fix(dotcom): increase zero-cache volume sizes and scope kill switch reload (tldraw#8407)

Production zero-cache (replication manager) was crash-looping due to
`SQLITE_FULL` — the 1GB default Fly.io volume filled up when the backup
replicator tried to write. This PR sets explicit volume sizes in the
Fly.io deploy templates (8GB for production, 1GB for staging/preview)
and narrows the zero kill switch reload to only affect users actually
running proper Zero.

### Change type

- [x] `bugfix`

### Test plan

1. Deploy to staging and verify zero-cache volumes are created with
correct `initial_size`
2. Enable `zero_kill_switch` flag and verify only users with
`zero_enabled` or localStorage override get reloaded

### Release notes

- Fix zero-cache crash loop caused by undersized Fly.io volumes in
production

### Code changes

| Section        | LOC change |
| -------------- | ---------- |
| Apps           | +14 / -4   |
| Config/tooling | +5 / -1    |

Author: MitjaBezensek

apps/dotcom/client/src/tla/utils/FeatureFlagPoller.test.ts

@@ -2,8 +2,12 @@ import { beforeAll, beforeEach, describe, expect, it, vi } from 'vitest'
 import type { FeatureFlags } from './FeatureFlagPoller'
 
 const mockFetch = vi.fn()
+let mockLocalStorage: Record<string, string> = {}
 vi.mock('tldraw', () => {
-	return { fetch: (...args: any[]) => mockFetch(...args) }
+	return {
+		fetch: (...args: any[]) => mockFetch(...args),
+		getFromLocalStorage: (key: string) => mockLocalStorage[key] ?? null,
+	}
 })
 
 function makeFlags(overrides: Partial<FeatureFlags> = {}): FeatureFlags {
@@ -33,19 +37,42 @@ describe('shouldReloadForFlagChange', () => {
 	})
 
 	describe('zero_kill_switch transitions', () => {
-		it('reloads when kill switch goes false → true', () => {
-			const prev = makeFlags({ zero_kill_switch: { enabled: false } })
+		it('reloads when kill switch goes false → true and user had zero enabled', () => {
+			const prev = makeFlags({
+				zero_kill_switch: { enabled: false },
+				zero_enabled: { enabled: true },
+			})
 			const next = makeFlags({ zero_kill_switch: { enabled: true } })
 			expect(shouldReloadForFlagChange(prev, next)).toBe(true)
 		})
 
-		it('reloads when kill switch goes undefined → true', () => {
-			const prev = makeFlags()
+		it('reloads when kill switch goes undefined → true and user had zero enabled', () => {
+			const prev = makeFlags({ zero_enabled: { enabled: true } })
 			delete (prev as any).zero_kill_switch
 			const next = makeFlags({ zero_kill_switch: { enabled: true } })
 			expect(shouldReloadForFlagChange(prev, next)).toBe(true)
 		})
 
+		it('does NOT reload when kill switch activates but user did not have zero enabled', () => {
+			const prev = makeFlags({
+				zero_kill_switch: { enabled: false },
+				zero_enabled: { enabled: false },
+			})
+			const next = makeFlags({ zero_kill_switch: { enabled: true } })
+			expect(shouldReloadForFlagChange(prev, next)).toBe(false)
+		})
+
+		it('reloads when kill switch activates and user has localStorage override', () => {
+			mockLocalStorage = { useProperZero: 'true' }
+			const prev = makeFlags({
+				zero_kill_switch: { enabled: false },
+				zero_enabled: { enabled: false },
+			})
+			const next = makeFlags({ zero_kill_switch: { enabled: true } })
+			expect(shouldReloadForFlagChange(prev, next)).toBe(true)
+			mockLocalStorage = {}
+		})
+
 		it('does NOT reload when kill switch stays true → true', () => {
 			const prev = makeFlags({ zero_kill_switch: { enabled: true } })
 			const next = makeFlags({ zero_kill_switch: { enabled: true } })

apps/dotcom/client/src/tla/utils/FeatureFlagPoller.tsx

@@ -1,6 +1,6 @@
 import { EvaluatedFeatureFlag, FeatureFlagKey } from '@tldraw/dotcom-shared'
 import { useEffect } from 'react'
-import { fetch } from 'tldraw'
+import { fetch, getFromLocalStorage } from 'tldraw'
 
