diff --git a/modules/kubernetes/metrics/jobs/README.md b/modules/kubernetes/metrics/jobs/README.md
index 09f6547..73180c6 100644
--- a/modules/kubernetes/metrics/jobs/README.md
+++ b/modules/kubernetes/metrics/jobs/README.md
@@ -36,7 +36,7 @@ The following jobs are completely isolated and have no dependencies on other mod
 | `keep_metrics` | `false` | [see module](./agent.river#L170) | A regex of metrics to keep |
 | `scrape_interval` | `false` | `60s` | How often to scrape metrics from the targets |
 | `scrape_timeout` | `false` | `10s` | How long before a scrape times out |
-| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient |
+| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache.  This should be at least 2x-5x your largest scrape target or samples appended rate. |
 | `clustering` | `false` | `false` | Whether or not clustering should be enabled |
 
 ---
@@ -56,7 +56,7 @@ The following jobs are completely isolated and have no dependencies on other mod
 | `drop_metrics` | `false` | `""` | A regex of metrics to drop |
 | `scrape_interval` | `false` | `60s` | How often to scrape metrics from the targets |
 | `scrape_timeout` | `false` | `10s` | How long before a scrape times out |
-| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient |
+| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache.  This should be at least 2x-5x your largest scrape target or samples appended rate. |
 | `clustering` | `false` | `false` | Whether or not clustering should be enabled |
 
 ---
@@ -76,7 +76,7 @@ The following jobs are completely isolated and have no dependencies on other mod
 | `scrape_port_named_metrics` | `false` | Whether or not to automatically scrape endpoints that have a port with 'metrics' in the name |
 | `scrape_interval` | `false` | `60s` | How often to scrape metrics from the targets |
 | `scrape_timeout` | `false` | `10s` | How long before a scrape times out |
-| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient |
+| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache.  This should be at least 2x-5x your largest scrape target or samples appended rate. |
 | `clustering` | `false` | `false` | Whether or not clustering should be enabled |
 
 ---
@@ -91,7 +91,7 @@ The following jobs are completely isolated and have no dependencies on other mod
 | `keep_metrics` | `false` | [see module](./agent.river#L115) | A regex of metrics to keep |
 | `scrape_interval` | `false` | `60s` | How often to scrape metrics from the targets |
 | `scrape_timeout` | `false` | `10s` | How long before a scrape times out |
-| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient |
+| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache.  This should be at least 2x-5x your largest scrape target or samples appended rate. |
 | `clustering` | `false` | `false` | Whether or not clustering should be enabled |
 
 ---
@@ -109,7 +109,7 @@ The following jobs are completely isolated and have no dependencies on other mod
 | `keep_metrics` | `false` | [see module](./agent.river#L128) | A regex of metrics to keep |
 | `scrape_interval` | `false` | `60s` | How often to scrape metrics from the targets |
 | `scrape_timeout` | `false` | `10s` | How long before a scrape times out |
-| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient |
+| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache.  This should be at least 2x-5x your largest scrape target or samples appended rate. |
 | `clustering` | `false` | `false` | Whether or not clustering should be enabled |
 
 ---
@@ -127,7 +127,7 @@ The following jobs are completely isolated and have no dependencies on other mod
 | `keep_metrics` | `false` | [see module](./agent.river#L128) | A regex of metrics to keep |
 | `scrape_interval` | `false` | `60s` | How often to scrape metrics from the targets |
 | `scrape_timeout` | `false` | `10s` | How long before a scrape times out |
-| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient |
+| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache.  This should be at least 2x-5x your largest scrape target or samples appended rate. |
 | `clustering` | `false` | `false` | Whether or not clustering should be enabled |
 
 ---
@@ -146,7 +146,7 @@ The following jobs are completely isolated and have no dependencies on other mod
 | `drop_les` | `false` | [see module](./kube-apiserver.river#163) | Regex of metric les label values to drop |
 | `scrape_interval` | `false` | `60s` | How often to scrape metrics from the targets |
 | `scrape_timeout` | `false` | `10s` | How long before a scrape times out |
-| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient |
+| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache.  This should be at least 2x-5x your largest scrape target or samples appended rate. |
 | `clustering` | `false` | `false` | Whether or not clustering should be enabled |
 
