feat: implement Waku API Health spec (WIP) #3689
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,19 @@ | ||
| {.push raises: [].} | ||
|
|
||
| import system, std/json | ||
| import ./json_base_event | ||
| import ../../waku/api/types | ||
|
|
||
| type JsonConnectionStatusChangeEvent* = ref object of JsonEvent | ||
| status*: ConnectionStatus | ||
|
|
||
| proc new*( | ||
| T: type JsonConnectionStatusChangeEvent, status: ConnectionStatus | ||
| ): T = | ||
| return JsonConnectionStatusChangeEvent( | ||
| eventType: "node_health_change", | ||
| status: status | ||
| ) | ||
|
|
||
| method `$`*(event: JsonConnectionStatusChangeEvent): string = | ||
| $(%*event) |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change | ||||
|---|---|---|---|---|---|---|
| @@ -0,0 +1,324 @@ | ||||||
| {.used.} | ||||||
|
|
||||||
| import | ||||||
| std/[json, options, sequtils, strutils, tables], testutils/unittests, chronos, results | ||||||
|
|
||||||
| import | ||||||
| waku/[ | ||||||
| waku_core, | ||||||
| common/waku_protocol, | ||||||
| node/waku_node, | ||||||
| node/peer_manager, | ||||||
| node/health_monitor/health_status, | ||||||
| node/health_monitor/connection_status, | ||||||
| node/health_monitor/protocol_health, | ||||||
| node/health_monitor/node_health_monitor, | ||||||
| node/kernel_api/relay, | ||||||
| node/kernel_api/store, | ||||||
| node/kernel_api/lightpush, | ||||||
| node/kernel_api/filter, | ||||||
| waku_archive, | ||||||
| ] | ||||||
|
|
||||||
| import ../testlib/[wakunode, wakucore], ../waku_archive/archive_utils | ||||||
|
|
||||||
| proc p(kind: WakuProtocol, health: HealthStatus): ProtocolHealth = | ||||||
| var ph = ProtocolHealth.init(kind) | ||||||
| if health == HealthStatus.READY: | ||||||
| return ph.ready() | ||||||
| else: | ||||||
| return ph.notReady("mock") | ||||||
|
|
||||||
| suite "Health Monitor - health state calculation": | ||||||
| test "Disconnected, zero peers": | ||||||
| let protocols = | ||||||
| @[ | ||||||
| p(RelayProtocol, HealthStatus.NOT_READY), | ||||||
| p(StoreClientProtocol, HealthStatus.NOT_READY), | ||||||
| p(FilterClientProtocol, HealthStatus.NOT_READY), | ||||||
| p(LightpushClientProtocol, HealthStatus.NOT_READY), | ||||||
| ] | ||||||
| let strength = initTable[WakuProtocol, int]() | ||||||
| let state = | ||||||
| calculateConnectionState(protocols, strength, DefaultRelayFailoverThreshold) | ||||||
|
Collaborator
There was a problem hiding this comment. Shouldn't be I'd assume state == Connected when DefaultRelayFailoverThreshold
Contributor
Author
There was a problem hiding this comment. This test is mocking basically everything and is just testing the math behind the ConnectionStatus calculation. Here |
||||||
| check state == ConnectionStatus.Disconnected | ||||||
|
|
||||||
| test "PartiallyConnected, weak relay": | ||||||
| let weakCount = DefaultRelayFailoverThreshold - 1 | ||||||
| let protocols = | ||||||
| @[ | ||||||
| p(RelayProtocol, HealthStatus.READY), p(StoreClientProtocol, HealthStatus.READY) | ||||||
| ] | ||||||
| var strength = initTable[WakuProtocol, int]() | ||||||
| strength[RelayProtocol] = weakCount | ||||||
| strength[StoreClientProtocol] = 1 | ||||||
| let state = | ||||||
| calculateConnectionState(protocols, strength, DefaultRelayFailoverThreshold) | ||||||
| check state == ConnectionStatus.PartiallyConnected | ||||||
|
Collaborator
There was a problem hiding this comment. Maybe a comment explaining why "PartiallyConnected" would help. |
||||||
|
|
||||||
| test "Connected, robust relay": | ||||||
| let protocols = | ||||||
| @[ | ||||||
| p(RelayProtocol, HealthStatus.READY), p(StoreClientProtocol, HealthStatus.READY) | ||||||
| ] | ||||||
| var strength = initTable[WakuProtocol, int]() | ||||||
| strength[RelayProtocol] = DefaultRelayFailoverThreshold | ||||||
| strength[StoreClientProtocol] = FailoverThreshold | ||||||
|
Collaborator
There was a problem hiding this comment. Maybe :)?
