Other problems

This article introduces the problems and solutions encountered during the use of Rainbond.

Insufficient disk space

When the disk space exceeds 80%, the application cannot be deployed normally

When the disk root partition space exceeds 80%, Kubernetes will automatically enter the eviction state, and all Pods will be evicted. At that time, Rainbond will not work normally and the disk space needs to be cleaned up.

Clean up resources no longer used by Docker:

  1. docker system prune

Clean up the images stored in the Rainbond mirror warehouse. If you use the default mirror warehouse provided by Rainbond, please refer to Clean up the rbd-hub mirror.

Disk has space, not enough storage

The server system disk still has enough space, but when looking at the storage on Rainbond, the disk is not enough. This is because Rainbond will create an nfs-provisioner by default to dynamically create PVCs, and this nfs-provisioner will be fixed on a certain node, and the default storage directory is /opt/rainbond.

Query the node where nfs-provisioner is located:

  1. $ kubectl get pod -l name=nfs-provisioner -n rbd-system -o wide
  2. NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
  3. nfs-provisioner-0 1/1 Running 1 (13h ago) 6d21h 10.42.0.228 192.168.3.33 <none> <none>

View the storage directory of nfs-provisioner:

  1. du -sh /opt/rainbond

You can increase the disk space for the /opt/rainbond directory, or mount /opt/rainbond to other disks through soft links.