Add metrics for tracking cache hit ratios

sample-config/pgmon-metrics.yml

@@ -19,10 +19,12 @@ metrics:
     type: value
     query:
       0: SHOW server_version_num
+
   max_frozen_age:
     type: row
     query:
-      0: SELECT max(age(datfrozenxid)) AS xid_age, max(mxid_age(datminmxid)) AS mxid_age FROM pg_database
+      0: SELECT max(age(datfrozenxid)) AS xid_age, 0 AS mxid_age FROM pg_database
+      90600: SELECT max(age(datfrozenxid)) AS xid_age, max(mxid_age(datminmxid)) AS mxid_age FROM pg_database
 
   # Per-database metrics
   db_stats:
@@ -33,6 +35,11 @@ metrics:
     test_args:
       dbname: postgres
 
+  hit_ratios:
+    type: row
+    query:
+      0: SELECT sum(heap_blks_read)::float / sum(heap_blks_read + heap_blks_hit) AS avg_heap_hit_ratio, sum(idx_blks_hit)::float / sum(idx_blks_read + idx_blks_hit) AS avg_idx_hit_ratio, sum(toast_blks_hit)::float / sum(toast_blks_read + toast_blks_hit) AS avg_toast_hit_ratio, sum(tidx_blks_hit)::float / sum(tidx_blks_read + tidx_blks_hit) AS avg_tidx_hit_ratio FROM pg_statio_all_tables
+
   # Debugging
   ntables:
     type: value

src/pgmon.py

@@ -656,7 +656,11 @@ def test_queries():
         # If the metric has arguments to use while testing, grab those
         args = metric.get("test_args", {})
         # Run the query without the ability to retry.
-        res = sample_metric(dbname, name, args, retry=False)
+        try:
+            res = sample_metric(dbname, name, args, retry=False)
+        except MetricVersionError:
+            print("{} -> Unsupported for this version".format(name))
+            continue
         # Compare the result to the provided sample results
         # TODO
         print("{} -> {}".format(name, res))
@@ -714,6 +718,7 @@ class SimpleHTTPRequestHandler(BaseHTTPRequestHandler):
             except LatestVersionCheckError as e:
                 log.error("Failed to retrieve latest version information: {}".format(e))
                 self._reply(503, "Failed to retrieve latest version info")
+            return
 
         # Note: parse_qs returns the values as a list.  Since we always expect
         #       single values, just grab the first from each.

zabbix_templates/pgmon_templates.yaml

@@ -673,6 +673,98 @@ zabbix_export:
                   value: '{#DBNAME}'
                 - tag: Type
                   value: Calculated
+            - uuid: bb0690e123e94e35a9bbd0f05ee322b4
+              name: 'Database Hit Ratios for {#DBNAME}'
+              type: HTTP_AGENT
+              key: 'pgmon_db_hit_ratios[{#DBNAME}]'
+              history: '0'
+              value_type: TEXT
+              trends: '0'
+              url: 'http://localhost:{$AGENT_PORT}/hit_ratios'
+              query_fields:
+                - name: dbname
+                  value: '{#DBNAME}'
+              tags:
+                - tag: Application
+                  value: PostgreSQL
+                - tag: Database
+                  value: '{#DBNAME}'
+            - uuid: e0edbaf6f98249a6b357ade2ab41a76a
+              name: 'Average heap hit ratio on {#DBNAME}'
+              type: DEPENDENT
+              key: 'pgmon_db_hit_ratio[heap,{#DBNAME}]'
+              delay: '0'
+              history: 90d
+              value_type: FLOAT
+              description: 'The cache (shared_buffers) hit ratio for heap entries for the given database'
+              preprocessing:
+                - type: JSONPATH
+                  parameters:
+                    - $.avg_heap_hit_ratio
+              master_item:
+                key: 'pgmon_db_hit_ratios[{#DBNAME}]'
+              tags:
+                - tag: Application
+                  value: PostgreSQL
+                - tag: Database
+                  value: '{#DBNAME}'
+            - uuid: e4956d433a47455ca0c00d67a0f59355
+              name: 'Average index hit ratio on {#DBNAME}'
+              type: DEPENDENT
+              key: 'pgmon_db_hit_ratio[index,{#DBNAME}]'
+              delay: '0'
+              history: 90d
+              value_type: FLOAT
+              description: 'The cache (shared_buffers) hit ratio for index entries for the given database'
+              preprocessing:
+                - type: JSONPATH
+                  parameters:
+                    - $.avg_idx_hit_ratio
+              master_item:
+                key: 'pgmon_db_hit_ratios[{#DBNAME}]'
+              tags:
+                - tag: Application
+                  value: PostgreSQL
+                - tag: Database
+                  value: '{#DBNAME}'
+            - uuid: 0335267c14a8410a80a16b38167efcb3
+              name: 'Average toast hit ratio on {#DBNAME}'
+              type: DEPENDENT
+              key: 'pgmon_db_hit_ratio[toast,{#DBNAME}]'
+              delay: '0'
+              history: 90d
+              value_type: FLOAT
+              description: 'The cache (shared_buffers) hit ratio for heap entries for the given database'
+              preprocessing:
+                - type: JSONPATH
+                  parameters:
+                    - $.avg_toast_hit_ratio
+              master_item:
+                key: 'pgmon_db_hit_ratios[{#DBNAME}]'
+              tags:
+                - tag: Application
+                  value: PostgreSQL
+                - tag: Database
+                  value: '{#DBNAME}'
+            - uuid: 3f7da45e351443528c6c36145d38f9b8
+              name: 'Average toast index hit ratio on {#DBNAME}'
+              type: DEPENDENT
+              key: 'pgmon_db_hit_ratio[toast_index,{#DBNAME}]'
+              delay: '0'
+              history: 90d
+              value_type: FLOAT
+              description: 'The cache (shared_buffers) hit ratio for toast index entries for the given database'
+              preprocessing:
+                - type: JSONPATH
+                  parameters:
+                    - $.avg_tidx_hit_ratio
+              master_item:
+                key: 'pgmon_db_hit_ratios[{#DBNAME}]'
+              tags:
+                - tag: Application
+                  value: PostgreSQL
+                - tag: Database
+                  value: '{#DBNAME}'
             - uuid: 492b3cac15f348c2b85f97b69c114d1b
               name: 'Database Stats for {#DBNAME}'
               type: HTTP_AGENT