Suggested change
|
||||||
| let state = | ||||||
| calculateConnectionState(protocols, strength, DefaultRelayFailoverThreshold) | ||||||
| check state == ConnectionStatus.Connected | ||||||
|
|
||||||
| test "Connected, robust edge": | ||||||
| let protocols = | ||||||
| @[ | ||||||
| p(RelayProtocol, HealthStatus.NOT_MOUNTED), | ||||||
| p(LightpushClientProtocol, HealthStatus.READY), | ||||||
| p(FilterClientProtocol, HealthStatus.READY), | ||||||
| p(StoreClientProtocol, HealthStatus.READY), | ||||||
| ] | ||||||
| var strength = initTable[WakuProtocol, int]() | ||||||
| strength[LightpushClientProtocol] = FailoverThreshold | ||||||
| strength[FilterClientProtocol] = FailoverThreshold | ||||||
| strength[StoreClientProtocol] = FailoverThreshold | ||||||
| let state = | ||||||
| calculateConnectionState(protocols, strength, DefaultRelayFailoverThreshold) | ||||||
| check state == ConnectionStatus.Connected | ||||||
|
|
||||||
| test "Disconnected, edge missing store": | ||||||
| let protocols = | ||||||
| @[ | ||||||
| p(LightpushClientProtocol, HealthStatus.READY), | ||||||
| p(FilterClientProtocol, HealthStatus.READY), | ||||||
| p(StoreClientProtocol, HealthStatus.NOT_READY), | ||||||
| ] | ||||||
| var strength = initTable[WakuProtocol, int]() | ||||||
| strength[LightpushClientProtocol] = FailoverThreshold | ||||||
| strength[FilterClientProtocol] = FailoverThreshold | ||||||
| strength[StoreClientProtocol] = 0 | ||||||
| let state = | ||||||
| calculateConnectionState(protocols, strength, DefaultRelayFailoverThreshold) | ||||||
| check state == ConnectionStatus.Disconnected | ||||||
|
|
||||||
| test "PartiallyConnected, edge meets minimum failover requirement": | ||||||
| let weakCount = max(1, FailoverThreshold - 1) | ||||||
| let protocols = | ||||||
| @[ | ||||||
| p(LightpushClientProtocol, HealthStatus.READY), | ||||||
| p(FilterClientProtocol, HealthStatus.READY), | ||||||
| p(StoreClientProtocol, HealthStatus.READY), | ||||||
| ] | ||||||
| var strength = initTable[WakuProtocol, int]() | ||||||
| strength[LightpushClientProtocol] = weakCount | ||||||
| strength[FilterClientProtocol] = weakCount | ||||||
| strength[StoreClientProtocol] = weakCount | ||||||
| let state = | ||||||
| calculateConnectionState(protocols, strength, DefaultRelayFailoverThreshold) | ||||||
| check state == ConnectionStatus.PartiallyConnected | ||||||
|
|
||||||
| test "Connected, robust relay ignores store server": | ||||||
| let protocols = | ||||||
| @[p(RelayProtocol, HealthStatus.READY), p(StoreProtocol, HealthStatus.READY)] | ||||||
| var strength = initTable[WakuProtocol, int]() | ||||||
| strength[RelayProtocol] = DefaultRelayFailoverThreshold | ||||||
| strength[StoreProtocol] = 0 | ||||||
| let state = | ||||||
| calculateConnectionState(protocols, strength, DefaultRelayFailoverThreshold) | ||||||
| check state == ConnectionStatus.Connected | ||||||
|
|
||||||
| test "Connected, robust relay ignores store client": | ||||||
| let protocols = | ||||||
| @[ | ||||||
| p(RelayProtocol, HealthStatus.READY), | ||||||
| p(StoreProtocol, HealthStatus.READY), | ||||||
| p(StoreClientProtocol, HealthStatus.NOT_READY), | ||||||
| ] | ||||||
| var strength = initTable[WakuProtocol, int]() | ||||||
| strength[RelayProtocol] = DefaultRelayFailoverThreshold | ||||||
| strength[StoreProtocol] = 0 | ||||||
| strength[StoreClientProtocol] = 0 | ||||||
| let state = | ||||||
| calculateConnectionState(protocols, strength, DefaultRelayFailoverThreshold) | ||||||
| check state == ConnectionStatus.Connected | ||||||
|
|
||||||
| suite "Health Monitor - events": | ||||||
| asyncTest "Core (relay) health update": | ||||||
| let | ||||||
| nodeAKey = generateSecp256k1Key() | ||||||
| nodeA = newTestWakuNode(nodeAKey, parseIpAddress("127.0.0.1"), Port(0)) | ||||||
|
|
