simplification of peerset api #2123

debris · 2019-03-26T21:03:07Z

not using Arc or Mutex in a peerset module
renaming Peerset to PeersetHandle
renaming PeersetMut to Peerset
tests
updated linked_hash_map crate to 0.5.2
bugfixes/changes:
- overriding slots in add_reserved_peer
- connecting to bootnodes before connecting to reserved nodes
- not connecting to reserved nodes in alloc_slots if reserved_only is set
- set_reserved_only(false) tries to reconnect us to not reserved nodes
- remove_reserved_peer disconnects us from the peer if reserved_only = true
- incoming accepts connections only from reserved peers in discovered queue in reserved_only = true

gterzian · 2019-03-27T12:10:55Z

core/peerset/src/lib.rs


 	fn poll(&mut self) -> Poll<Option<Self::Item>, Self::Error> {
-		self.rx.poll()
+		loop {


Do I understand it correctly that the loop is necessary because the various calls to self.message_queue.push_back will not wake-up the current task?

Have you considered instead having the various on_ methods return a Option<Self::Item> directly? Since if let Some(message) immediately returns, I assume there isn't a case where any of the on_ methods will create more than one message?

Do I understand it correctly that the loop is necessary because the various calls to self.message_queue.push_back will not wake-up the current task?

It's necessary, cause on_ methods may add multiple messages or may not add them at all. In the later case, we need to process next Action.

Have you considered instead having the various on_ methods return a Option<Self::Item> directly? Since if let Some(message) immediately returns, I assume there isn't a case where any of the on_ methods will create more than one message?

Yes, I considered it, but it's less efficient and may be confusing.

on_ methods would require &mut self even if they don't touch self.message_queue. So we would have an on_ method that mutates self and returns the result that should be always appended to self.message_queue. It would be confusing and could lead to incorrect behaviours

on_ methods would have to return Vec<Self::Item> so we would have an additional allocation of on a heap of something that is rewritten to self.message_queue a second after

Ok, I see, wouldn't it be possible for messsages in message_queue to be stuck there in case nothing new is sent on the receiver? Let's say one of the _on methods adds 2 items to message_queue, the first one get's returned by the if let Some(message) and the second one is still inside message_queue. If nothing wakes up the current task, the second message in message_qeue could never be handled? Maybe schedule a wake-up if the message_queue is non-empty before returning from poll?

Ok, I see, wouldn't it be possible for messsages in message_queue to be stuck there in case nothing new is sent on the receiver?

It's a good question, but no.

Let's say one of the _on methods adds 2 items to message_queue, the first one get's returned by the if let Some(message) and the second one is still inside message_queue. If nothing wakes up the current task, the second message in message_qeue could never be handled? Maybe schedule a wake-up if the message_queue is non-empty before returning from poll

If a stream returns a single item, it will be waken up subsequently until it returns Async::NotReady or Async::Ready(None). So if we push multiple messages to message_queue, it will be polled until all the items from message_queue are drained

Ok, I didn't know one could rely on the stream being woken-up later by returning Async::Ready(Some). I thought one had to have another way to register the current task for being awaken, for example in the current case by calling self.rx.poll(), which registers the current task for being woken-up if a message is sent on the corrseponding sender.

tomusdrw

Looks nice, a bit of boilerplate code that I think could be replaced by a library (isn't there one?)

tomusdrw · 2019-03-28T09:49:00Z

core/network-libp2p/src/custom_proto/behaviour.rs

 			PeerState::DisabledPendingEnable { open, connected_point, timer } => {
 				debug!(target: "sub-libp2p", "PSM <= Dropped({:?})", peer_id);
-				self.peerset.dropped(peer_id);
+				self.peerset.dropped(peer_id.clone());


Just out of curiosity, why do you think it's better to clone on the call site? My preference would be to pass a reference and clone only if needed. In that case it doesn't really matter, since we know that current implementation requires ownership anyway, but IMHO passing reference is more future proof (shall the impl change).

Just out of curiosity, why do you think it's better to clone on the call site?

