From c1782c50042cb271665ec357625b33f23a363323 Mon Sep 17 00:00:00 2001
From: "Joshua M. Boniface" <joshua@boniface.me>
Date: Fri, 28 Apr 2023 10:48:28 -0400
Subject: [PATCH] Add full/nearfull OSD health detection

---
 daemon-common/cluster.py | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/daemon-common/cluster.py b/daemon-common/cluster.py
index 645dc165..ce448e22 100644
--- a/daemon-common/cluster.py
+++ b/daemon-common/cluster.py
@@ -51,6 +51,8 @@ def getClusterHealth(zkhandler, node_list, vm_list, ceph_osd_list):
         "vm_stopped": 10,
         "osd_out": 50,
         "osd_down": 10,
+        "osd_full": 50,
+        "osd_nearfull": 10,
         "memory_overprovisioned": 50,
         "ceph_err": 50,
         "ceph_warn": 10,
@@ -110,6 +112,18 @@ def getClusterHealth(zkhandler, node_list, vm_list, ceph_osd_list):
                 f"cluster: Ceph OSD {ceph_osd['id']} in {up_texts[ceph_osd['stats']['up']].upper()} state"
             )
 
+        # Handle full or nearfull OSDs (>85%)
+        if ceph_osd["stats"]["utilization"] >= 90:
+            cluster_health_value -= health_delta_map["osd_full"]
+            cluster_health_messages.append(
+                f"cluster: Ceph OSD {ceph_osd['id']} is FULL ({ceph_osd['stats']['utilization']:.1f}% > 90%)"
+            )
+        elif ceph_osd["stats"]["utilization"] >= 85:
+            cluster_health_value -= health_delta_map["osd_nearfull"]
+            cluster_health_messages.append(
+                f"cluster: Ceph OSD {ceph_osd['id']} is NEARFULL ({ceph_osd['stats']['utilization']:.1f}% > 85%)"
+            )
+
     # Check for (n-1) overprovisioning
     #   Assume X nodes. If the total VM memory allocation (counting only running VMss) is greater than
     #   the total memory of the (n-1) smallest nodes, trigger this warning.