||||||
| (await nodeA.mountRelay()).expect("Node A failed to mount Relay") | ||||||
|
|
||||||
| nodeA.mountStoreClient() | ||||||
|
Collaborator
There was a problem hiding this comment. This is not needed. That happens in |
||||||
|
|
||||||
| await nodeA.start() | ||||||
|
|
||||||
| let monitorA = NodeHealthMonitor.new() | ||||||
| monitorA.setNodeToHealthMonitor(nodeA) | ||||||
|
|
||||||
| var | ||||||
| lastStatus = ConnectionStatus.Disconnected | ||||||
| callbackCount = 0 | ||||||
| healthChangeSignal = newFuture[void]() | ||||||
|
|
||||||
| monitorA.onConnectionStatusChange = proc(status: ConnectionStatus) {.async.} = | ||||||
| lastStatus = status | ||||||
| callbackCount.inc() | ||||||
| if not healthChangeSignal.finished: | ||||||
| healthChangeSignal.complete() | ||||||
|
|
||||||
| monitorA.startHealthMonitor().expect("Health monitor failed to start") | ||||||
|
|
||||||
| let | ||||||
| nodeBKey = generateSecp256k1Key() | ||||||
| nodeB = newTestWakuNode(nodeBKey, parseIpAddress("127.0.0.1"), Port(0)) | ||||||
|
|
||||||
| let driver = newSqliteArchiveDriver() | ||||||
| nodeB.mountArchive(driver).expect("Node B failed to mount archive") | ||||||
|
|
||||||
| (await nodeB.mountRelay()).expect("Node B failed to mount relay") | ||||||
| await nodeB.mountStore() | ||||||
|
|
||||||
| await nodeB.start() | ||||||
|
|
||||||
| await nodeA.connectToNodes(@[nodeB.switch.peerInfo.toRemotePeerInfo()]) | ||||||
|
|
||||||
| proc dummyHandler(topic: PubsubTopic, msg: WakuMessage): Future[void] {.async.} = | ||||||
| discard | ||||||
|
|
||||||
| nodeA.subscribe((kind: PubsubSub, topic: DefaultPubsubTopic), dummyHandler).expect( | ||||||
| "Node A failed to subscribe" | ||||||
| ) | ||||||
| nodeB.subscribe((kind: PubsubSub, topic: DefaultPubsubTopic), dummyHandler).expect( | ||||||
| "Node B failed to subscribe" | ||||||
| ) | ||||||
|
Comment on lines
+177
to
+182
Collaborator
There was a problem hiding this comment. These might not be needed as
Contributor
Author
There was a problem hiding this comment. Taking these out breaks the test unfortunately. The newTestWakuNode helper just passes this to the "enrBuilder"; it doesn't seem to execute relay subscriptions with it, like |
||||||
|
|
||||||
| let connectTimeLimit = Moment.now() + 10.seconds | ||||||
|
Collaborator
There was a problem hiding this comment. We cannot wait so long :)
Suggested change
|
||||||
| var gotConnected = false | ||||||
|
|
||||||
| while Moment.now() < connectTimeLimit: | ||||||
| if lastStatus != ConnectionStatus.Disconnected: | ||||||
| gotConnected = true | ||||||
| break | ||||||
|
|
||||||
| if healthChangeSignal.finished: | ||||||
| healthChangeSignal = newFuture[void]() | ||||||
|
Collaborator
There was a problem hiding this comment. Maybe not needed as it happens a few lines below
Contributor
Author
There was a problem hiding this comment. Changed from void Future to AsyncEvent to make it cleaner and clearer |
||||||
|
|
||||||
| discard await healthChangeSignal.withTimeout(connectTimeLimit - Moment.now()) | ||||||
|
|
||||||
| check: | ||||||
| gotConnected == true | ||||||
| callbackCount >= 1 | ||||||
|
|
||||||
| if healthChangeSignal.finished: | ||||||
| healthChangeSignal = newFuture[void]() | ||||||
|
|
||||||
| await nodeB.stop() | ||||||
| await nodeA.disconnectNode(nodeB.switch.peerInfo.toRemotePeerInfo()) | ||||||
|
|
||||||
| let disconnectTimeLimit = Moment.now() + 10.seconds | ||||||
| var gotDisconnected = false | ||||||
|
|
||||||
| while Moment.now() < disconnectTimeLimit: | ||||||
| if lastStatus == ConnectionStatus.Disconnected: | ||||||
| gotDisconnected = true | ||||||
| break | ||||||
|
|
||||||
| if healthChangeSignal.finished: | ||||||
| healthChangeSignal = newFuture[void]() | ||||||
|
|
||||||
