fix(guard): replace internal caches with moka

[MasterPtato]

Changes

[MasterPtato]

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

• feat: db sh for workflows #2570
• fix(pegboard): fix enabling root users #2572
• fix(toolchain): fix compat with deploying using podman #2571
• feat: implement clickhouse-user-query #2554
• chore: add Linux ARM64 and Windows build targets #2551
• fix: uncomment function type filter in ACTORS_FILTER #2533
• chore(examples): multitenant deploys example #2527
• feat: add non-interactive mode and route management options to CLI #2532
• fix(pegboard): fix container runner orphaning #2575
• fix: dont include installing servers in service discovery #2569
• fix(pegboard): fix allocation metrics #2568
• fix: fix edge dc eligible calculation #2567
• fix(cluster): fix scaling logic and dedup metrics #2566
• feat(pegboard): add draining state to alloc metrics #2565
• fix(guard): dont route to installing api nodes #2564
• fix: improve error handling in status check #2563
• fix: add vector http config to edge configs #2553
• chore: fix dc wf #2552
• chore(vector): add vector pipeline to ship clickhouse events from the edge #2526
• fix(pegboard): increase actor log ttl to 14 days #2544
• feat(clusters): add margin per pool #2543
• Fix(guard): return error response on router error #2540
• fix(pegboard): wait for actor allocation not creation #2539
• fix(pegboard): increase signal timeout, add check for actor exit #2538
• fix(pegboard): revise actor rescheduling algorithm, add client metrics #2531
• fix: fix build cache key #2530
• feat: add retries to sql macros #2529
• fix: add create ts to server list #2523
• fix: fix pb metrics standalone #2518
• chore: disable OTEL by default in guard service #2522
• chore: increase file watch limit for rivet guard #2521
• fix(pegboard): continue exporting metrics even if fetching actor fails #2520
• fix(pegboard): fix collecting metrics for actors without workflows #2519
• feat(guard): add http ping endpoint to edge api hosts #2506
• chore: update workspace #2505
• chore: fix guard metrics #2504
• feat: add pb usage metrics, server state #2503
• fix(cache): add traces #2489
• fix(guard): add more tokio runtime metrics, remove labels from metrics #2485
• fix(guard): replace internal caches with moka #2481 👈 (View in Graphite)
• fix(guard): add metrics #2480
• chore: expose namespace in otel attributes #2501 : 1 other dependent PR (#2502 )
• chore: add standalone toolchain build scripts #2400 : 1 other dependent PR (#2500 )
• refactor: replace unsafe File::from_raw_fd with safe File::from #2479
• main

---

How to use the Graphite Merge Queue

Add the label merge-queue to this PR to add it to the merge queue.

You must have a Graphite account in order to use the merge queue. Sign up using this link.

An organization admin has enabled the Graphite Merge Queue in this repository.

Please do not merge from GitHub as this will restart CI on PRs being processed by the merge queue.

This stack of pull requests is managed by Graphite. Learn more about stacking.

[greptile-apps]

PR Summary

This PR replaces the internal caching mechanism in the guard service with the Moka caching library, providing better cache management and automatic TTL expiration.

• Added moka dependency with future support in packages/edge/infra/guard/core/Cargo.toml
• Implemented cache TTL constants: 10 minutes for route cache, 1 hour for proxy state cache in proxy_service.rs
• Replaced SccHashMap with moka::future::Cache for route caching, rate limiters, and in-flight counters
• Added mutex protection around rate limiters and in-flight counters for thread safety
• Updated metrics to use Moka's entry_count() for accurate cache size tracking

_{2 file(s) reviewed, 1 comment(s)
Edit PR Review Bot Settings | Greptile}

[greptile-apps]

logic: Potential deadlock risk if the limiter lock is held during await points. Consider using a timeout on the lock acquisition.

Suggested change

	// Try to acquire from the limiter
	let result = {
	let mut limiter = limiter_arc.lock().await;
	limiter.try_acquire()
	};
	Ok(result)
	}
	Ok(result)
	// Try to acquire from the limiter with timeout
	let result = match tokio::time::timeout(
	Duration::from_secs(5),
	limiter_arc.lock()
	).await {
	Ok(Ok(mut limiter)) => limiter.try_acquire(),
	Ok(Err(_)) => false, // Lock poisoned
	Err(_) => false, // Lock timeout
	};
	Ok(result)

[cloudflare-workers-and-pages]

Deploying rivet with    Cloudflare Pages

Latest commit:	6caa5bf
Status:	✅  Deploy successful!
Preview URL:	https://0d946734.rivet.pages.dev
Branch Preview URL:	https://05-28-fix-guard-replace-inte.rivet.pages.dev

View logs

[cloudflare-workers-and-pages]

Deploying rivet-hub with    Cloudflare Pages

Latest commit:	6caa5bf
Status:	✅  Deploy successful!
Preview URL:	https://85366007.rivet-hub-7jb.pages.dev
Branch Preview URL:	https://05-28-fix-guard-replace-inte.rivet-hub-7jb.pages.dev

View logs

[cloudflare-workers-and-pages]

Deploying rivet-studio with    Cloudflare Pages

Latest commit:	6caa5bf
Status:	✅  Deploy successful!
Preview URL:	https://3ab60aec.rivet-studio.pages.dev
Branch Preview URL:	https://05-28-fix-guard-replace-inte.rivet-studio.pages.dev

View logs