 ---
@@ -161,7 +161,7 @@ The following jobs are completely isolated and have no dependencies on other mod
 | `keep_metrics` | `false` | [see module](./kube-probes.river#L117) | A regex of metrics to keep |
 | `scrape_interval` | `false` | `60s` | How often to scrape metrics from the targets |
 | `scrape_timeout` | `false` | `10s` | How long before a scrape times out |
-| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient |
+| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache.  This should be at least 2x-5x your largest scrape target or samples appended rate. |
 | `clustering` | `false` | `false` | Whether or not clustering should be enabled |
 
 ---
@@ -179,7 +179,7 @@ The following jobs are completely isolated and have no dependencies on other mod
 | `keep_metrics` | `false` | [see module](./kube-proxy.river#L124) | A regex of metrics to drop |
 | `scrape_interval` | `false` | `60s` | How often to scrape metrics from the targets |
 | `scrape_timeout` | `false` | `10s` | How long before a scrape times out |
-| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient |
+| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache.  This should be at least 2x-5x your largest scrape target or samples appended rate. |
 | `clustering` | `false` | `false` | Whether or not clustering should be enabled |
 
 ---
@@ -194,7 +194,7 @@ The following jobs are completely isolated and have no dependencies on other mod
 | `keep_metrics` | `false` | [see module](./kube-resource.river#L117) | A regex of metrics to keep |
 | `scrape_interval` | `false` | `60s` | How often to scrape metrics from the targets |
 | `scrape_timeout` | `false` | `10s` | How long before a scrape times out |
-| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient |
+| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache.  This should be at least 2x-5x your largest scrape target or samples appended rate. |
 | `clustering` | `false` | `false` | Whether or not clustering should be enabled |
 
 ---
@@ -212,7 +212,7 @@ The following jobs are completely isolated and have no dependencies on other mod
 | `keep_metrics` | `false` | [see module](./kube-state-metrics.river#L127) | A regex of metrics to keep |
 | `scrape_interval` | `false` | `60s` | How often to scrape metrics from the targets |
 | `scrape_timeout` | `false` | `10s` | How long before a scrape times out |
-| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient |
+| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache.  This should be at least 2x-5x your largest scrape target or samples appended rate. |
 | `clustering` | `false` | `false` | Whether or not clustering should be enabled |
 
 ---
@@ -227,7 +227,7 @@ The following jobs are completely isolated and have no dependencies on other mod
 | `keep_metrics` | `false` | [see module](./kubelet.river#L116) | A regex of metrics to keep |
 | `scrape_interval` | `false` | `60s` | How often to scrape metrics from the targets |
 | `scrape_timeout` | `false` | `10s` | How long before a scrape times out |
-| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient |
+| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache.  This should be at least 2x-5x your largest scrape target or samples appended rate. |
 | `clustering` | `false` | `false` | Whether or not clustering should be enabled |
 
 ---
@@ -245,7 +245,7 @@ The following jobs are completely isolated and have no dependencies on other mod
 | `keep_metrics` | `false` | [see module](./loki.river#L186) | A regex of metrics to keep |
 | `scrape_interval` | `false` | `60s` | How often to scrape metrics from the targets |
 | `scrape_timeout` | `false` | `10s` | How long before a scrape times out |
-| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient |
+| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache.  This should be at least 2x-5x your largest scrape target or samples appended rate. |
 | `clustering` | `false` | `false` | Whether or not clustering should be enabled |
 
 ---
@@ -263,7 +263,7 @@ The following jobs are completely isolated and have no dependencies on other mod
 | `keep_metrics` | `false` | [see module](./memcached.river#L170) | A regex of metrics to keep |
 | `scrape_interval` | `false` | `60s` | How often to scrape metrics from the targets |
 | `scrape_timeout` | `false` | `10s` | How long before a scrape times out |
-| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient |
+| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache.  This should be at least 2x-5x your largest scrape target or samples appended rate. |
 | `clustering` | `false` | `false` | Whether or not clustering should be enabled |
 
