Bug #21122
closedKraken to luminous upgrade: Error EINVAL: key for mgr.vpm037 exists but cap mds does not match
0%
Description
Steps:
1. Installed stable version of kraken using ceph-deploy and setup a 4 node cluster
2. To workaround issue http://tracker.ceph.com/issues/20950 , key is pushed to disk as below (and the jewel -> luminous upgrade test works fine )
2017-08-24T21:44:45.156 INFO:teuthology.orchestra.run.vpm037:Running: 'sudo ceph auth get client.bootstrap-mgr | sudo tee /var/lib/ceph/bootstrap-mgr/ceph.keyring' 2017-08-24T21:44:45.457 INFO:teuthology.orchestra.run.vpm037.stderr:exported keyring for client.bootstrap-mgr 2017-08-24T21:44:45.471 INFO:teuthology.orchestra.run.vpm037.stdout:[client.bootstrap-mgr] 2017-08-24T21:44:45.474 INFO:teuthology.orchestra.run.vpm037.stdout: key = AQDMSJ9ZBSlOMxAA1Fje8hHXRaitROQK8vnTHw== 2017-08-24T21:44:45.477 INFO:teuthology.orchestra.run.vpm037.stdout: caps mon = "allow profile bootstrap-mgr" 2017-08-24T21:44:45.481 INFO:teuthology.orchestra.run.vpm001:Running: 'sudo ceph auth get client.bootstrap-mgr | sudo tee /var/lib/ceph/bootstrap-mgr/ceph.keyring' 2017-08-24T21:44:45.806 INFO:teuthology.orchestra.run.vpm001.stderr:exported keyring for client.bootstrap-mgr 2017-08-24T21:44:45.854 INFO:teuthology.orchestra.run.vpm001.stdout:[client.bootstrap-mgr] 2017-08-24T21:44:45.858 INFO:teuthology.orchestra.run.vpm001.stdout: key = AQDMSJ9ZBSlOMxAA1Fje8hHXRaitROQK8vnTHw== 2017-08-24T21:44:45.860 INFO:teuthology.orchestra.run.vpm001.stdout: caps mon = "allow profile bootstrap-mgr" 2017-08-24T21:44:45.872 INFO:teuthology.orchestra.run.vpm125:Running: 'sudo ceph auth get client.bootstrap-mgr | sudo tee /var/lib/ceph/bootstrap-mgr/ceph.keyring' 2017-08-24T21:44:46.252 INFO:teuthology.orchestra.run.vpm125.stderr:exported keyring for client.bootstrap-mgr 2017-08-24T21:44:46.259 INFO:teuthology.orchestra.run.vpm125.stdout:[client.bootstrap-mgr] 2017-08-24T21:44:46.305 INFO:teuthology.orchestra.run.vpm125.stdout: key = AQDMSJ9ZBSlOMxAA1Fje8hHXRaitROQK8vnTHw== 2017-08-24T21:44:46.308 INFO:teuthology.orchestra.run.vpm125.stdout: caps mon = "allow profile bootstrap-mgr"
3. ceph-deploy mgr create is then run to create the mgr, at this stage it fails with below error
2017-08-24T21:44:55.428 INFO:teuthology.orchestra.run.vpm037:Running: 'cd /home/ubuntu/cephtest/ceph-deploy && ./ceph-deploy mgr create vpm037 vpm001' 2017-08-24T21:44:55.725 INFO:teuthology.orchestra.run.vpm037.stderr:[ceph_deploy.conf][DEBUG ] found configuration file at: /home/ubuntu/.cephdeploy.conf 2017-08-24T21:44:55.729 INFO:teuthology.orchestra.run.vpm037.stderr:[ceph_deploy.cli][INFO ] Invoked (1.5.38): ./ceph-deploy mgr create vpm037 vpm001 2017-08-24T21:44:55.731 INFO:teuthology.orchestra.run.vpm037.stderr:[ceph_deploy.cli][INFO ] ceph-deploy options: 2017-08-24T21:44:55.734 INFO:teuthology.orchestra.run.vpm037.stderr:[ceph_deploy.cli][INFO ] username : None 2017-08-24T21:44:55.736 INFO:teuthology.orchestra.run.vpm037.stderr:[ceph_deploy.cli][INFO ] verbose : False 2017-08-24T21:44:55.739 