This message was deleted.
# longhorn-storagea
adamant-kite-43734
05/31/2023, 12:32 AMThis message was deleted.
q
quiet-area-89381
05/31/2023, 12:32 AMthat node has scheduling enabled though
quiet-area-89381
05/31/2023, 12:34 AMit seems to complain about disk pressure
Copy code
Last Transition Time: 6 days ago
Message: the disk default-disk-f4fc945ae2cd6216(/var/lib/longhorn/) on the node has 49807360000 available, but requires reserved 62378975232, minimal 25% to schedule more replicas
Reason: DiskPressure
Status: Falsequiet-area-89381
05/31/2023, 12:46 AMis there some possible undeleted files of volumes that may have crash or other things taking space I could remove manually??
quiet-area-89381
05/31/2023, 12:59 AMI guess pruning container images helps sometimes... Microk8s/containerd have been keep those too long it seems.
b
blue-kitchen-51801
05/31/2023, 6:35 AMi’m not sure if I understand correctly your problem, but keep in mind that Longhorn is storing all the changes on disk (for example if you have a folder of 1 GB and you’re updating a 500 MB file, Longhorn sometimes might store on disk 1.5 GB, this is needed for snapshots/backups etc)
For more information please check out this link https://longhorn.io/docs/1.4.2/volumes-and-nodes/volume-size/
👍 1
f
faint-sunset-36608
06/01/2023, 6:52 PMI have found that the interaction between Longhorn's reserved capacity requirement and kubelet's garbage collection thresholds can be less than ideal. In general, I agree with @blue-kitchen-51801, but if you are using relatively small nodes and pulling lots of different images, you might also run into a need to prune container images. See my discussion topic at https://github.com/longhorn/longhorn/discussions/5962 for more details.
q
quiet-area-89381
06/01/2023, 6:56 PMThanks for the pointer! It seems to be related. We are indeed using a lot of different images and we were not doing a good job at pruning fast enough.
144 Views