 ___
@@ -281,7 +281,7 @@ ___
 | `keep_metrics` | `false` | [see module](./mimir.river#L186) | A regex of metrics to keep |
 | `scrape_interval` | `false` | `60s` | How often to scrape metrics from the targets |
 | `scrape_timeout` | `false` | `10s` | How long before a scrape times out |
-| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient |
+| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache.  This should be at least 2x-5x your largest scrape target or samples appended rate. |
 | `clustering` | `false` | `false` | Whether or not clustering should be enabled |
 
 ---
@@ -299,7 +299,7 @@ ___
 | `keep_metrics` | `false` | [see module](./node-exporter.river#L141) | A regex of metrics to keep |
 | `scrape_interval` | `false` | `60s` | How often to scrape metrics from the targets |
 | `scrape_timeout` | `false` | `10s` | How long before a scrape times out |
-| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient |
+| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache.  This should be at least 2x-5x your largest scrape target or samples appended rate. |
 | `clustering` | `false` | `false` | Whether or not clustering should be enabled |
 
 ---
@@ -317,7 +317,7 @@ ___
 | `keep_metrics` | `false` | [see module](./opencost.river#L123) | A regex of metrics to keep |
 | `scrape_interval` | `false` | `60s` | How often to scrape metrics from the targets |
 | `scrape_timeout` | `false` | `10s` | How long before a scrape times out |
-| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient |
+| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache.  This should be at least 2x-5x your largest scrape target or samples appended rate. |
 | `clustering` | `false` | `false` | Whether or not clustering should be enabled |
 
 ---
@@ -349,5 +349,5 @@ ___
 | `keep_metrics` | `false` | [see module](./tempo.river#L183) | A regex of metrics to keep |
 | `scrape_interval` | `false` | `60s` | How often to scrape metrics from the targets |
 | `scrape_timeout` | `false` | `10s` | How long before a scrape times out |
-| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient |
+| `max_cache_size` | `false` | `100000` | The maximum number of elements to hold in the relabeling cache.  This should be at least 2x-5x your largest scrape target or samples appended rate. |
 | `clustering` | `false` | `false` | Whether or not clustering should be enabled |
diff --git a/modules/kubernetes/metrics/jobs/agent.river b/modules/kubernetes/metrics/jobs/agent.river
index 684fb0e..6638fb0 100644
--- a/modules/kubernetes/metrics/jobs/agent.river
+++ b/modules/kubernetes/metrics/jobs/agent.river
@@ -53,7 +53,7 @@ argument "scrape_timeout" {
 }
 
 argument "max_cache_size" {
-  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient"
+  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  This should be at least 2x-5x your largest scrape target or samples appended rate."
   optional = true
 }
 
diff --git a/modules/kubernetes/metrics/jobs/annotations-probe.river b/modules/kubernetes/metrics/jobs/annotations-probe.river
index 792ed76..1f744a2 100644
--- a/modules/kubernetes/metrics/jobs/annotations-probe.river
+++ b/modules/kubernetes/metrics/jobs/annotations-probe.river
@@ -119,7 +119,7 @@ argument "scrape_timeout" {
 }
 
 argument "max_cache_size" {
-  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient"
+  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  This should be at least 2x-5x your largest scrape target or samples appended rate."
   optional = true
 }
 
diff --git a/modules/kubernetes/metrics/jobs/annotations-scrape.river b/modules/kubernetes/metrics/jobs/annotations-scrape.river
index e94d10e..ceba188 100644
--- a/modules/kubernetes/metrics/jobs/annotations-scrape.river
+++ b/modules/kubernetes/metrics/jobs/annotations-scrape.river
@@ -146,7 +146,7 @@ argument "scrape_timeout" {
 }
 
 argument "max_cache_size" {
-  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient"
+  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  This should be at least 2x-5x your largest scrape target or samples appended rate."
   optional = true
 }
 
