whiptail: command not found ⚠ User exited script

that will not work as the movie scraper store the pictures in the same folder as the movie - otherwise it would be super easy to create an flash only share. the mover worked and moved all files above 14M size to array.

fresh reinstalled and a run with test mode: Jul 24 21:19:17 unraid plugin-manager: ca.mover.tuning.plg installed Jul 24 21:20:03 unraid kernel: mdcmd (37): nocheck pause Jul 24 21:20:03 unraid kernel: md: recovery thread: exit status: -4 Jul 24 21:20:06 unraid emhttpd: shcmd (1338): /usr/local/sbin/mover start |& logger -t move & Jul 24 21:20:06 unraid move: Log Level: 1 Jul 24 21:20:06 unraid move: /tmp/Mover Directory does not exist. Creating it Jul 24 21:20:06 unraid move: mover: started Jul 24 21:20:10 unraid flash_backup: adding task: /usr/local/emhttp/plugins/dynamix.my.servers/scripts/UpdateFlashBackup update Jul 24 21:20:13 unraid move: mover: finished on my cache disk are around 400GB to be moved .. dont know what the test mode should report - i expected filenames to be moved i am not using "age" as criteria, i am using size above 14M to keep media library pictures on the flash drives

./test.sh Creating Soft link for 7.0 or newer since /usr/local/sbin/move is now located in /usr/libexec/unraid/move

no changes with this one: https://github.com/R3yn4ld/ca.mover.tuning/blob/794d8b0fad8ea1c2db8096dbd753ecf75c8a6e22/plugins/ca.mover.tuning.plg First test yes - then with test no:

i removed the old and installed this version from the community apps (not manually with a link from the forum). Jul 22 20:38:27 unraid emhttpd: shcmd (329893): /usr/local/sbin/mover start |& logger -t move & Jul 22 20:38:27 unraid kernel: mdcmd (37): set md_write_method 1 Jul 22 20:38:27 unraid kernel: Jul 22 20:38:28 unraid move: Log Level: 1 Jul 22 20:38:28 unraid move: mover: started Jul 22 20:38:29 unraid move: /usr/local/emhttp/plugins/ca.mover.tuning/age_mover: line 260: /usr/local/sbin/move: No such file or directory *repeated About 20 times* Jul 22 20:38:30 unraid move: mover: finished

Die meisten hier schauen schon auf Stromverbrauch - diese alten Mellanoxkarten brauchen mehr Strom als mancher ganze Server. Nur falls Du da noch nicht drüber gestolpert bist

no answer for your question but maybe an alternative - all the diy builds lacked ilo or ipmi - so i tried pikvm in the last days and .. damn .. runs out of the box seriously good i bought early last year a PiKVM v3 BliKVM for a raspberry that was laying around and finally tried it out as i had some freezes with the jellyfin bugged image weeks ago.

does not need to be a mover feature - the mover is just a workaround how new files are handled. it could be a base feature for all files to define storage pools based on performance/technology. For movie streaming spindles are still perfect - for browsing the library flash is a huge benefit. Maybe a bit like Lustre can do it, there is a feature data on metadata which is basically .. metadata of files is stored on flash, but files below a defined size are stored on the flash portion of the lustre solution as well. https://wiki.lustre.org/Data_on_MDT_Solution_Architecture

no, the request was not for mover tuning plugin - mover is base unraid - the request was to enrich the mover with the capability or .. a policy that handles small files different than large files, mover independant

With RC7 beta it seems we need to disable mover tuning plugin to work flawless - that means my setup i use today will be broken. Today new files are created on the cache disk and only files larger than size x are moved to array. This prevents spnning of the disks and enables quick access to pictures, movie poster etc. Would be nice to have move by filesize or "all" mover Tuning capabilities to unraid mover base.

so a few hours late i played around with Wireguard MTU sizes and it improved a bit, but not good enough to play 4k movies without stuttering. Then i stumbled over the info that WireGuard can benefit a lot by AVX512 support which my server supports, but me friend's doesnt (10th Gen Intel vs. AVX512 introduced 11th Gen). Will update after hardware updates

My friend has a very good internet connection and we are connected via WireGuard. Today i found out that for nice large files WireGuard is heavily throttling the speed to around 40-60 Mbps while he has 300 Mbps upload and 600 Mbps download. The speedtests on his side are validated and not the issue - same on my side - and when i copy files for example via the Anydesk servers it is like twice of WireGuard VPN. What is the best way to connect the 2 unraid servers without WireGuard throttling ? Both sides IPv6 und just unassigned devices (i was not succesful mounting an IPv6 share).

The limitation of threads seems not to work on this linuxserver.io container (jellyfin developer statement at github). So if someone has issues with too many ffprobe threads in parallel, switch to official repository.

My issue is "solved" - caused by Jellyfin full library scan and a ridiculous default setting

In case anyone fails to scan his library and or crashes the whole unraid server .. what a mess... i found the issue. I installed jellyfin - and it does not matter which container, bbergle or linuxserver or whatever , all behaved same. My movie collection is around 1k 4k movies and a few hundred lower resolution. Each time i scanned my movie folder, my unraid server became unresponsive and even root login was no more possible, after typing in user in console it did not even promp for password (timed out). I tried local as well via KVM, same. So i did memtest, all stuff, my hardware is fine. Next step was going by single folders instead all at once - worked better but still freezing sometimes and needed hard reset, awful. After a while i found out that it may caused be out of memory as once unraid freezed with 100% mem usage. So i added a swapfile but did not help. Next step was to limit the memory of the container --memory=12G and then by surprise, it uses the swap file on top. so i set a swap file to 32G just to test .. after 20 minutes of scan it used up the container 12G + 32G swap ... but it did not immediately crash (it crashed after 20 minutes utilizing all swap). So i had time to dig into the container and found ~300 ffprobe processes ... like endless processes to scan media in parallel. I am sure spinning disks like that a lot ... what a shit. So i stumbled over the (default) parameter JELLYFIN_FFmpeg__analyzeduration=200M which is like how many frames per movies are analyzed for aspect ratio, codecs etc .... So i reduced this a lot setting a new environment variable to 3M instead 200M -> voila, scan rushed through but failed on many codeds. Then increased to 10M .. again, it used 12GB container mem and 10GB swap ... what a horrible setup. Next step was a hint in the jellyfin forum, to limit the "Parallel library scan tasks limit" - seems they introduced this in jellyfin 10.9 and thd default is "" (maybe endless ... killing all disk based storage systems). Prevous to 10.9 it was like 1 scan parallel and it just took a while but it worked. so i set this now to 4 and check what happens ... it looks promising. I filed an issue on github to change the default parameter of parallel scans.. or many many spinning disk setups will fail with full lib scans https://github.com/jellyfin/jellyfin/issues/12203 *edit* default parallel scans is the logical core count .. in my case 20 ... no wonder it crashes

All OK, but when the freeze appears there IS No interesting entry

This as well, correct. Additional to parity speed limitation the network is obviously faulty as well.

ok, that helps a lot - added now the mover tuning plugin to limit which files are moved and i can copy the small media files back to cache Thanks !

Hi all, last night my unraid somewhen became unresponsive via webinterface, ssh login not possible - local console login not possible as well (after typign in username it did not prompt for pass -> timeout after a while). So i enabled syslog server but i cannot find anything odd in the log. Server crashed 3x after that within minutes. Any i idea ? unraid-diagnostics-20240624-0745.zip