-
Notifications
You must be signed in to change notification settings - Fork 876
Documentation: Capacity Planning - states nonexistant metric 'prometheus_tsdb_head_series' #7334
Describe the bug
The Documentation under https://cortexmetrics.io/docs/guides/capacity-planning/ states a metric 'prometheus_tsdb_head_series' that does not exist at all!
To Reproduce
Try to find prometheus_tsdb_head_series in /metrics on your cortex.
Expected behavior
It exists on /metrics
Environment:
Docker in single-binary mode
Additional Context
There is a closed PR that tried to add this metric to the system, but the PR was never merged: #4308
There is a comment on this PR, that states, an alternative to the given metric: #4308 (comment)
So maybe the Documentation can get fixed to that?
Instead of mentioning 'prometheus_tsdb_head_series' it could say 'cortex_ingester_memory_series_created_total - cortex_ingester_memory_series_removed_total' - if this query is still correct?
All reactions
Replies: 1 comment
the key point in the documentation: If you have Prometheus already
This is about migrating current workloads running in Prometheus to cortex.
if you need a head series equivalent the correct value is
sum(
cortex_ingester_memory_series_created_total
-
cortex_ingester_memory_series_removed_total
) by (user) / 3
Remember the replica. Yes I can try fixing the docs to also include this metric
All reactions
-
❤️ 1