diff --git a/modules/kubernetes/metrics/jobs/cadvisor.river b/modules/kubernetes/metrics/jobs/cadvisor.river
index c08c2b5..443e99c 100644
--- a/modules/kubernetes/metrics/jobs/cadvisor.river
+++ b/modules/kubernetes/metrics/jobs/cadvisor.river
@@ -37,7 +37,7 @@ argument "scrape_timeout" {
 }
 
 argument "max_cache_size" {
-  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient"
+  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  This should be at least 2x-5x your largest scrape target or samples appended rate."
   optional = true
 }
 
diff --git a/modules/kubernetes/metrics/jobs/gitlab-exporter.river b/modules/kubernetes/metrics/jobs/gitlab-exporter.river
index dde929f..e7f0f87 100644
--- a/modules/kubernetes/metrics/jobs/gitlab-exporter.river
+++ b/modules/kubernetes/metrics/jobs/gitlab-exporter.river
@@ -57,7 +57,7 @@ argument "scrape_timeout" {
 }
 
 argument "max_cache_size" {
-  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient"
+  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  This should be at least 2x-5x your largest scrape target or samples appended rate."
   optional = true
 }
 
diff --git a/modules/kubernetes/metrics/jobs/grafana.river b/modules/kubernetes/metrics/jobs/grafana.river
index 387c151..c421a13 100644
--- a/modules/kubernetes/metrics/jobs/grafana.river
+++ b/modules/kubernetes/metrics/jobs/grafana.river
@@ -53,7 +53,7 @@ argument "scrape_timeout" {
 }
 
 argument "max_cache_size" {
-  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient"
+  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  This should be at least 2x-5x your largest scrape target or samples appended rate."
   optional = true
 }
 
diff --git a/modules/kubernetes/metrics/jobs/kube-apiserver.river b/modules/kubernetes/metrics/jobs/kube-apiserver.river
index 31f3bcc..585b0fc 100644
--- a/modules/kubernetes/metrics/jobs/kube-apiserver.river
+++ b/modules/kubernetes/metrics/jobs/kube-apiserver.river
@@ -60,7 +60,7 @@ argument "scrape_timeout" {
 }
 
 argument "max_cache_size" {
-  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient"
+  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  This should be at least 2x-5x your largest scrape target or samples appended rate."
   optional = true
 }
 
diff --git a/modules/kubernetes/metrics/jobs/kube-probes.river b/modules/kubernetes/metrics/jobs/kube-probes.river
index c2e2ff6..a077b71 100644
--- a/modules/kubernetes/metrics/jobs/kube-probes.river
+++ b/modules/kubernetes/metrics/jobs/kube-probes.river
@@ -38,7 +38,7 @@ argument "scrape_timeout" {
 }
 
 argument "max_cache_size" {
-  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient"
+  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  This should be at least 2x-5x your largest scrape target or samples appended rate."
   optional = true
 }
 
diff --git a/modules/kubernetes/metrics/jobs/kube-proxy.river b/modules/kubernetes/metrics/jobs/kube-proxy.river
index f3216f7..0b6acf1 100644
--- a/modules/kubernetes/metrics/jobs/kube-proxy.river
+++ b/modules/kubernetes/metrics/jobs/kube-proxy.river
@@ -54,7 +54,7 @@ argument "scrape_timeout" {
 }
 
 argument "max_cache_size" {
-  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient"
+  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  This should be at least 2x-5x your largest scrape target or samples appended rate."
   optional = true
 }
 
diff --git a/modules/kubernetes/metrics/jobs/kube-resource.river b/modules/kubernetes/metrics/jobs/kube-resource.river
index d117d7c..71d850e 100644
--- a/modules/kubernetes/metrics/jobs/kube-resource.river
+++ b/modules/kubernetes/metrics/jobs/kube-resource.river
@@ -38,7 +38,7 @@ argument "scrape_timeout" {
 }
 
 argument "max_cache_size" {
-  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient"
+  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  This should be at least 2x-5x your largest scrape target or samples appended rate."
   optional = true
 }
 
