mixin: Remove support for removed cache metrics (#9674)

Remove backwards compatibility support for `thanos_memcached_` prefixed metrics that were removed over a year ago in favor of `thanos_cache_` metrics. Signed-off-by: Nick Pillitteri <[email protected]>
grafana · Oct 18, 2024 · 7a69789 · 7a69789
1 parent d534195
commit 7a69789
Show file tree

Hide file tree

Showing 10 changed files with 374 additions and 272 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -54,6 +54,7 @@
 
 ### Mixin
 
+* [CHANGE] Remove backwards compatibility for `thanos_memcached_` prefixed metrics in dashboards and alerts removed in 2.12. #9674
 * [ENHANCEMENT] Unify ingester autoscaling panels on 'Mimir / Writes' dashboard to work for both ingest-storage and non-ingest-storage autoscaling. #9617
 * [BUGFIX] Dashboards: Fix autoscaling metrics joins when series churn. #9412 #9450 #9432
 * [BUGFIX] Alerts: Fix autoscaling metrics joins in `MimirAutoscalerNotActive` when series churn. #9412

diff --git a/...oring-values-generated/mimir-distributed/templates/metamonitoring/grafana-dashboards.yaml b/...oring-values-generated/mimir-distributed/templates/metamonitoring/grafana-dashboards.yaml
diff --git a/...amonitoring-values-generated/mimir-distributed/templates/metamonitoring/mixin-alerts.yaml b/...amonitoring-values-generated/mimir-distributed/templates/metamonitoring/mixin-alerts.yaml
@@ -119,14 +119,10 @@ spec:
             expr: |
               (
                 sum by(cluster, namespace, name, operation) (
-                  rate(thanos_memcached_operation_failures_total{operation!="add"}[1m])
-                  or
                   rate(thanos_cache_operation_failures_total{operation!="add"}[1m])
                 )
                 /
                 sum by(cluster, namespace, name, operation) (
-                  rate(thanos_memcached_operations_total{operation!="add"}[1m])
-                  or
                   rate(thanos_cache_operations_total{operation!="add"}[1m])
                 )
               ) * 100 > 5

diff --git a/operations/mimir-mixin-compiled-baremetal/alerts.yaml b/operations/mimir-mixin-compiled-baremetal/alerts.yaml
@@ -107,14 +107,10 @@ groups:
           expr: |
             (
               sum by(cluster, namespace, name, operation) (
-                rate(thanos_memcached_operation_failures_total{operation!="add"}[1m])
-                or
                 rate(thanos_cache_operation_failures_total{operation!="add"}[1m])
               )
               /
               sum by(cluster, namespace, name, operation) (
-                rate(thanos_memcached_operations_total{operation!="add"}[1m])
-                or
                 rate(thanos_cache_operations_total{operation!="add"}[1m])
               )
             ) * 100 > 5

diff --git a/operations/mimir-mixin-compiled-baremetal/dashboards/mimir-reads.json b/operations/mimir-mixin-compiled-baremetal/dashboards/mimir-reads.json
diff --git a/operations/mimir-mixin-compiled/alerts.yaml b/operations/mimir-mixin-compiled/alerts.yaml
@@ -107,14 +107,10 @@ groups:
           expr: |
             (
               sum by(cluster, namespace, name, operation) (
-                rate(thanos_memcached_operation_failures_total{operation!="add"}[1m])
-                or
                 rate(thanos_cache_operation_failures_total{operation!="add"}[1m])
               )
               /
               sum by(cluster, namespace, name, operation) (
-                rate(thanos_memcached_operations_total{operation!="add"}[1m])
-                or
                 rate(thanos_cache_operations_total{operation!="add"}[1m])
               )
             ) * 100 > 5

