Add index and table metrics to vttablet #17570

rafer · 2025-01-18T02:22:26Z

Description

This adds the following metrics to vttablet's schema engine.

TableRows (by table) - The estimated number of rows in the table
TableClusteredIndexSize (by table) - The size of the clustered index (i.e. the size of the "row data")
IndexCardinality (by index) - The estimated number of unique values in the index
IndexBytes (by index) - The size of the index

metrics sample

curl -sS http://localhost:15100/debug/vars | jq -S '{TableRows, TableClusteredIndexSize, IndexCardinality, IndexBytes}'
{
  "IndexBytes": {
    "corder.PRIMARY": 16384,
    "customer.PRIMARY": 16384,
    "product.PRIMARY": 16384
  },
  "IndexCardinality": {
    "corder.PRIMARY": 0,
    "customer.PRIMARY": 0,
    "product.PRIMARY": 0
  },
  "TableClusteredIndexSize": {
    "corder": 16384,
    "customer": 16384,
    "product": 16384
  },
  "TableRows": {
    "corder": 0,
    "customer": 0,
    "product": 0
  }
}

These values are read from mysql system tables when Engine#Reload is called, and made available via Env#Exporter

Partition names

Since mysql implements partitions internally as one table per partition, the "table name" referenced in the stats tables mostly look like TABLE_NAME#p#PARTITION_NAME. We want to report statistics per logical table (not per partition) so we need a way to extract the underlying table name. Unfortunately, it's not as simple as splitting the composite name by #p#, because it's possible to create a table or partition with #p# as a part of the name.

The approach I took here is to load the information_schema.partitions table, which includes separate columns for the table and partition name. Then, as we read through the stats tables that contain the composite name, if we encounter a #p#, we attempt to find a matching table/partition in the loaded list, and, if it is present, we can find the base table name.

An alternative method would be to make use of information_schema.innodb_tables.name column (like we do here), which provides the table name with special characters encoded. This makes it possible to confidently split on #p#, because the # characters would be encoded if they were part of the real table or partition name. I opted for the "loading all partitions" approach instead because the joins required to get an encoded table name made the queries too slow when there are many tables, indexes or partitions.

Performance

Since performance of the queries run on #Reload has been an issue in the past, I tested the queries added in this PR with the following test setup.

5,000 tables with 6 indexes each (5 secondary indexes)
5 tables with 5,000 partitions each, 4 indexes each (3 secondary indexes)

Results:

Entire updateTableIndexMetrics method: 716ms
- Partitions query: 37ms
- Table stats query (table row count and clustered index size): 15ms
- Index bytes query (byte size of each index): 205ms
- Cardinality query: 427ms

For comparison the InnoDBTableSizes query takes 2.0s for the same test setup.

Related Issue(s)

Fixes: Feature Request: Additional table and index metrics #17569

Checklist

"Backport to:" labels have been added if this change should be back-ported to release branches
If this change is to be back-ported to previous releases, a justification is included in the PR description
Tests were added or are not required
Did the new or modified tests pass consistently locally and on CI?
Documentation was added or is not required

Deployment Notes

None

* IndexBytes * IndexCardinality * TableClusteredIndexSize * TableRows Signed-off-by: Rafer Hazen <rafer@ralua.com>

vitess-bot · 2025-01-18T02:22:29Z

Signed-off-by: Rafer Hazen <rafer@ralua.com>

codecov · 2025-01-21T23:58:21Z

Codecov Report

Attention: Patch coverage is 84.03361% with 19 lines in your changes missing coverage. Please review.

Project coverage is 67.65%. Comparing base (71ccd6d) to head (8f83d68).
Report is 14 commits behind head on main.

Files with missing lines	Patch %	Lines
go/vt/vttablet/tabletserver/schema/engine.go	84.03%	19 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #17570      +/-   ##
==========================================
- Coverage   67.71%   67.65%   -0.06%     
==========================================
  Files        1584     1585       +1     
  Lines      254721   256592    +1871     
==========================================
+ Hits       172473   173608    +1135     
- Misses      82248    82984     +736

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

rafer · 2025-01-22T00:14:11Z

go/vt/vttablet/tabletserver/schema/engine.go

@@ -432,7 +444,7 @@ func (se *Engine) reload(ctx context.Context, includeStats bool) error {
 			// We therefore don't want to query for table sizes in getTableData()
 			includeStats = false

-			innodbResults, err := conn.Conn.Exec(ctx, innodbTableSizesQuery, maxTableCount, false)
+			innodbResults, err := conn.Conn.Exec(ctx, innodbTableSizesQuery, maxTableCount*maxPartitionsPerTable, false)


Increased this because even with our limit of 10k tables, it's possible that there are more partitions than that (and this query returns one row per partition).

rafer · 2025-01-22T00:17:00Z

go/vt/vttablet/tabletserver/health_streamer_test.go

@@ -404,6 +404,10 @@ func TestReloadView(t *testing.T) {
 	db.AddQuery("SELECT TABLE_NAME, CREATE_TIME FROM _vt.`tables`", &sqltypes.Result{})
 	// adding query pattern for udfs
 	db.AddQueryPattern("SELECT name.*", &sqltypes.Result{})
+	db.AddQuery("select table_name, partition_name from information_schema.partitions where table_schema = database() and partition_name is not null", &sqltypes.Result{})


I don't love that these queries have leaked out of the schema package, but I can't think of better way to make this test pass (unless we want to make the query constants public). Interested if this causes anyone concern.

rafer · 2025-01-22T00:18:30Z

go/vt/vttablet/tabletserver/schema/engine.go

@@ -466,8 +478,12 @@ func (se *Engine) reload(ctx context.Context, includeStats bool) error {
 					}
 				}
 			}
+			if err := se.updateTableIndexMetrics(ctx, conn.Conn); err != nil {
+				log.Errorf("Updating index/table statistics failed, error: %v", err)