No, but we always need a value, cause we send it over the channel. That's why I changed the function signature

tomusdrw · 2019-03-28T09:50:21Z

core/peerset/src/lib.rs


-/// Message that can be sent by the peer set manager (PSM).
 #[derive(Debug)]
+enum Action {


Sounds like a good idea for proc-macro library :)

definitely :) if we only have more messages or if we do it in more than one place. Right now proc-macro library would probably require writing more code than this boilerplate does

tomusdrw · 2019-03-28T09:52:04Z

core/peerset/src/lib.rs

+	message_queue: VecDeque<Message>,
+}
+
+impl ops::Deref for Peerset {


I don't think it's a good use case for Deref, from the docs:

Deref should only be implemented for smart pointers to avoid confusion.

hmm... I was not aware of this. I added Deref for convenience

should I remove this ops::Deref?

Yeah, if it's ok to use it directly. Alternatively add AsRef if you want to pass it to some generic code.

tomusdrw · 2019-03-28T09:52:38Z

core/peerset/src/lib.rs

-	/// Reports an adjustement to the reputation of the given peer.
-	pub fn report_peer(&self, _peer_id: &PeerId, _score_diff: i32) {
+	fn on_report_peer(&self, _peer_id: PeerId, _score_diff: i32) {
+		//unimplemented!();


Should most likely be removed

it needs implementation, I'm adding it right now

bkchr · 2019-03-28T11:13:47Z

core/peerset/src/lib.rs

-			*slot = Some(elem.clone());
-			let _ = tx.unbounded_send(Message::Connect(elem));
+		for slot in self.data.out_slots.iter_mut() {
+			if slot.is_some() {


move into a filter?

the implementation here remained unchanged, I just moved the function

Still could be placed into a filter :)

yes! I was supposed to give you thumbs up, not confused emoji :)

bkchr · 2019-03-28T11:18:13Z

core/peerset/src/lib.rs

+			if let Some(message) = self.message_queue.pop_front() {
+				return Ok(Async::Ready(Some(message)));
+			}
+			match self.rx.poll()? {


Suggested change

match self.rx.poll()? {

match try_ready!(self.rx.poll()) {

debris · 2019-03-28T12:04:00Z

core/peerset/src/lib.rs

-		if let Some(pos) = self.data.out_slots.iter().position(|s| s.as_ref().map(|n| !self.data.reserved.contains(n)).unwrap_or(true)) {
+		// TODO: override slots that are occupied by not reserved peers
+		// send Message::Drop in those cases
+		if let Some(pos) = self.data.out_slots.iter().position(Option::is_none) {


this is a bugfix. bootnodes were overwritten by reserved peers, but no message was created

tomaka · 2019-03-28T12:25:41Z

core/peerset/src/lib.rs

+	///
+	/// Because of concurrency issues, it is acceptable to call `incoming` with a `PeerId` the
+	/// peerset is already connected to, in which case it must not answer.
+	pub fn incoming(&self, peer_id: PeerId, index: IncomingIndex) {


The incoming, dropped and discovered methods were only on the PeersetMut (now Peerset) because it doesn't make sense for them to be accessible from anything else than the networking. If someone calls them you will just crate an inconsistent state in the network layer. If for some reason we have to keep that as is, please put a huge warning that says "don't call these functions".

Why would anyone want to call this functions? Calling them is a logical error, same as doing

let a = None; if let Some(value) = a { // ... }

Peerset is a futures::Stream and I believe it should be responsible only for producing items, nothing else.

If we want to limit how and where the methods are used, I believe we should have 2 different handles. One with incoming, dropped and discovered methods and the second one with everything else

Calling them is a logical error

Yeah, and it would be even better if the type system prevents these logic errors. I don't really want to have to deal with someone calling dropped thinking it disconnects a node, and I'm pretty sure it will happen.

Peerset is a futures::Stream and I believe it should be responsible only for producing items, nothing else.

Why is that?

Peerset is a futures::Stream and I believe it should be responsible only for producing items, nothing else.

Why is that?