diff --git a/operations/mimir-mixin-compiled/dashboards/mimir-reads.json b/operations/mimir-mixin-compiled/dashboards/mimir-reads.json
diff --git a/operations/mimir-mixin/alerts/alerts.libsonnet b/operations/mimir-mixin/alerts/alerts.libsonnet
@@ -207,14 +207,10 @@ local utils = import 'mixin-utils/utils.libsonnet';
           expr: |||
             (
               sum by(%(group_by)s, name, operation) (
-                rate(thanos_memcached_operation_failures_total{operation!="add"}[%(range_interval)s])
-                or
                 rate(thanos_cache_operation_failures_total{operation!="add"}[%(range_interval)s])
               )
               /
               sum by(%(group_by)s, name, operation) (
-                rate(thanos_memcached_operations_total{operation!="add"}[%(range_interval)s])
-                or
                 rate(thanos_cache_operations_total{operation!="add"}[%(range_interval)s])
               )
             ) * 100 > 5

diff --git a/operations/mimir-mixin/dashboards/dashboard-utils.libsonnet b/operations/mimir-mixin/dashboards/dashboard-utils.libsonnet
@@ -1198,13 +1198,6 @@ local utils = import 'mixin-utils/utils.libsonnet';
       $.queryPanel(
         |||
           sum by(operation) (
-            # Backwards compatibility
-            rate(thanos_memcached_operations_total{
-              %(jobMatcher)s,
-              component="%(component)s",
-              name="%(cacheName)s"
-            }[$__rate_interval])
-            or ignoring(backend)
             rate(thanos_cache_operations_total{
               %(jobMatcher)s,
               component="%(component)s",
@@ -1219,8 +1212,7 @@ local utils = import 'mixin-utils/utils.libsonnet';
     )
     .addPanel(
       $.timeseriesPanel('Latency (getmulti)') +
-      $.backwardsCompatibleLatencyPanel(
-        'thanos_memcached_operation_duration_seconds',
+      $.latencyPanel(
         'thanos_cache_operation_duration_seconds',
         |||
           {
@@ -1237,13 +1229,6 @@ local utils = import 'mixin-utils/utils.libsonnet';
       $.queryPanel(
         |||
           sum(
-            # Backwards compatibility
-            rate(thanos_cache_memcached_hits_total{
-              %(jobMatcher)s,
-              component="%(component)s",
-              name="%(cacheName)s"
-            }[$__rate_interval])
-            or
             rate(thanos_cache_hits_total{
               %(jobMatcher)s,
               component="%(component)s",
@@ -1252,13 +1237,6 @@ local utils = import 'mixin-utils/utils.libsonnet';
           )
           /
           sum(
-            # Backwards compatibility
-            rate(thanos_cache_memcached_requests_total{
-              %(jobMatcher)s,
-              component="%(component)s",
-              name="%(cacheName)s"
-            }[$__rate_interval])
-            or
             rate(thanos_cache_requests_total{
               %(jobMatcher)s,
               component="%(component)s",
@@ -1318,76 +1296,6 @@ local utils = import 'mixin-utils/utils.libsonnet';
       },
     },
 