This just logs instead of failing the method. I figure if updateTableIndexMetrics fails it's better to continue on since the only impact is missing metrics.

mncaudill · 2025-02-04T20:06:51Z

@deepthi We're hoping to keep moving forward on https://github.com/planetscale/surfaces/issues/2679 and this change is (unfortunately!) a blocker. Would you be able to help us get it into the queue for review? If it already is, ignore me!

harshit-gangal · 2025-02-17T10:52:00Z

go/vt/vttablet/tabletserver/schema/engine.go

+		if strings.Contains(tableName, "#p#") {
+			if partition, ok := partitions[tableName]; ok {
+				tableName = partition.table
+			}
+		}


can we not directly check in partitions map instead of doing a contains check?

harshit-gangal · 2025-02-17T10:53:01Z

go/vt/vttablet/tabletserver/schema/engine.go

+		if strings.Contains(tableName, "#p#") {
+			if partition, ok := partitions[tableName]; ok {
+				tableName = partition.table
+			}
+		}


same as above

harshit-gangal · 2025-02-17T10:55:51Z

go/vt/vttablet/tabletserver/schema/db.go

+
+	// fetch a list of all partitions
+	fetchPartitions = `select table_name, partition_name from information_schema.partitions where table_schema = database() and partition_name is not null`
+
+	// fetch the estimated number of rows and the clustered index byte size for all tables
+	fetchTableRowCountClusteredIndex = `select table_name, n_rows, clustered_index_size * @@innodb_page_size from mysql.innodb_table_stats where database_name = database()`
+
+	// fetch the byte size of all indexes
+	fetchIndexSizes = `select table_name, index_name, stat_value * @@innodb_page_size from mysql.innodb_index_stats where database_name = database() and stat_name = 'size'`
+
+	// fetch the cardinality of all indexes
+	fetchIndexCardinalities = `select table_name, index_name, max(cardinality) from information_schema.statistics s where table_schema = database() group by s.table_name, s.index_name`


this should be added to mysql flavor, otherwise import from mysql 5.7 will complain if the table or column does not exist

harshit-gangal · 2025-02-17T10:57:49Z

go/vt/vttablet/endtoend/misc_test.go

@@ -1090,6 +1090,30 @@ func TestEngineReload(t *testing.T) {
 	})
 }

+func TestUpdateTableIndexMetrics(t *testing.T) {


do we have tables with partitions in this e2e test.
Is it possible to validate the value than checking for nil?

GuptaManan100

LGTM

Add the following metrics to vttablet

64b606d

* IndexBytes * IndexCardinality * TableClusteredIndexSize * TableRows Signed-off-by: Rafer Hazen <rafer@ralua.com>

rafer requested review from harshit-gangal, systay, shlomi-noach, rohit-nayak-ps, timvaillancourt and mattlord as code owners January 18, 2025 02:22

github-actions bot added this to the v22.0.0 milestone Jan 18, 2025

rafer added 2 commits January 21, 2025 15:15

Extract table/index metrics queries, add them to Engine test mocks

bfb0eef

Signed-off-by: Rafer Hazen <rafer@ralua.com>

Ignore table/index metrics queries in health streamer tests

8f83d68

Signed-off-by: Rafer Hazen <rafer@ralua.com>

rafer commented Jan 22, 2025

View reviewed changes

deepthi removed the request for review from shlomi-noach January 22, 2025 01:27

harshit-gangal reviewed Feb 17, 2025

View reviewed changes

GuptaManan100 reviewed Feb 19, 2025

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add index and table metrics to vttablet #17570

Add index and table metrics to vttablet #17570

rafer commented Jan 18, 2025 •

edited by mattlord

Loading

vitess-bot bot commented Jan 18, 2025

codecov bot commented Jan 21, 2025

rafer Jan 22, 2025

rafer Jan 22, 2025 •

edited

Loading

rafer Jan 22, 2025

mncaudill commented Feb 4, 2025

harshit-gangal Feb 17, 2025

harshit-gangal Feb 17, 2025

harshit-gangal Feb 17, 2025 •

edited

Loading

harshit-gangal Feb 17, 2025

GuptaManan100 left a comment

Add index and table metrics to vttablet #17570

Are you sure you want to change the base?

Add index and table metrics to vttablet #17570

Conversation

rafer commented Jan 18, 2025 • edited by mattlord Loading

Description

Partition names

Performance

Related Issue(s)

Checklist

Deployment Notes

vitess-bot bot commented Jan 18, 2025

Review Checklist

General

Tests

Documentation

New flags

If a workflow is added or modified:

Backward compatibility

codecov bot commented Jan 21, 2025

Codecov Report

rafer Jan 22, 2025

Choose a reason for hiding this comment

rafer Jan 22, 2025 • edited Loading

Choose a reason for hiding this comment

rafer Jan 22, 2025

Choose a reason for hiding this comment

mncaudill commented Feb 4, 2025

harshit-gangal Feb 17, 2025

Choose a reason for hiding this comment

harshit-gangal Feb 17, 2025

Choose a reason for hiding this comment

harshit-gangal Feb 17, 2025 • edited Loading

Choose a reason for hiding this comment

harshit-gangal Feb 17, 2025

Choose a reason for hiding this comment

GuptaManan100 left a comment

Choose a reason for hiding this comment

rafer commented Jan 18, 2025 •

edited by mattlord

Loading

rafer Jan 22, 2025 •

edited

Loading

harshit-gangal Feb 17, 2025 •

edited

Loading