ChatNoir

Then using the Unassigned Devices plugin.

Please setup a syslog server and post the file along your diagnostics after the next crash.

Thanks for that, but I do not see any MACVLAN or any other issues, I was just answering on your thread title about RAM increase. And on that topic, it is still higher (though not problematic with my usecase). Here is the view from the last 90 days. The main culprit seems to be Plex being around 300MB previously to 1 to 11GB now. Maybe the --no-healthcheck command could help, I'll look into it (maybe) when I have some vacation time at home.

@capt.shitface

Hi, not really. One advantage would be that HBAs generally support more drives. But otherwise a good SATA controller should work as well. Can't vouch for that model as the manufacturer does not indicate what controller is used. From the images, it does not seem to use a port multiplier. It might be visible in your diagnostics though.

Docker containers are split in two major elements : the elements that can change (settings, DBs, etc.), that's on your appdata and those that do not change, that's in the docker image For performance reasons, it is generally better to have both on SSDs and so on a Pool. And they would stay on the Pool with the settings above. Note that with 6.12, the wording and organisation of those settings in Shares has changed. While the ideas behind the choice are the same, the things are organized differently and might be easier to understand. (the Exclusive access part is optional)

Did you check the /logs directory of your flash drive ? Any command line diagnostics should be stored there.

Maybe for 2.5 inch drives, for 3.5 it was OK the last time I shucked ones

Ca semble lié au plugin Unraid Connect. Tu auras plus de chance d'avoir des réponses sur le sous-forum officel : https://forums.unraid.net/forum/94-connect-plugin-support/ (par contre, c'est en anglais)

Moved to the German section, if you want to post in the general forum, please use English.

Changed Status to Closed Changed Priority to Other

This is not a bug, it's the way it is designed. The configuration is stored in a cookie on your browser. So a different browser can be configured differently depending on your use case or resolution.

Are you still seeing the problem on 6.12.2 ? Because nothing will be changed on 6.11.x

Not really, Flash is the used size of the flash drive, etc.

RAM usage is only the first line, the last line is about you docker image size if I am not mistaking.

did you

Don't hesitate to report trolls so moderators and administrators can act on it. It does not guarantee a moderation action but we don't see every message and one or several reports can make us reconsider our initial judgement on some users messages and behavior.

Il te faut activer le syslog server pour avoir des journaux persistants (sinon ils sont en RAM comme le reste de l'OS). Tu peux faire un miroir sur ta clé USB mais uniquement temporairement le temps de débuggé, ce n'est pas conseillé à long terme pour la durée de vie de la clé.

Et si tu mets 'Automatically split any directory as required' ? Après, possible que Rclone crée d'abord tous les fichiers à 0 octets (donc sur un seul disque vu que ça ne prend pas de place) avant de les écrire vraiment ? Si c'est ça, Unraid remplace toujours un fichier où le trouve initialement ce qui pourrait créer ton problème.

Et même pour les pools multi-disques, le BTRFS est beaucoup plus flexible de mon point de vue que le ZFS. Par exemple, mon cache est un RAID10 sur lequel je pourrais rajouter facilement un 5e disque (de même taille ou différente) et utiliser toute sa capacité sans perte (selon la taille utilisée). En ZFS, je pense que je devrais ajouter 4 nouveaux disques. Perso je vais rester en XFS pour l'Array et BTRFS pour les pools tant que je n'aurais pas de vrai bénéfice. Et sinon, aucun problème sur la 6.12.

Bizarre @dam448, ça marchait très bien sous 6.11.5 et maintenant sur 6.12.2. Par contre, ça ne protège que des suppressions depuis le réseau sur les dossiers partagés en SMB.

@royce Contact support

https://unraid.net/download

I can confirm on that on FF114. Clearing cached data on the browser did not change the behavior. Same thing on Chrome.

Not sure to see a significant improvement from 6.12.2. It seems that some containers are more impacted than others (Plex, Jellyfin for example for me) On the other end, I am wondering it is an actual RAM usage problem or an issue of reporting from docker and/or kernel interpretation. (my table above uses 'docker_container_mem' ) On the same time period than the image above (15 days), my system RAM usage does not seem to be fluctuating much while docker is supposed to have reached ~27GB from an average of 8-9GB on 6.11. My total RAM is 64GB, so going from 9 to 27 should be clearly visible on the system. The Unraid Dashboard is consistant with my Grafana dashboard.