fuel 6.x ceph low disk space warning on controller

[Baboune]

Hi,

Due to partitioning on the controller node, the ceph directory is located in a small partition. In a situation where the controller node has 700gb, the following is automatically created:
/dev/mapper/os-root 50G 38G 12G 77% /
/dev/sda3 196M 54M 141M 28% /boot
/dev/mapper/mongo-mongodb
                      668G 25G 610G 4% /var/lib/mongo

The result is this warning:
ceph status
    cluster 40b4db6d-beaf-4bef-b0a4-e4e1ac12e33d
     health HEALTH_WARN mon.node-55 low disk space
     monmap e1: 1 mons at {node-55=192.168.40.4:6789/0}, election epoch 1, quorum 0 node-55
     osdmap e666: 31 osds: 31 up, 31 in
      pgmap v41095: 2496 pgs, 12 pools, 40578 MB data, 5157 objects
            143 GB used, 25907 GB / 26050 GB avail
                2496 active+clean
[root@node-55 ~]# cd /var/lib/c

node-55 is the controller.

There are different possible solutions for this. I suppose one possible implementation would be to have a dedicated sizeable ceph partition. A quick fix is to ensure the mongodb gets less space by default in the Fuel controller.

[Vladimir Sharshov (vsharshov)]

Covered by blueprint, that is why moved to 8.0: https://blueprints.launchpad.net/fuel/+spec/volume-manager-refactoring after discussion with Alexander Gordeev

[Alexander Kislitsky (akislitsky)]

We passed SCF in 8.0. Moving the bug to 9.0.