So we can use combinators (.map, .then) on the stream

The only thing that will use the Peerset is the networking code which already exists and doesn't use combinators. I don't think we should have a weird API and add a performance hit because of usability, especially since usability here is a non-issue.

tomaka · 2019-03-28T12:27:32Z

core/peerset/src/lib.rs

 impl Peerset {
 	/// Builds a new peerset from the given configuration.
-	pub fn from_config(config: PeersetConfig) -> (Arc<Peerset>, PeersetMut) {
+	pub fn from_config(config: PeersetConfig) -> Peerset {


Being able to create handles from the Peerset means that we will forever have to keep in the Peerset the building blocks for a PeersetHandler unless we break the API. Whereas if we return a tuple, they can be separate.

you can always create PeersetHandle by cloning an existing one

I'm not strongly opinionated here. For me, this method could also return a tuple. But I see nothing bad, in Peerset keeping the building blocks for the PeersetHandler

gavofyork · 2019-04-02T07:59:47Z

core/peerset/src/lib.rs

+		let _ = self.tx.unbounded_send(Action::SetReservedOnly(reserved));
+	}
+
+	/// Reports an adjustement to the reputation of the given peer.


Suggested change

/// Reports an adjustement to the reputation of the given peer.

/// Reports an adjustment to the reputation of the given peer.

suggestions applied, please review again

debris · 2019-04-04T09:37:04Z

done, please review again

arkpar · 2019-04-04T12:35:43Z

Would love to see detailed tracing in the peerset module.

tomaka

Lots of comments but only the Accept/Reject need fixing.

tomaka · 2019-04-04T12:36:25Z

core/peerset/src/slots.rs

+	}
+
+	/// Returns Ok if we successfully connected to a given peer.
+	pub fn add_peer(&mut self, peer_id: PeerId, slot_type: SlotType) -> Result<(), SlotError>  {


From looking at the return values, I'm not sure if this should be a Result. You return Err even in situations that I would consider successful and where you change the state.
That can be improved later, though.

Fair enough, I'll change the Result

tomaka · 2019-04-04T12:39:26Z

core/peerset/src/slots.rs

+		self.slots.remove(peer_id);
+	}
+
+	pub fn is_reserved(&self, peer_id: &PeerId) -> bool {


Should be is_connected_and_reserved maybe?

all peers in slots are connected

Hence the name change to make that explicit.

tomaka · 2019-04-04T12:50:58Z

core/peerset/src/lib.rs

+		self.data.reserved_only = reserved_only;
+		if self.data.reserved_only {
+			for peer_id in self.data.in_slots.clear_common_slots().into_iter().chain(self.data.out_slots.clear_common_slots().into_iter()) {
+				// peer will be removed from `in_slots` or `out_slots` in `on_dropped` method


Comment is obsolete.

tomaka · 2019-04-04T12:51:11Z

core/peerset/src/lib.rs

+		if self.data.reserved_only {
+			for peer_id in self.data.in_slots.clear_common_slots().into_iter().chain(self.data.out_slots.clear_common_slots().into_iter()) {
+				// peer will be removed from `in_slots` or `out_slots` in `on_dropped` method
+				self.data.in_slots.clear_slot(&peer_id);


Is this line useful? in_slots should be cleared in the for, no?

good catch. it's the leftover from previous version of this function :)

tomaka · 2019-04-04T12:52:08Z

core/peerset/src/lib.rs

-	}
-}
+	fn on_report_peer(&mut self, peer_id: PeerId, score_diff: i32) {
+		let score = self.data.scores.entry(peer_id.clone()).or_default();


I think scores grows forever, right? For long running nodes it will take up a lot of memory.

true, we can use some lru_cache here.

The same problem exists with discovered. Should we limit the number of peers over there as well?

tomaka · 2019-04-04T12:52:42Z

core/peerset/src/lib.rs

-			continue;
+		if *score < 0 {
+			// peer will be removed from `in_slots` or `out_slots` in `on_dropped` method
+			if self.data.in_slots.contains(&peer_id) || self.data.out_slots.contains(&peer_id) {


Could combine contains and clear_slot into one method.

We can't. We need to know whether we have a connection to this slot to call Message::Drop

What I mean is that clear_slot could return an enum indicating whether we cleared something or not.

tomaka · 2019-04-04T12:53:12Z

core/peerset/src/lib.rs

+				Ok(_) => {
+					self.message_queue.push_back(Message::Connect(peer_id));
+				},
+				Err(SlotError::AlreadyConnected(_)) => (),


I'd print an error here, since there's an obvious problem somewhere.

good point! being here means that we have a logic error somewhere

tomaka · 2019-04-04T12:53:48Z

core/peerset/src/lib.rs

+
+		// check if we are already connected to this peer
+		if self.data.out_slots.contains(&peer_id) {
+			self.message_queue.push_back(Message::Reject(index));


Must not send Reject, see the doc-comment of the function.

