[QueryThrottler]Replace polling with topo server watch for config updates

[siddharth16396]

This PR is clone of #18905 which i messed up pretty badly while rebasing for DCO commits 😢

Description

This PR Replaces periodic config polling (60s intervals) with event-driven topology server watches for QueryThrottler configuration updates. Instead of continuously reading config files on a timer, the throttler now subscribes to SrvKeyspace changes and reacts immediately when configs are updated in the topo server.

Changes Made

Removed the polling model:

• Deleted ConfigLoader interface and FileBasedConfigLoader implementation.
• Removed 60-second refresh loop from QueryThrottler.
• Eliminated file-based config loading entirely.

Added event-driven topo watch:

• Introduced startSrvKeyspaceWatch() that uses srvtopo.WatchSrvKeyspace() for real-time updates
• Added HandleConfigUpdate() callback for processing config changes
• Implemented smart deduplication to avoid redundant strategy reloads
• Added resilient error handling (continues watching on transient errors, stops only on permanent ones like NoNode)

New proto definitions:

• Created querythrottler.proto with Config, ThrottlingStrategy, and nested rule structures
• Added incoming_query_throttler_config field to SrvKeyspace message in topodata.proto
• Implemented ConfigFromProto() conversion between protobuf and internal config representation

Architecture improvements:

• Config now lives in topology server (centralized, versioned, observable)
• Tablets get instant updates when operators change throttling rules
• Initial config loaded synchronously during InitDBConfig() to ensure correct state on startup/restart
• Watch continues indefinitely with auto-retry for transient failures

How it works

1. Startup: InitDBConfig() sets the keyspace and immediately fetches initial config via GetSrvKeyspace()
2. Watch: Background goroutine subscribes to WatchSrvKeyspace() for the cell+keyspace
3. Updates: When config changes in topo, HandleConfigUpdate() fires, deduplicates, and hot-swaps the strategy
4. Resilience: Network blips are retried automatically; only fatal errors (context canceled, keyspace deleted) stop the watch

Benefits achieved:

• Faster propagation: Config changes apply immediately instead of waiting up to 60s
• Less load: No more periodic file reads from every tablet
• Better ops: Centralized config in topo server (can use vtctldclient to update)
• Consistency: All tablets see changes at roughly the same time
• Debugging: Topo changes are auditable (vs. scattered config files)

Related Issue(s)

• Question: v23.0.0 Error loading config: open /config/throttler-config.json: no such file or directory #18869

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

[vitess-bot]

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.

[mattlord]

The description looks good to me! I didn't get to this before my holiday vacation, but I promise to do it the week that I'm back. Thank you for the contribution! I apologize for the delay.

[mattlord]

LGTM! Just a number of minor comments/questions/suggestions. Let me know what you think

[mattlord]

Why not use a sync.Once instance?

[siddharth16396]

sync.Once cannot be reset. Shutdown() resets watchStarted to false to allow potential restarts (useful for testing and defensive design).

Though we're certain QueryThrottler is never restarted after shutdown and maybe we could switch to sync.Once, but the current approach provides more flexibility for tests and future use cases and hence we used atomic.Bool

[mattlord]

Any reason not to use Equal instead? It's another reason to use the proto types directly as we can then use proto.Equal()

[siddharth16396]

The reason for not using "Equal" was:

The Config proto contains both:

1. QueryThrottler-level fields: enabled, strategy, dry_run
2. Strategy-specific configs: nested configs for TabletThrottler, custom strategies, etc.

isConfigUpdateRequired intentionally compares only the QueryThrottler-level fields because:

1. Separation of concerns: QueryThrottler treats strategy internals as opaque. It only cares whether throttling is enabled, which strategy type to use, and whether dry-run is on.
2. Strategy self-management: Each strategy instance watches and handles its own configuration updates independently. When TabletThrottler's internal thresholds change, it reloads them itself - QueryThrottler doesn't need to recreate the strategy.
3. Avoiding unnecessary churn: Using proto.Equal() would trigger a full strategy swap whenever any nested field changes, even though the existing strategy instance can handle those updates internally.

[codecov]

Codecov Report

❌ Patch coverage is 98.16514% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 69.89%. Comparing base (fa770f9) to head (942d7e8).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
...vttablet/tabletserver/querythrottler/test_utils.go	88.23%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #18979      +/-   ##
==========================================
+ Coverage   69.87%   69.89%   +0.01%     
==========================================
  Files        1613     1611       -2     
  Lines      216002   216058      +56     
==========================================
+ Hits       150939   151013      +74     
+ Misses      65063    65045      -18     

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.