INFO:teuthology.orchestra.run.vpm037.stderr:[ceph_deploy.cli][INFO ] mgr : [('vpm037', 'vpm037'), ('vpm001', 'vpm001')] 2017-08-24T21:44:55.753 INFO:teuthology.orchestra.run.vpm037.stderr:[ceph_deploy.cli][INFO ] overwrite_conf : False 2017-08-24T21:44:55.756 INFO:teuthology.orchestra.run.vpm037.stderr:[ceph_deploy.cli][INFO ] subcommand : create 2017-08-24T21:44:55.758 INFO:teuthology.orchestra.run.vpm037.stderr:[ceph_deploy.cli][INFO ] quiet : False 2017-08-24T21:44:55.760 INFO:teuthology.orchestra.run.vpm037.stderr:[ceph_deploy.cli][INFO ] cd_conf : <ceph_deploy.conf.cephdeploy.Conf instance at 0x7f0fd3c31cb0> 2017-08-24T21:44:55.763 INFO:teuthology.orchestra.run.vpm037.stderr:[ceph_deploy.cli][INFO ] cluster : ceph 2017-08-24T21:44:55.765 INFO:teuthology.orchestra.run.vpm037.stderr:[ceph_deploy.cli][INFO ] func : <function mgr at 0x7f0fd44b0c08> 2017-08-24T21:44:55.768 INFO:teuthology.orchestra.run.vpm037.stderr:[ceph_deploy.cli][INFO ] ceph_conf : None 2017-08-24T21:44:55.785 INFO:teuthology.orchestra.run.vpm037.stderr:[ceph_deploy.cli][INFO ] default_release : False 2017-08-24T21:44:55.787 INFO:teuthology.orchestra.run.vpm037.stderr:[ceph_deploy.mgr][DEBUG ] Deploying mgr, cluster ceph hosts vpm037:vpm037 vpm001:vpm001 2017-08-24T21:44:55.792 INFO:teuthology.orchestra.run.vpm037.stderr:[vpm037][DEBUG ] connection detected need for sudo 2017-08-24T21:44:55.798 INFO:teuthology.orchestra.run.vpm037.stderr:[vpm037][DEBUG ] connected to host: vpm037 2017-08-24T21:44:55.804 INFO:teuthology.orchestra.run.vpm037.stderr:[vpm037][DEBUG ] detect platform information from remote host 2017-08-24T21:44:55.840 INFO:teuthology.orchestra.run.vpm037.stderr:[vpm037][DEBUG ] detect machine type 2017-08-24T21:44:56.021 INFO:teuthology.orchestra.run.vpm037.stderr:[ceph_deploy.mgr][INFO ] Distro info: Ubuntu 16.04 xenial 2017-08-24T21:44:56.033 INFO:teuthology.orchestra.run.vpm037.stderr:[ceph_deploy.mgr][DEBUG ] remote host will use systemd 2017-08-24T21:44:56.044 INFO:teuthology.orchestra.run.vpm037.stderr:[ceph_deploy.mgr][DEBUG ] deploying mgr bootstrap to vpm037 2017-08-24T21:44:56.047 INFO:teuthology.orchestra.run.vpm037.stderr:[vpm037][DEBUG ] write cluster configuration to /etc/ceph/{cluster}.conf 2017-08-24T21:44:57.254 INFO:teuthology.orchestra.run.vpm037.stderr:[vpm037][DEBUG ] create path if it doesn't exist 2017-08-24T21:44:58.196 INFO:teuthology.orchestra.run.vpm037.stderr:[vpm037][INFO ] Running command: sudo ceph --cluster ceph --name client.bootstrap-mgr --keyring /var/lib/ceph/bootstrap-mgr/ceph.keyring auth get-or-create mgr.vpm037 mon allow profile mgr osd allow * mds allow * -o /var/lib/ceph/mgr/ceph-vpm037/keyring 2017-08-24T21:44:58.349 INFO:teuthology.orchestra.run.vpm037.stderr:[vpm037][ERROR ] Error EINVAL: key for mgr.vpm037 exists but cap mds does not match
Updated by Nicolas Drufin over 6 years ago
I have the same error when I migrate my cluster and create mgr
Updated by John Spray over 6 years ago
- Status changed from New to Closed
This kind of looks like the mgr daemon with that name just already existed?
