ceph / ceph-medic

find common issues in ceph clusters
MIT License
22 stars 18 forks source link

`oc` is not aliased to `openshift` #114

Closed alfredodeza closed 5 years ago

alfredodeza commented 5 years ago

This is a problem, remoto supports it, there is no reason why we can't make it work.

In addition, when no hosts are able to connect, explain what connection method was used, this is already a thing for openshift, but should be as well for other methods.

djgalloway commented 5 years ago
$ cat /home/jenkins-build/build/workspace/ocs4-metal3/18/ceph-medic.conf 
[global]
--log-path = /home/jenkins-build/build/workspace/ocs4-metal3/18
deployment_type = oc

[check]
--ignore = ECOM8,WOSD2
[jenkins-build@bruuni008 18]$ ceph-medic --config=/home/jenkins-build/build/workspace/ocs4-metal3/18/ceph-medic.conf check --ignore ECOM8,WOSD2
Collection failed!                                                                                
--> RuntimeError: All nodes failed to connect. Cannot run any checks
[jenkins-build@bruuni008 18]$ sed -i "s/deployment_type.*/deployment_type = openshift/g" /home/jenkins-build/build/workspace/ocs4-metal3/18/ceph-medic.conf
[jenkins-build@bruuni008 18]$ ceph-medic --config=/home/jenkins-build/build/workspace/ocs4-metal3/18/ceph-medic.conf check --ignore ECOM8,WOSD2
Host: rook-ceph-mgr-a-6755457dcb-qt87b          connection: [connected  ]          
alfredodeza commented 5 years ago

This is weird because we do the translation already at https://github.com/ceph/ceph-medic/blob/master/ceph_medic/connection.py#L28-L29

Maybe a unit test or walking through the code while it figures it out is what we need to find out why this isn't working