Ceph

I am writing to report an issue that occurred with our Ceph cluster after adding two new machines, one MDS (Metadata Server), one MGR (Manager), and several OSDs (Object Storage Devices) one each new machine. Following the addition of these new components, I made changes to the ceph.conf configuration file.

Subsequently, I restarted the Ceph service on the existing instances within the Ceph cluster, which resulted in the cluster becoming temporarily unresponsive and caused an unexpected increase in client load.

We request your assistance in addressing the following concerns:

Ceph Cluster Unresponsiveness: After restarting the Ceph service on the existing instances, there was a period during which the cluster became unresponsive.

High Client Load: The restart of the Ceph service caused an unexpected increase in client load, affecting the performance of other services relying on the Ceph cluster. Although the cluster is now responsive, we would appreciate assistance in optimizing the client load and ensuring its stability.

Below are the relevant details of our Ceph environment:

Ceph Cluster Configuration (ceph.conf): Added mon initial members and mon host to the [global] section and the mds on the bottom of the file

Ceph Cluster Components:
Before the add : 5 MDS / 2 active, 5 MON, 5 MGR and 13 OSD
After the add : 7MDS / 2 active, 7 mon, 5MGR and 21 osd (only 13 Up during the crash)

We appreciate your attention to this matter and any assistance you can provide in optimizing the Ceph cluster's performance and stability. If there are any additional details or logs required from our side, please let us know, and we will provide them promptly.