This article will demonstrate the steps to replace a OSD in your Petasan cluster after you have had a disk failure.
- Remove the failed disk from the Storinator, and replace it with the new replacement disk.
The first step to replace this OSD is to delete the old OSD that no longer exists. This can be done from the Petasan dashboard.
First, go to Manage Nodes, and click on the disks where the failed OSD was hosted.
From here, click on the X on the OSD that is marked as down. It will ask you if you are sure. Click OK.
Next, it is time to create the new OSD to take its place, but first wait until Petasan marks the OSD as deleted in the Dashboard. Find the newly added Disk, and click on the plus button.
From here, you will choose what you want to add this disk as. If this is to be used as an OSD, choose to add as OSD. Next you must decide if you want to use a journal and/or cache. If you are currently using journals and cache on your other OSD’s you will want to add this one with the same. Change the journal to Enabled and the cache to Write Cache. You may leave the chosen disks to auto. This will let Petasan choose which SSD’s it will use for these, or you can specify yourself if you know specifically which drives you want to use. Once you finish, click Add.
Petasan will ask if you are sure you want to add this as an OSD. If you are sure, click OK.
Your OSD will now be added to the cluster. Once this OSD is added, your Petasan cluster will go unhealthy momentarily while Ceph rebuilds the data onto the new OSD. This is normal and will return to Healthy on its own.
Go back to the main Dashboard page, and wait for Ceph to return to Healthy. Also, if you want to SSH into any of your Petasan nodes, you can run the command “ceph osd tree” This will show you all of your OSD’s in the cluster and if they are up.