I'm going to close this as the bootstrap-mgr issue has been fixed elsewhere, so I suspect the sequence of actions (including workarounds) that led to this might not exist any more.
Updated by Vasu Kulkarni over 6 years ago
- Status changed from Closed to New
Hi John,
The issue from Kraken(Stable) -> Luminous(dev) is still not fixed, I have removed the workaround of recreating the mgr key which was required before but not in the recent dev. The same steps works for Jewel(Stable) -> Luminous(dev) but not for Kraken to Luminous
2017-11-21T21:26:12.938 INFO:teuthology.orchestra.run.vpm019:Running: 'cd /home/ubuntu/cephtest/ceph-deploy && ./ceph-deploy mgr create vpm019 vpm181' 2017-11-21T21:26:13.179 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.conf][DEBUG ] found configuration file at: /home/ubuntu/.cephdeploy.conf 2017-11-21T21:26:13.180 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.cli][INFO ] Invoked (1.5.39): ./ceph-deploy mgr create vpm019 vpm181 2017-11-21T21:26:13.180 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.cli][INFO ] ceph-deploy options: 2017-11-21T21:26:13.181 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.cli][INFO ] username : None 2017-11-21T21:26:13.181 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.cli][INFO ] verbose : False 2017-11-21T21:26:13.181 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.cli][INFO ] mgr : [('vpm019', 'vpm019'), ('vpm181', 'vpm181')] 2017-11-21T21:26:13.210 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.cli][INFO ] overwrite_conf : False 2017-11-21T21:26:13.210 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.cli][INFO ] subcommand : create 2017-11-21T21:26:13.210 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.cli][INFO ] quiet : False 2017-11-21T21:26:13.210 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.cli][INFO ] cd_conf : <ceph_deploy.conf.cephdeploy.Conf instance at 0x7f87c38dccb0> 2017-11-21T21:26:13.210 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.cli][INFO ] cluster : ceph 2017-11-21T21:26:13.210 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.cli][INFO ] func : <function mgr at 0x7f87c415cd70> 2017-11-21T21:26:13.210 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.cli][INFO ] ceph_conf : None 2017-11-21T21:26:13.210 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.cli][INFO ] default_release : False 2017-11-21T21:26:13.211 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.mgr][DEBUG ] Deploying mgr, cluster ceph hosts vpm019:vpm019 vpm181:vpm181 2017-11-21T21:26:13.221 INFO:teuthology.orchestra.run.vpm019.stderr:[vpm019][DEBUG ] connection detected need for sudo 2017-11-21T21:26:13.244 INFO:teuthology.orchestra.run.vpm019.stderr:[vpm019][DEBUG ] connected to host: vpm019 2017-11-21T21:26:13.245 INFO:teuthology.orchestra.run.vpm019.stderr:[vpm019][DEBUG ] detect platform information from remote host 2017-11-21T21:26:13.300 INFO:teuthology.orchestra.run.vpm019.stderr:[vpm019][DEBUG ] detect machine type 2017-11-21T21:26:13.300 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.mgr][INFO ] Distro info: Ubuntu 16.04 xenial 2017-11-21T21:26:13.301 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.mgr][DEBUG ] remote host will use systemd 2017-11-21T21:26:13.301 