Thanks for pointing this out. May I accept why is it designed like that?

I don't remember, to be honest. There was a long debate about this and this was the outcome.

tomaka · 2019-04-04T12:54:29Z

core/peerset/src/lib.rs

+					return;
+				},
+				Err(SlotError::DemandReroute { disconnect, .. }) => {
+					// disconnect not reserved node


Should send back Accept here, no?

no, we are in the loop. Accept will be send on the next loop iteration

…ved_only

debris · 2019-04-04T14:38:05Z

@tomaka please review again. I applied all suggestions/fixes

* Introduction of PeersetHandle * integrate PeersetHandle with the rest of the codebase * fix compilation errors * more tests for peerset, fixed overwriting bug in add_reserved_peer * Slots data structure and bugfixes for peerset * bend to pressure * updated lru-cache to 0.1.2 and updated linked-hash-map to 0.5.2 * peerset discovered list is now a LinkedHashMap * fix review suggestions * split back Peerset and PeersetHandle * test for Peerset::discovered * applied review suggestions * fixes to peerset::incoming * peerset disconnects are all instantaneous * instantaneous drop in peerset finished * Peerset::set_reserved_only can also reconnect nodes * Peerset scores cache uses lru-cache * remove redundant function call and comment from Peerset::on_set_reserved_only * add_peer returns SlotState enum * apply review suggestions * is_reserved -> is_connected_and_reserved

