Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Vtgate metrics: Added plan and query type metrics #17727

Open
wants to merge 16 commits into
base: main
Choose a base branch
from

Conversation

harshit-gangal
Copy link
Member

@harshit-gangal harshit-gangal commented Feb 10, 2025

Description

This PR adds classification of query plans based on their execution strategy.
Additionally, this PR introduces a new metrics counter, QueryProcessed, which tracks queries processed at VTGate with labels for Query, Plan, and Tablet type. This metric provides better observability into query distribution and execution patterns

Plan Classification

Query plans are categorized into different types based on their execution behavior:

  • Local – Queries executed locally on VTGate without involving any shard.
  • Passthrough – Queries forwarded to single shard without having any additional processing at VTGate.
  • MultiShard – Queries executed across multiple shards with controlled routing.
  • Lookup – Queries using lookup vindexes to resolve keyspace IDs efficiently and route to specific shards.
  • Scatter – Queries broadcast to all shards.
  • JoinOp – Queries involving join operations across multiple shards, with Join on VTGate.
  • ForeignKey – Queries handling foreign key constraints, such as cascades and validations.
  • Complex – Queries with intricate execution logic requiring VTGate to process results.
  • OnlineDDL – DDLs executed though Online Schema change workflow.
  • DirectDDL – DDLs directly executed on the shards.
  • Transaction – Queries managing transactions, including BEGIN, COMMIT, ROLLBACK, and SAVEPOINT.
  • Topology - Queries that involves accessing Topology Server for Get/Put.

Published Metrics

Sample metrics from VTGate's /metrics path:

# HELP vtgate_query_processed Query processed at vtgate by query, plan, and tablet type
# TYPE vtgate_query_processed counter
vtgate_query_processed{plan="Complex",query="SET",tablet="PRIMARY"} 1
vtgate_query_processed{plan="Local",query="USE",tablet="PRIMARY"} 1
vtgate_query_processed{plan="MultiShard",query="DELETE",tablet="PRIMARY"} 1
vtgate_query_processed{plan="MultiShard",query="UPDATE",tablet="PRIMARY"} 1
vtgate_query_processed{plan="Passthrough",query="SHOW",tablet="PRIMARY"} 1
vtgate_query_processed{plan="Scatter",query="DELETE",tablet="PRIMARY"} 20
vtgate_query_processed{plan="Scatter",query="SELECT",tablet="PRIMARY"} 1
vtgate_query_processed{plan="Transaction",query="ROLLBACK",tablet="PRIMARY"} 1
vtgate_query_processed{plan="Transaction",query="SAVEPOINT",tablet="PRIMARY"} 1

# HELP vtgate_query_routed Query routed from vtgate to vttablet by query, plan, and tablet type
# TYPE vtgate_query_routed counter
vtgate_query_routed{plan="Complex",query="SET",tablet="PRIMARY"} 0
vtgate_query_routed{plan="Local",query="USE",tablet="PRIMARY"} 0
vtgate_query_routed{plan="MultiShard",query="DELETE",tablet="PRIMARY"} 2
vtgate_query_routed{plan="MultiShard",query="UPDATE",tablet="PRIMARY"} 2
vtgate_query_routed{plan="Passthrough",query="SHOW",tablet="PRIMARY"} 1
vtgate_query_routed{plan="Scatter",query="DELETE",tablet="PRIMARY"} 56
vtgate_query_routed{plan="Scatter",query="SELECT",tablet="PRIMARY"} 2
vtgate_query_routed{plan="Transaction",query="ROLLBACK",tablet="PRIMARY"} 0
vtgate_query_routed{plan="Transaction",query="SAVEPOINT",tablet="PRIMARY"} 0

Related Issue(s)

Checklist

Signed-off-by: Harshit Gangal <harshit@planetscale.com>
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
Copy link
Contributor

vitess-bot bot commented Feb 10, 2025

Review Checklist

Hello reviewers! 👋 Please follow this checklist when reviewing this Pull Request.

General

  • Ensure that the Pull Request has a descriptive title.
  • Ensure there is a link to an issue (except for internal cleanup and flaky test fixes), new features should have an RFC that documents use cases and test cases.

Tests

  • Bug fixes should have at least one unit or end-to-end test, enhancement and new features should have a sufficient number of tests.

Documentation

  • Apply the release notes (needs details) label if users need to know about this change.
  • New features should be documented.
  • There should be some code comments as to why things are implemented the way they are.
  • There should be a comment at the top of each new or modified test to explain what the test does.

New flags

  • Is this flag really necessary?
  • Flag names must be clear and intuitive, use dashes (-), and have a clear help text.