diff --git a/modules/kubernetes/metrics/jobs/kube-state-metrics.river b/modules/kubernetes/metrics/jobs/kube-state-metrics.river
index 529d7c6..b32a4b3 100644
--- a/modules/kubernetes/metrics/jobs/kube-state-metrics.river
+++ b/modules/kubernetes/metrics/jobs/kube-state-metrics.river
@@ -57,7 +57,7 @@ argument "scrape_timeout" {
 }
 
 argument "max_cache_size" {
-  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient"
+  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  This should be at least 2x-5x your largest scrape target or samples appended rate."
   optional = true
 }
 
diff --git a/modules/kubernetes/metrics/jobs/kubelet.river b/modules/kubernetes/metrics/jobs/kubelet.river
index bb21df8..d4423b8 100644
--- a/modules/kubernetes/metrics/jobs/kubelet.river
+++ b/modules/kubernetes/metrics/jobs/kubelet.river
@@ -38,7 +38,7 @@ argument "scrape_timeout" {
 }
 
 argument "max_cache_size" {
-  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient"
+  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  This should be at least 2x-5x your largest scrape target or samples appended rate."
   optional = true
 }
 
diff --git a/modules/kubernetes/metrics/jobs/loki.river b/modules/kubernetes/metrics/jobs/loki.river
index d4feb20..6b2956f 100644
--- a/modules/kubernetes/metrics/jobs/loki.river
+++ b/modules/kubernetes/metrics/jobs/loki.river
@@ -53,7 +53,7 @@ argument "scrape_timeout" {
 }
 
 argument "max_cache_size" {
-  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient"
+  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  This should be at least 2x-5x your largest scrape target or samples appended rate."
   optional = true
 }
 
diff --git a/modules/kubernetes/metrics/jobs/memcached.river b/modules/kubernetes/metrics/jobs/memcached.river
index f8d5c8e..da58f74 100644
--- a/modules/kubernetes/metrics/jobs/memcached.river
+++ b/modules/kubernetes/metrics/jobs/memcached.river
@@ -53,7 +53,7 @@ argument "scrape_timeout" {
 }
 
 argument "max_cache_size" {
-  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient"
+  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  This should be at least 2x-5x your largest scrape target or samples appended rate."
   optional = true
 }
 
diff --git a/modules/kubernetes/metrics/jobs/mimir.river b/modules/kubernetes/metrics/jobs/mimir.river
index ff4b0cc..6adbde7 100644
--- a/modules/kubernetes/metrics/jobs/mimir.river
+++ b/modules/kubernetes/metrics/jobs/mimir.river
@@ -53,7 +53,7 @@ argument "scrape_timeout" {
 }
 
 argument "max_cache_size" {
-  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient"
+  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  This should be at least 2x-5x your largest scrape target or samples appended rate."
   optional = true
 }
 
diff --git a/modules/kubernetes/metrics/jobs/node-exporter.river b/modules/kubernetes/metrics/jobs/node-exporter.river
index ea11cc2..86e430a 100644
--- a/modules/kubernetes/metrics/jobs/node-exporter.river
+++ b/modules/kubernetes/metrics/jobs/node-exporter.river
@@ -58,7 +58,7 @@ argument "scrape_timeout" {
 }
 
 argument "max_cache_size" {
-  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient"
+  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  This should be at least 2x-5x your largest scrape target or samples appended rate."
   optional = true
 }
 
diff --git a/modules/kubernetes/metrics/jobs/opencost.river b/modules/kubernetes/metrics/jobs/opencost.river
index 651f811..47b0682 100644
--- a/modules/kubernetes/metrics/jobs/opencost.river
+++ b/modules/kubernetes/metrics/jobs/opencost.river
@@ -56,7 +56,7 @@ argument "scrape_timeout" {
 }
 
 argument "max_cache_size" {
-  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  Only increase if the amount of metrics returned is extremely large, the default will almost always be sufficient"
+  comment = "The maximum number of elements to hold in the relabeling cache (default: 100000).  This should be at least 2x-5x your largest scrape target or samples appended rate."
   optional = true
 }