…y-by-default' Squashed commit of the following: commit 6de583a Author: Xiliang Chen <[email protected]> Date: Wed Apr 10 20:26:29 2019 +1200 update authers for rest of the node-template cargo.toml files (#2242) commit 5240bc1 Author: Bastian Köcher <[email protected]> Date: Tue Apr 9 10:31:18 2019 +0200 Throw a compile error for `on_finalise` and `on_initialise` (#2236) commit 67d2e71 Author: Pierre Krieger <[email protected]> Date: Tue Apr 9 05:30:43 2019 -0300 Add warning when using default protocol ID (#2234) * Add warning when using default protocol ID * Update core/service/src/lib.rs commit 1421fed Author: Xiliang Chen <[email protected]> Date: Tue Apr 9 17:22:20 2019 +1200 update name and authors to placeholder text for node-template (#2222) * update name and authors to placeholder text * revert package name change commit 6617f23 Author: André Silva <[email protected]> Date: Mon Apr 8 12:50:34 2019 +0100 grandpa: Voter persistence and upgrade to finality-grandpa v0.7 (#2139) * core: grandpa: migrate to grandpa 0.7 * core: grandpa: store current round votes and load them on startup * core: grandpa: resend old persisted votes for the current round * core: grandpa: store base and votes for last completed round * core: grandpa: fix latest grandpa 0.7 changes * core: grandpa: update to grandpa 0.7.1 * core: grandpa: persist votes for last two completed rounds * core: grandpa: simplify VoterSetState usage * core: grandpa: use Environment::update_voter_set_state * core: grandpa: fix aux_schema test * core: grandpa: add docs * core: grandpa: add note about environment assumption * core: grandpa: don't update voter set state on ignored votes * core: grandpa: add test for v1 -> v2 aux_schema migration * core: grandpa: add test for voter vote persistence * core: grandpa: use grandpa 0.7.1 from crates.io * core: grandpa: use try_init in test * core: grandpa: add comment about block_import in test * core: grandpa: avoid cloning HasVoted * core: grandpa: add missing docs * core: grandpa: cleanup up can_propose/prevote/precommit commit 21e0877 Author: Gregory Terzian <[email protected]> Date: Mon Apr 8 13:17:00 2019 +0200 remove clone bound on specialization in testnet factory (#2157) commit 7c64746 Author: Andrew Jones <[email protected]> Date: Sat Apr 6 12:23:56 2019 +0100 Contract import/export validation (#2203) * Reject validation of contract with unknown exports * Validate imports eagerly * Increment spec version commit 12718fa Author: Pierre Krieger <[email protected]> Date: Fri Apr 5 14:07:09 2019 -0300 Fix state inconsistency between handler and behaviour (#2220) * Fix state inconsistency between handler and behaviour * Fix the error! being in the wrong place commit f917d12 Author: Bastian Köcher <[email protected]> Date: Fri Apr 5 18:50:38 2019 +0200 Use `storage_root` of newly calculated header (#2216) Instead of calculating the `storage_root` a second time, we just can take the `storage_root` from the new header. commit 3359ce0 Author: Marek Kotewicz <[email protected]> Date: Fri Apr 5 14:44:46 2019 +0200 Peerset::discovered accepts many peer ids (#2213) * Peerset::discovered accepts many peer ids * Improve tracing in peerset commit dd82e0e Author: Marek Kotewicz <[email protected]> Date: Thu Apr 4 19:40:40 2019 +0200 simplification of peerset api (#2123) * Introduction of PeersetHandle * integrate PeersetHandle with the rest of the codebase * fix compilation errors * more tests for peerset, fixed overwriting bug in add_reserved_peer * Slots data structure and bugfixes for peerset * bend to pressure * updated lru-cache to 0.1.2 and updated linked-hash-map to 0.5.2 * peerset discovered list is now a LinkedHashMap * fix review suggestions * split back Peerset and PeersetHandle * test for Peerset::discovered * applied review suggestions * fixes to peerset::incoming * peerset disconnects are all instantaneous * instantaneous drop in peerset finished * Peerset::set_reserved_only can also reconnect nodes * Peerset scores cache uses lru-cache * remove redundant function call and comment from Peerset::on_set_reserved_only * add_peer returns SlotState enum * apply review suggestions * is_reserved -> is_connected_and_reserved commit d90833d Author: Arkadiy Paronyan <[email protected]> Date: Thu Apr 4 18:01:28 2019 +0200 Disconnect on protocol timeout (#2212) commit c0a46b5 Author: André Silva <[email protected]> Date: Thu Apr 4 15:56:49 2019 +0100 core: grandpa: verify commit target in justification (#2201) commit 3a4901a Author: Bastian Köcher <[email protected]> Date: Thu Apr 4 16:56:16 2019 +0200 Introduce `original_storage` and `original_storage_hash` (#2211) Both functions will ignore any overlayed changes and access the backend directly. commit a7a469f Author: Xiliang Chen <[email protected]> Date: Fri Apr 5 03:55:55 2019 +1300 code cleanup (#2206) commit 26c7b44 Author: Arkadiy Paronyan <[email protected]> Date: Wed Apr 3 15:52:46 2019 +0200 Emberic elm testnet (#2197)

* Introduction of PeersetHandle * integrate PeersetHandle with the rest of the codebase * fix compilation errors * more tests for peerset, fixed overwriting bug in add_reserved_peer * Slots data structure and bugfixes for peerset * bend to pressure * updated lru-cache to 0.1.2 and updated linked-hash-map to 0.5.2 * peerset discovered list is now a LinkedHashMap * fix review suggestions * split back Peerset and PeersetHandle * test for Peerset::discovered * applied review suggestions * fixes to peerset::incoming * peerset disconnects are all instantaneous * instantaneous drop in peerset finished * Peerset::set_reserved_only can also reconnect nodes * Peerset scores cache uses lru-cache * remove redundant function call and comment from Peerset::on_set_reserved_only * add_peer returns SlotState enum * apply review suggestions * is_reserved -> is_connected_and_reserved

* simplification of peerset api (#2123) * Introduction of PeersetHandle * integrate PeersetHandle with the rest of the codebase * fix compilation errors * more tests for peerset, fixed overwriting bug in add_reserved_peer * Slots data structure and bugfixes for peerset * bend to pressure * updated lru-cache to 0.1.2 and updated linked-hash-map to 0.5.2 * peerset discovered list is now a LinkedHashMap * fix review suggestions * split back Peerset and PeersetHandle * test for Peerset::discovered * applied review suggestions * fixes to peerset::incoming * peerset disconnects are all instantaneous * instantaneous drop in peerset finished * Peerset::set_reserved_only can also reconnect nodes * Peerset scores cache uses lru-cache * remove redundant function call and comment from Peerset::on_set_reserved_only * add_peer returns SlotState enum * apply review suggestions * is_reserved -> is_connected_and_reserved * Fix state inconsistency between handler and behaviour (#2220) * Fix state inconsistency between handler and behaviour * Fix the error! being in the wrong place * Apply negative rating on explicit ban (#2247) * Apply negative rating on explicit ban * Update core/network/src/service.rs Co-Authored-By: arkpar <[email protected]>