 export type FeatureFlags = Record<FeatureFlagKey, EvaluatedFeatureFlag>
 
@@ -52,12 +52,16 @@ const REFETCH_INTERVAL = 60000 // 1 minute
 
 /**
  * Determines whether a flag change should trigger a page reload.
- * Only reloads when zero_kill_switch transitions to true from any non-true state.
+ * Only reloads when zero_kill_switch transitions to true AND the user
+ * was actually using proper Zero (no point reloading polyfill users).
  */
 export function shouldReloadForFlagChange(prev: FeatureFlags, next: FeatureFlags): boolean {
 	const prevKillSwitch = prev.zero_kill_switch?.enabled
 	const nextKillSwitch = next.zero_kill_switch?.enabled
-	return nextKillSwitch === true && prevKillSwitch !== true
+	if (nextKillSwitch !== true || prevKillSwitch === true) return false
+	// Only reload if this user was actually on proper Zero
+	const wasUsingZero = prev.zero_enabled?.enabled || getFromLocalStorage('useProperZero') === 'true'
+	return !!wasUsingZero
 }
 
 /**

apps/dotcom/zero-cache/flyio-replication-manager.template.toml

@@ -28,6 +28,7 @@ cpus = __VM_CPUS
 [mounts]
 source = "sqlite_repl"
 destination = "/data"
+initial_size = "__VOLUME_SIZE"
 
 [[files]]
 guest_path = "/etc/litestream.yml"

apps/dotcom/zero-cache/flyio-view-syncer.template.toml

@@ -31,6 +31,7 @@ cpus = __VM_CPUS
 [mounts]
 source = "sqlite_db"
 destination = "/data"
+initial_size = "__VOLUME_SIZE"
 
 [[files]]
 guest_path = "/etc/litestream.yml"

apps/dotcom/zero-cache/flyio.template.toml

@@ -25,6 +25,7 @@ cpus = 2
 [mounts]
 source = "sqlite_db"
 destination = "/data"
+initial_size = "1gb"
 
 [env]
 ZERO_REPLICA_FILE = "/data/sync-replica.db"

internal/scripts/deploy-dotcom.ts

@@ -285,8 +285,12 @@ const zeroQueryUrl = `${env.MULTIPLAYER_SERVER.replace(/^ws/, 'http')}/app/zero/
 // Fly.io shared-cpu VM sizes per environment.
 // Max shared-cpu is 8x (8 CPUs, 16 GB RAM).
 const zeroVmSizes = {
-	staging: { rm: { cpus: 1, memory: '2gb' }, vs: { cpus: 2, memory: '4gb' } },
-	production: { rm: { cpus: 4, memory: '8gb' }, vs: { cpus: 8, memory: '16gb' } },
+	staging: { rm: { cpus: 1, memory: '2gb' }, vs: { cpus: 2, memory: '4gb' }, volumeSize: '1gb' },
+	production: {
+		rm: { cpus: 4, memory: '8gb' },
+		vs: { cpus: 8, memory: '16gb' },
+		volumeSize: '8gb',
+	},
 	preview: { single: { cpus: 2, memory: '2gb' } },
 } as const
 
@@ -329,6 +333,7 @@ interface SingleNodeVmSizes {
 interface MultiNodeVmSizes {
 	rm: VmSize
 	vs: VmSize
+	volumeSize: string
 }
 const zeroVm = zeroVmSizes[env.TLDRAW_ENV as keyof typeof zeroVmSizes] as
 	| SingleNodeVmSizes
@@ -732,6 +737,7 @@ function updateFlyioReplicationManagerToml(appName: string, backupPath: string):
 		.replaceAll('__RM_CHANGE_MAX_CONNS', String(zeroConns.rm.change))
 		.replaceAll('__VM_CPUS', String(zeroVm.rm.cpus))
 		.replaceAll('__VM_MEMORY', zeroVm.rm.memory)
+		.replaceAll('__VOLUME_SIZE', zeroVm.volumeSize)
 
 	fs.writeFileSync(flyioTomlFile, updatedContent, 'utf-8')
 }
@@ -767,6 +773,7 @@ function updateFlyioViewSyncerToml(
 		.replaceAll('__VS_CHANGE_MAX_CONNS', String(zeroConns.vs.change))
 		.replaceAll('__VM_CPUS', String(zeroVm.vs.cpus))
 		.replaceAll('__VM_MEMORY', zeroVm.vs.memory)
+		.replaceAll('__VOLUME_SIZE', zeroVm.volumeSize)
 
 	fs.writeFileSync(flyioTomlFile, updatedContent, 'utf-8')