-  // Copy/paste of latencyPanel from grafana-builder so that we can migrate between two different
-  // names for the same metric. When enough time has passed and we no longer care about the old
-  // metric name, this method can be removed and replaced with $.latencyPanel
-  backwardsCompatibleLatencyPanel(oldMetricName, newMetricName, selector, multiplier='1e3'):: {
-    targets: [
-      {
-        expr: |||
-          histogram_quantile(0.99, sum(
-            # Backwards compatibility
-            rate(%s_bucket%s[$__rate_interval])
-            or
-            rate(%s_bucket%s[$__rate_interval])
-          ) by (le)) * %s
-        ||| % [oldMetricName, selector, newMetricName, selector, multiplier],
-        format: 'time_series',
-        intervalFactor: 2,
-        legendFormat: '99th Percentile',
-        refId: 'A',
-      },
-      {
-        expr: |||
-          histogram_quantile(0.50, sum(
-            # Backwards compatibility
-            rate(%s_bucket%s[$__rate_interval])
-            or
-            rate(%s_bucket%s[$__rate_interval])
-          ) by (le)) * %s
-        ||| % [oldMetricName, selector, newMetricName, selector, multiplier],
-        format: 'time_series',
-        intervalFactor: 2,
-        legendFormat: '50th Percentile',
-        refId: 'B',
-      },
-      {
-        expr: |||
-          sum(
-            # Backwards compatibility
-            rate(%s_sum%s[$__rate_interval])
-            or
-            rate(%s_sum%s[$__rate_interval])
-          ) * %s
-          /
-          sum(
-            # Backwards compatibility
-            rate(%s_count%s[$__rate_interval])
-            or
-            rate(%s_count%s[$__rate_interval])
-          )
-        ||| % [
-          oldMetricName,
-          selector,
-          newMetricName,
-          selector,
-          multiplier,
-          oldMetricName,
-          selector,
-          newMetricName,
-          selector,
-        ],
-        format: 'time_series',
-        intervalFactor: 2,
-        legendFormat: 'Average',
-        refId: 'C',
-      },
-    ],
-    fieldConfig+: {
-      defaults+: { unit: 'ms', noValue: 0 },
-    },
-  },
-
   latencyRecordingRulePanel(metric, selectors, extra_selectors=[], multiplier='1e3', sum_by=[])::
     utils.latencyRecordingRulePanel(metric, selectors, extra_selectors, multiplier, sum_by) + {
       // Hide yaxes from JSON Model; it's not supported by timeseriesPanel.
@@ -1735,8 +1643,6 @@ local utils = import 'mixin-utils/utils.libsonnet';
           $.queryPanel(
             |||
               sum (
-                rate(thanos_memcached_operations_total{name="frontend-cache", %(frontend)s}[$__rate_interval])
-                or ignoring(backend)
                 rate(thanos_cache_operations_total{name="frontend-cache", %(frontend)s}[$__rate_interval])
               )
             ||| % {
@@ -1748,8 +1654,7 @@ local utils = import 'mixin-utils/utils.libsonnet';
         )
         .addPanel(
           $.timeseriesPanel('Latency') +
-          $.backwardsCompatibleLatencyPanel(
-            'thanos_memcached_operation_duration_seconds',
+          $.latencyPanel(
             'thanos_cache_operation_duration_seconds',
             '{%s, name="frontend-cache"}' % $.jobMatcher(queryFrontendJobName)
           )

diff --git a/operations/mimir-mixin/dashboards/reads.libsonnet b/operations/mimir-mixin/dashboards/reads.libsonnet
@@ -345,15 +345,6 @@ local filename = 'mimir-reads.json';
         $.queryPanel(
           |||
             sum by(operation) (
-              # Backwards compatibility
-              rate(
-                thanos_memcached_operations_total{
-                  component="store-gateway",
-                  name="index-cache",
-                  %s
-                }[$__rate_interval]
-              )
-              or ignoring(backend)
               rate(
                 thanos_cache_operations_total{
                   component="store-gateway",
@@ -362,19 +353,15 @@ local filename = 'mimir-reads.json';
                 }[$__rate_interval]
               )
             )
-          ||| % [
-            $.jobMatcher($._config.job_names.store_gateway),
-            $.jobMatcher($._config.job_names.store_gateway),
-          ],
+          ||| % $.jobMatcher($._config.job_names.store_gateway),
           '{{operation}}'
         ) +
         $.stack +
         { fieldConfig+: { defaults+: { unit: 'ops' } } }
       )
       .addPanel(
         $.timeseriesPanel('Latency (getmulti)') +
-        $.backwardsCompatibleLatencyPanel(
-          'thanos_memcached_operation_duration_seconds',
+        $.latencyPanel(
           'thanos_cache_operation_duration_seconds',
           |||
             {