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.mgr][DEBUG ] deploying mgr bootstrap to vpm019 2017-11-21T21:26:13.301 INFO:teuthology.orchestra.run.vpm019.stderr:[vpm019][DEBUG ] write cluster configuration to /etc/ceph/{cluster}.conf 2017-11-21T21:26:13.301 INFO:teuthology.orchestra.run.vpm019.stderr:[vpm019][DEBUG ] create path if it doesn't exist 2017-11-21T21:26:13.306 INFO:teuthology.orchestra.run.vpm019.stderr:[vpm019][INFO ] Running command: sudo ceph --cluster ceph --name client.bootstrap-mgr --keyring /var/lib/ceph/bootstrap-mgr/ceph.keyring auth get-or-create mgr.vpm019 mon allow profile mgr osd allow * mds allow * -o /var/lib/ceph/mgr/ceph-vpm019/keyring 2017-11-21T21:26:13.622 INFO:teuthology.orchestra.run.vpm019.stderr:[vpm019][ERROR ] Error EINVAL: key for mgr.vpm019 exists but cap mds does not match 2017-11-21T21:26:13.623 INFO:teuthology.orchestra.run.vpm019.stderr:[vpm019][ERROR ] exit code from command was: 22 2017-11-21T21:26:13.623 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.mgr][ERROR ] could not create mgr 2017-11-21T21:26:13.749 INFO:teuthology.orchestra.run.vpm019.stderr:Warning: Permanently added 'vpm181,172.21.2.181' (ECDSA) to the list of known hosts. 2017-11-21T21:26:13.964 INFO:teuthology.orchestra.run.vpm019.stderr:[vpm181][DEBUG ] connection detected need for sudo 2017-11-21T21:26:14.092 INFO:teuthology.orchestra.run.vpm019.stderr:Warning: Permanently added 'vpm181,172.21.2.181' (ECDSA) to the list of known hosts. 2017-11-21T21:26:14.310 INFO:teuthology.orchestra.run.vpm019.stderr:[vpm181][DEBUG ] connected to host: vpm181 2017-11-21T21:26:14.310 INFO:teuthology.orchestra.run.vpm019.stderr:[vpm181][DEBUG ] detect platform information from remote host 2017-11-21T21:26:14.362 INFO:teuthology.orchestra.run.vpm019.stderr:[vpm181][DEBUG ] detect machine type 2017-11-21T21:26:14.372 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.mgr][INFO ] Distro info: Ubuntu 16.04 xenial 2017-11-21T21:26:14.372 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.mgr][DEBUG ] remote host will use systemd 2017-11-21T21:26:14.372 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.mgr][DEBUG ] deploying mgr bootstrap to vpm181 2017-11-21T21:26:14.372 INFO:teuthology.orchestra.run.vpm019.stderr:[vpm181][DEBUG ] write cluster configuration to /etc/ceph/{cluster}.conf 2017-11-21T21:26:14.373 INFO:teuthology.orchestra.run.vpm019.stderr:[vpm181][DEBUG ] create path if it doesn't exist 2017-11-21T21:26:14.376 INFO:teuthology.orchestra.run.vpm019.stderr:[vpm181][INFO ] Running command: sudo ceph --cluster ceph --name client.bootstrap-mgr --keyring /var/lib/ceph/bootstrap-mgr/ceph.keyring auth get-or-create mgr.vpm181 mon allow profile mgr osd allow * mds allow * -o /var/lib/ceph/mgr/ceph-vpm181/keyring 2017-11-21T21:26:14.695 INFO:teuthology.orchestra.run.vpm019.stderr:[vpm181][ERROR ] Error EINVAL: key for mgr.vpm181 exists but cap mds does not match 2017-11-21T21:26:14.695 INFO:teuthology.orchestra.run.vpm019.stderr:[vpm181][ERROR ] exit code from command was: 22 2017-11-21T21:26:14.696 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy.mgr][ERROR ] could not create mgr 2017-11-21T21:26:14.696 INFO:teuthology.orchestra.run.vpm019.stderr:[ceph_deploy][ERROR ] GenericError: Failed to create 2 MGRs