You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Update the metric names to match Prometheus conventions
I found a few improvements to the metrics based on https://prometheus.io/docs/practices/naming/
These are good recommendations in general, and having standard units makes it easier
to use the metrics in queries.
For example, having everything in bytes, means you can just divide one metric by another
without conversion.
Having the timestamp in seconds allows easy use with the `time()` function.
Signed-off-by: Goutham <[email protected]>
Copy file name to clipboardExpand all lines: README.md
+4-4
Original file line number
Diff line number
Diff line change
@@ -46,15 +46,15 @@ It exposes the following metrics:
46
46
| Metric | Description |
47
47
|-|-|
48
48
|`unused_disks_count`| How many unused disks are in this provider |
49
-
|`unused_disks_size_gb`| Total size of unused disks in this provider in GB|
49
+
|`unused_disks_total_size_bytes`| Total size of unused disks in this provider in bytes|
50
50
|`unused_disk_size_bytes`| Size of each disk in bytes |
51
-
|`unused_disks_last_used_at`| Last timestamp (unix ms) when this disk was used. GCP only! |
52
-
|`unused_provider_duration_ms`| How long in milliseconds took to fetch this provider information |
51
+
|`unused_disks_last_used_timestamp_seconds`| Last timestamp (unix seconds) when this disk was used. GCP only! |
52
+
|`unused_provider_duration_seconds`| How long in milliseconds took to fetch this provider information |
53
53
|`unused_provider_info`| CSP information |
54
54
|`unused_provider_success`| Static metric indicating if collecting the metrics succeeded or not |
55
55
56
56
All metrics have the `provider` and `provider_id` labels to identify to which provider instance they belong.
57
-
The `unused_disks_count` and `unused_disks_size_gb` metrics have an additional `k8s_namespace` metric mapped to the `kubernetes.io/created-for/pvc/namespace` annotation assigned to persistent disks created by Kubernetes.
57
+
The `unused_disks_count` and `unused_disks_total_size_bytes` metrics have an additional `k8s_namespace` metric mapped to the `kubernetes.io/created-for/pvc/namespace` annotation assigned to persistent disks created by Kubernetes.
58
58
59
59
Information about each unused disk is currently logged to stdout given that it contains more changing information that could lead to cardinality explosion.
0 commit comments