Skip to content

Commit f56bbe3

Browse files
committed
fix formatting for the naming page
1 parent aa63992 commit f56bbe3

File tree

1 file changed

+6
-0
lines changed

1 file changed

+6
-0
lines changed

docs/concepts/naming.md

Lines changed: 6 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -32,6 +32,7 @@ or `jvm.gc.pause`. The prefix should generally have no more than 2 levels to kee
3232
This is not a package hierarchy like in Java - it's simply a way to group related metrics.
3333

3434
Examples of good prefixes:
35+
3536
* `ipc.*` for inter-process communication metrics
3637
* `jvm.*` for Java Virtual Machine metrics
3738
* `db.*` for database metrics
@@ -79,10 +80,12 @@ and each combination consumes storage and processing resources. Tag combinations
7980
over time to avoid constantly creating new time series.
8081

8182
Consider the cardinality impact:
83+
8284
* A metric with 3 tag keys, each with 10 possible values = 1,000 potential time series
8385
* A metric with 5 tag keys, each with 10 possible values = 100,000 potential time series
8486

8587
Guidelines for managing cardinality:
88+
8689
* **Limit high-cardinality dimensions.** Avoid tags with unbounded or very large value sets
8790
* **Use stable identifiers.** Tag values should remain consistent over time
8891

@@ -93,11 +96,13 @@ be queried simply and allow users to incrementally drill into the data. This imp
9396
performance and user experience.
9497

9598
Good query patterns:
99+
96100
* `name,threadpool.size,:eq` - exact match on name
97101
* `name,threadpool.size,:eq,id,server-requests,:eq,:and` - add exact tag filter
98102
* `name,threadpool.*,:re` - simple prefix pattern (use sparingly)
99103

100104
Avoid patterns that require expensive operations:
105+
101106
* Complex regex patterns that must scan many metric names
102107
* Queries that require examining all tag combinations to find matches
103108
* Dynamic name construction that makes direct queries impossible
@@ -142,6 +147,7 @@ see the total number of threads in all pools. You can then group by or select an
142147
filter the data to a subset in which you have an interest.
143148

144149
This approach also supports simple queries without regex patterns:
150+
145151
* `name,threadpool.size,:eq` gives you all thread pool sizes
146152
* `name,db.size,:eq` gives you all database sizes
147153
* `name,threadpool.size,:eq,id,server-requests,:eq,:and` drills down to a specific pool

0 commit comments

Comments
 (0)