If a workflow is added or modified:

  • Each item in Jobs should be named in order to mark it as required.
  • If the workflow needs to be marked as required, the maintainer team must be notified.

Backward compatibility

  • Protobuf changes should be wire-compatible.
  • Changes to _vt tables and RPCs need to be backward compatible.
  • RPC changes should be compatible with vitess-operator
  • If a flag is removed, then it should also be removed from vitess-operator and arewefastyet, if used there.
  • vtctl command output order should be stable and awk-able.

@vitess-bot vitess-bot bot added NeedsBackportReason If backport labels have been applied to a PR, a justification is required NeedsDescriptionUpdate The description is not clear or comprehensive enough, and needs work NeedsIssue A linked issue is missing for this Pull Request NeedsWebsiteDocsUpdate What it says labels Feb 10, 2025
@github-actions github-actions bot added this to the v22.0.0 milestone Feb 10, 2025
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
Copy link

codecov bot commented Feb 10, 2025

Codecov Report

Attention: Patch coverage is 94.75983% with 12 lines in your changes missing coverage. Please review.

Project coverage is 67.45%. Comparing base (000fbbe) to head (6c55d04).
Report is 8 commits behind head on main.

Files with missing lines Patch % Lines
go/vt/vtgate/engine/plan.go 93.02% 12 Missing ⚠️
Additional details and impacted files
@@            Coverage Diff             @@
##             main   #17727      +/-   ##
==========================================
- Coverage   67.95%   67.45%   -0.50%     
==========================================
  Files        1586     1592       +6     
  Lines      255208   258189    +2981     
==========================================
+ Hits       173423   174165     +742     
- Misses      81785    84024    +2239     

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

Signed-off-by: Harshit Gangal <harshit@planetscale.com>
@harshit-gangal harshit-gangal added Type: Enhancement Logical improvement (somewhere between a bug and feature) Component: Query Serving and removed NeedsIssue A linked issue is missing for this Pull Request NeedsBackportReason If backport labels have been applied to a PR, a justification is required labels Feb 10, 2025
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
…st for metrics

Signed-off-by: Harshit Gangal <harshit@planetscale.com>
@harshit-gangal harshit-gangal removed the NeedsDescriptionUpdate The description is not clear or comprehensive enough, and needs work label Feb 11, 2025
@harshit-gangal harshit-gangal marked this pull request as ready for review February 11, 2025 08:20
…nd using lookup vindex as lookup plan type

Signed-off-by: Harshit Gangal <harshit@planetscale.com>
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
@harshit-gangal
Copy link
Member Author

Website Doc PR: vitessio/website#1935

@deepthi
Copy link
Member

deepthi commented Feb 12, 2025

I understand that we can't re-use QueriesProcessed and QueriesRouted, because that might break people's dashboards, but we need to come up with some other names for QueryProcessed and QueryRouted which are grammatical. Leaving this comment as a place-holder while I try to think of alternatives.

harshit-gangal and others added 7 commits February 13, 2025 20:25
… for Set and Show Execution commands

Signed-off-by: Harshit Gangal <harshit@planetscale.com>
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
Signed-off-by: Harshit Gangal <harshit@planetscale.com>
Signed-off-by: Andres Taylor <andres@planetscale.com>
Signed-off-by: Andres Taylor <andres@planetscale.com>
@deepthi
Copy link
Member

deepthi commented Feb 18, 2025

I understand that we can't re-use QueriesProcessed and QueriesRouted, because that might break people's dashboards, but we need to come up with some other names for QueryProcessed and QueryRouted which are grammatical. Leaving this comment as a place-holder while I try to think of alternatives.

Here are my suggestions

  • QueryProcessed -> QueryCounts
  • QueryRouted -> QueryRouteCounts

@harshit-gangal
Copy link
Member Author

  • QueryProcessed -> QueryCounts
  • QueryRouted -> QueryRouteCounts

Count is self-explanatory, as it already implies a counter. Using Counts looks redundant in this context.

Comment on lines +57 to +71
const (
PlanUnknown PlanType = iota
PlanLocal
PlanPassthrough
PlanMultiShard
PlanLookup
PlanScatter
PlanJoinOp
PlanForeignKey
PlanComplex
PlanOnlineDDL
PlanDirectDDL
PlanTransaction
PlanTopoOp
)
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It could be good to have a link to the docs here as a comment (https://github.com/vitessio/website/pull/1935/files#diff-4badcfb66b7411dea206a299af9b9f8e3d349d817ad332798f747b0b9a40ad9f), both to help us know what each value mean and also to remind us that modifying this list implies that we need to change the docs.