| discard await healthChangeSignal.withTimeout(disconnectTimeLimit - Moment.now()) | ||||||
|
|
||||||
| check: | ||||||
| gotDisconnected == true | ||||||
|
|
||||||
| await monitorA.stopHealthMonitor() | ||||||
| await nodeA.stop() | ||||||
|
|
||||||
| asyncTest "Edge (light client) health update": | ||||||
| let | ||||||
| nodeAKey = generateSecp256k1Key() | ||||||
| nodeA = newTestWakuNode(nodeAKey, parseIpAddress("127.0.0.1"), Port(0)) | ||||||
|
|
||||||
| nodeA.mountLightpushClient() | ||||||
| await nodeA.mountFilterClient() | ||||||
| nodeA.mountStoreClient() | ||||||
|
|
||||||
| await nodeA.start() | ||||||
|
|
||||||
| let monitorA = NodeHealthMonitor.new() | ||||||
| monitorA.setNodeToHealthMonitor(nodeA) | ||||||
|
|
||||||
| var | ||||||
| lastStatus = ConnectionStatus.Disconnected | ||||||
| callbackCount = 0 | ||||||
| healthChangeSignal = newFuture[void]() | ||||||
|
|
||||||
| monitorA.onConnectionStatusChange = proc(status: ConnectionStatus) {.async.} = | ||||||
| lastStatus = status | ||||||
| callbackCount.inc() | ||||||
| if not healthChangeSignal.finished: | ||||||
| healthChangeSignal.complete() | ||||||
|
|
||||||
| monitorA.startHealthMonitor().expect("Health monitor failed to start") | ||||||
|
|
||||||
| let | ||||||
| nodeBKey = generateSecp256k1Key() | ||||||
| nodeB = newTestWakuNode(nodeBKey, parseIpAddress("127.0.0.1"), Port(0)) | ||||||
|
|
||||||
| let driver = newSqliteArchiveDriver() | ||||||
| nodeB.mountArchive(driver).expect("Node B failed to mount archive") | ||||||
|
|
||||||
| (await nodeB.mountRelay()).expect("Node B failed to mount relay") | ||||||
|
|
||||||
| (await nodeB.mountLightpush()).expect("Node B failed to mount lightpush") | ||||||
| await nodeB.mountFilter() | ||||||
| await nodeB.mountStore() | ||||||
|
|
||||||
| await nodeB.start() | ||||||
|
|
||||||
| await nodeA.connectToNodes(@[nodeB.switch.peerInfo.toRemotePeerInfo()]) | ||||||
|
|
||||||
| let connectTimeLimit = Moment.now() + 10.seconds | ||||||
| var gotConnected = false | ||||||
|
|
||||||
| while Moment.now() < connectTimeLimit: | ||||||
| if lastStatus == ConnectionStatus.PartiallyConnected: | ||||||
| gotConnected = true | ||||||
| break | ||||||
|
|
||||||
| if healthChangeSignal.finished: | ||||||
| healthChangeSignal = newFuture[void]() | ||||||
|
|
||||||
| discard await healthChangeSignal.withTimeout(connectTimeLimit - Moment.now()) | ||||||
|
|
||||||
| check: | ||||||
| gotConnected == true | ||||||
| callbackCount >= 1 | ||||||
| lastStatus == ConnectionStatus.PartiallyConnected | ||||||
|
|
||||||
| if healthChangeSignal.finished: | ||||||
| healthChangeSignal = newFuture[void]() | ||||||
|
|
||||||
| await nodeB.stop() | ||||||
| await nodeA.disconnectNode(nodeB.switch.peerInfo.toRemotePeerInfo()) | ||||||
|
|
||||||
| let disconnectTimeLimit = Moment.now() + 10.seconds | ||||||
| var gotDisconnected = false | ||||||
|
|
||||||
|
Collaborator
There was a problem hiding this comment. I think is interesting to confirm the opposite state happens before entering the loop.
Suggested change
|
||||||
| while Moment.now() < disconnectTimeLimit: | ||||||
| if lastStatus == ConnectionStatus.Disconnected: | ||||||
| gotDisconnected = true | ||||||
| break | ||||||
|
|
||||||
| if healthChangeSignal.finished: | ||||||
| healthChangeSignal = newFuture[void]() | ||||||
|
|
||||||
| discard await healthChangeSignal.withTimeout(disconnectTimeLimit - Moment.now()) | ||||||
|
|
||||||
| check: | ||||||
| gotDisconnected == true | ||||||
| lastStatus == ConnectionStatus.Disconnected | ||||||
|
|
||||||
| await monitorA.stopHealthMonitor() | ||||||
| await nodeA.stop() | ||||||
|
|
||||||
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.
Oops, something went wrong.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Better use a longer name so it's easier to lookup.