This bug was initially created as a copy of Bug #1908065 I am copying this bug because: Bug #1908065 was using to track a ceph-volume fix https://github.com/ceph/ceph/pull/40201 which will not list devices as available if there are gpt headers. But even though the ceph-volume inventory returned a device as not available and ceph orch device ls returned as not available, ceph orch apply osd --all-available-devices tries to deploy on device which is not available and fails to deploy OSDs as ceph-volume will not be able to deploy OSDs on devices wit GPT labels Description of problem: Running on single node, bare-metal, "ceph orch apply osd --all-available-devices" fails to deploy devices which are reported as 'Available'. Version-Release number of selected component (if applicable): # cephadm version INFO:cephadm:Using recent ceph image docker.io/ceph/ceph:v15 ceph version 15.2.7 (88e41c6c49beb18add4fdb6b4326ca466d931db8) octopus (stable) How reproducible: yes Steps to Reproduce: 1. Follow steps here to "Bootstrap a New Cluster" Upstream Doc: https://docs.ceph.com/en/latest/cephadm/install/ 2. # ceph status ← HEALTH_WARN (no OSDs) 3. ADD OSDs # ceph orch device ls <- reports 12x HDDs AVAIL=True # ceph orch apply osd --all-available-devices ←NOTHING! 4. REVIEW LOG # ceph log last ceph-adm <...> INFO:cephadm:/bin/podman:stderr ceph-volume lvm batch: error: GPT headers found, they must be removed on: /dev/sda <...> Actual results: No OSDs get deployed Expected results: Available devices get added as OSDs Additional info: Three attachments provided: 1) cephSTATUS.txt 2) deviceLS.txt 3) cephadmLOG.txt
Sequence of operations - 1) Boot strapped cluster using - cephadm -v --image registry-proxy.engineering.redhat.com/rh-osbs/rhceph:ceph-5.0-rhel-8-containers-candidate-54312-20210519174049 bootstrap --mon-ip 10.8.129.101 --cluster-network 172.20.20.0/24 2) checked cephadm shell ceph-volume inventory and cephadm shell ceph orch device ls To make sure that all three devices on host we listed as available 3) Created gpt labels on one of the devices using parted /dev/sdd mklabel gpt and checked cephadm shell ceph-volume inventory and cephadm shell ceph orch device ls to make sure that the device with gpt label created was marked as not available. 'cephadm shell ceph orch device ls' had to be tried after intervals to see that changes updated (Did not know about --refresh) 4) Executed ceph orch apply osd --all-available-devices And observed mgr logs sudo journalctl -fu ceph-d66b64e4-b923-11eb-ac53-0cc47a6ee150.juyiug.service and observed that device with gpt label was not excluded from the list of available devices and operation failed as ceph-volume fails to configure OSD on devices with GPT headers.
doesn't block 5.0
This bug was not blocker for 5.0 , so i guess we can wait to have Cephadm agent in downstream. WDYT Sebastian?
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (Moderate: Red Hat Ceph Storage Security, Bug Fix, and Enhancement Update), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHSA-2022:5997