Copy link
Member

@frouioui frouioui left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks good to me, just QueryRouted -> "query" in singular feels a bit odd. But I don't have better suggestion for the name. Not approving as deepthi has an ongoing conversation.

Copy link
Member

@GuptaManan100 GuptaManan100 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM to me too

@deepthi
Copy link
Member

deepthi commented Feb 21, 2025

  • QueryProcessed -> QueryCounts
  • QueryRouted -> QueryRouteCounts

Count is self-explanatory, as it already implies a counter. Using Counts looks redundant in this context.

The problem is that we don't have any good options because QueriesProcessed is taken. I proposed Counts because we already have that on vttablet, so there is precedence for it.
If you don't like that, then the next proposal is simply "Queries" and "QueryRoutes" :)

// TODO: @harshit/@systay - Remove these deprecated stats once we have a replacement in a released version.
queriesProcessed = stats.NewCountersWithSingleLabel("QueriesProcessed", "Deprecated: Queries processed at vtgate by plan type", "Plan")
queriesRouted = stats.NewCountersWithSingleLabel("QueriesRouted", "Deprecated: Queries routed from vtgate to vttablet by plan type", "Plan")
queriesProcessedByTable = stats.NewCountersWithMultiLabels("QueriesProcessedByTable", "Deprecated: Queries processed at vtgate by plan type, keyspace and table", []string{"Plan", "Keyspace", "Table"})
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We seem to be dropping the table metrics. Can we add that as a dimension to the new metrics? They have been useful in the past in debugging production issues.

2. QueryRouted - This counts the number of shards the query was executed on.

Deprecated:
1. QueriesProcessed
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

In code you are deprecating the table metrics, but that is missing from here.

### <a id="vtgate-metrics"/>VTGate Metrics

Added two new metrics for queries with query type, plan type and tablet type as dimension.
1. QueryProcessed - This counts the number of query executed.
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
1. QueryProcessed - This counts the number of query executed.
1. QueryProcessed - This counts the number of queries processed by this vtgate.


Added two new metrics for queries with query type, plan type and tablet type as dimension.
1. QueryProcessed - This counts the number of query executed.
2. QueryRouted - This counts the number of shards the query was executed on.
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This says shards but the actual description of the metric says "routed to vttablets". Which one is it?


queriesProcessedByTable = stats.NewCountersWithMultiLabels("QueriesProcessedByTable", "Queries processed at vtgate by plan type, keyspace and table", []string{"Plan", "Keyspace", "Table"})
queriesRoutedByTable = stats.NewCountersWithMultiLabels("QueriesRoutedByTable", "Queries routed from vtgate to vttablet by plan type, keyspace and table", []string{"Plan", "Keyspace", "Table"})
queryProcessed = stats.NewCountersWithMultiLabels("QueryProcessed", "Query processed at vtgate by query, plan, and tablet type", []string{"Query", "Plan", "Tablet"})
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
queryProcessed = stats.NewCountersWithMultiLabels("QueryProcessed", "Query processed at vtgate by query, plan, and tablet type", []string{"Query", "Plan", "Tablet"})
queryProcessed = stats.NewCountersWithMultiLabels("QueryProcessed", "Count of queries processed by query, plan and tablet type", []string{"Query", "Plan", "Tablet"})

queriesProcessedByTable = stats.NewCountersWithMultiLabels("QueriesProcessedByTable", "Queries processed at vtgate by plan type, keyspace and table", []string{"Plan", "Keyspace", "Table"})
queriesRoutedByTable = stats.NewCountersWithMultiLabels("QueriesRoutedByTable", "Queries routed from vtgate to vttablet by plan type, keyspace and table", []string{"Plan", "Keyspace", "Table"})
queryProcessed = stats.NewCountersWithMultiLabels("QueryProcessed", "Query processed at vtgate by query, plan, and tablet type", []string{"Query", "Plan", "Tablet"})
queryRouted = stats.NewCountersWithMultiLabels("QueryRouted", "Query routed from vtgate to vttablet by query, plan, and tablet type", []string{"Query", "Plan", "Tablet"})
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change
queryRouted = stats.NewCountersWithMultiLabels("QueryRouted", "Query routed from vtgate to vttablet by query, plan, and tablet type", []string{"Query", "Plan", "Tablet"})
queryRouted = stats.NewCountersWithMultiLabels("QueryRouted", "Count of queries routed to vttablets by query, plan and tablet type", []string{"Query", "Plan", "Tablet"})

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Component: Query Serving Type: Enhancement Logical improvement (somewhere between a bug and feature)
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants