Description of problem: The main problems we have detected here are the following: https://github.com/ceph/ceph-ansible/blob/master/infrastructure-playbooks/shrink-osd.yml#L110 -In this case it seems that awk pattern it might not be the best option to use because this command will match all osd that start with the osd number. If you have more than 10 OSDs one will have problems to shrink OSDs using this method. Example: /usr/sbin/ceph-disk list | awk -v pattern=osd.1 '$0 ~ pattern {print $1}' will print all devices matching OSD names osd.1, osd.1{0..9}. [root@ceph2-5 ~]# ceph-disk list /dev/vdb1 ceph data, active, cluster ceph, osd.1, journal /dev/vdb2 /dev/vdc1 ceph data, active, cluster ceph, osd.11, journal /dev/vdc2 /dev/vdd1 ceph data, active, cluster ceph, osd.12, journal /dev/vdc2 /dev/vde1 ceph data, active, cluster ceph, osd.13, journal /dev/vdc2 /dev/vdf1 ceph data, active, cluster ceph, osd.14, journal /dev/vdc2 /dev/vdg1 ceph data, active, cluster ceph, osd.15, journal /dev/vdc2 /dev/vdh1 ceph data, active, cluster ceph, osd.16, journal /dev/vdc2 /dev/vdi1 ceph data, active, cluster ceph, osd.17, journal /dev/vdc2 /dev/vdj1 ceph data, active, cluster ceph, osd.18, journal /dev/vdc2 /dev/vdk1 ceph data, active, cluster ceph, osd.19, journal /dev/vdc2 [root@ceph2-5 ~]# ceph-disk list | awk -v pattern=osd.1 '$0 ~ pattern {print $1}' /dev/vdb1 /dev/vdc1 /dev/vdd1 /dev/vde1 /dev/vdf1 /dev/vdg1 /dev/vdh1 /dev/vdi1 /dev/vdj1 /dev/vdk1 https://github.com/ceph/ceph-ansible/blob/master/infrastructure-playbooks/shrink-osd.yml#L121 -In this case, when using NVMe devices, partitions are created using p+partition number, not just with the partition number. sda --> sda1 nvme0n1 --> nvme0n1p1 So, in this case this line when using NVMe devices should be name: "ceph-osd@{{ item.0.stdout[:-2] | regex_replace('/dev/', '') }}" Version-Release number of selected component (if applicable):RHCS 3.x Additional info: Additional BZ's for NVME devices with ceph-ansible. Should this be resolved in 3.0z1 as part of Ansible 3.0.25 as well? Same issue in the purge-docker-cluster.yml --- https://bugzilla.redhat.com/show_bug.cgi?id=1547999 should be fixed in 3.0z1 Install with NVME fails for the same reason --- https://bugzilla.redhat.com/show_bug.cgi?id=1541016 should be fixed in 3.0z1 Install of NVME in container fails -- https://bugzilla.redhat.com/show_bug.cgi?id=1537980 should be fixed in 3.0z1
*** This bug has been marked as a duplicate of bug 1561456 ***