* Send high-level consensus telemetry by default * Notify telemetry on finalized * Send used authority set to telemetry * Do not send commit message telemetry by default * Fix typo * Allow for notifications on telemetry connect ...and send the current authority set on each connect. * Send authority set to telemetry on change * Revert "Send used authority set to telemetry" This reverts commit 1deceea. * Merge branch 'master' into 'cmichi-send-high-level-consensus-telemetry-by-default' Squashed commit of the following: commit 6de583a Author: Xiliang Chen <[email protected]> Date: Wed Apr 10 20:26:29 2019 +1200 update authers for rest of the node-template cargo.toml files (#2242) commit 5240bc1 Author: Bastian Köcher <[email protected]> Date: Tue Apr 9 10:31:18 2019 +0200 Throw a compile error for `on_finalise` and `on_initialise` (#2236) commit 67d2e71 Author: Pierre Krieger <[email protected]> Date: Tue Apr 9 05:30:43 2019 -0300 Add warning when using default protocol ID (#2234) * Add warning when using default protocol ID * Update core/service/src/lib.rs commit 1421fed Author: Xiliang Chen <[email protected]> Date: Tue Apr 9 17:22:20 2019 +1200 update name and authors to placeholder text for node-template (#2222) * update name and authors to placeholder text * revert package name change commit 6617f23 Author: André Silva <[email protected]> Date: Mon Apr 8 12:50:34 2019 +0100 grandpa: Voter persistence and upgrade to finality-grandpa v0.7 (#2139) * core: grandpa: migrate to grandpa 0.7 * core: grandpa: store current round votes and load them on startup * core: grandpa: resend old persisted votes for the current round * core: grandpa: store base and votes for last completed round * core: grandpa: fix latest grandpa 0.7 changes * core: grandpa: update to grandpa 0.7.1 * core: grandpa: persist votes for last two completed rounds * core: grandpa: simplify VoterSetState usage * core: grandpa: use Environment::update_voter_set_state * core: grandpa: fix aux_schema test * core: grandpa: add docs * core: grandpa: add note about environment assumption * core: grandpa: don't update voter set state on ignored votes * core: grandpa: add test for v1 -> v2 aux_schema migration * core: grandpa: add test for voter vote persistence * core: grandpa: use grandpa 0.7.1 from crates.io * core: grandpa: use try_init in test * core: grandpa: add comment about block_import in test * core: grandpa: avoid cloning HasVoted * core: grandpa: add missing docs * core: grandpa: cleanup up can_propose/prevote/precommit commit 21e0877 Author: Gregory Terzian <[email protected]> Date: Mon Apr 8 13:17:00 2019 +0200 remove clone bound on specialization in testnet factory (#2157) commit 7c64746 Author: Andrew Jones <[email protected]> Date: Sat Apr 6 12:23:56 2019 +0100 Contract import/export validation (#2203) * Reject validation of contract with unknown exports * Validate imports eagerly * Increment spec version commit 12718fa Author: Pierre Krieger <[email protected]> Date: Fri Apr 5 14:07:09 2019 -0300 Fix state inconsistency between handler and behaviour (#2220) * Fix state inconsistency between handler and behaviour * Fix the error! being in the wrong place commit f917d12 Author: Bastian Köcher <[email protected]> Date: Fri Apr 5 18:50:38 2019 +0200 Use `storage_root` of newly calculated header (#2216) Instead of calculating the `storage_root` a second time, we just can take the `storage_root` from the new header. commit 3359ce0 Author: Marek Kotewicz <[email protected]> Date: Fri Apr 5 14:44:46 2019 +0200 Peerset::discovered accepts many peer ids (#2213) * Peerset::discovered accepts many peer ids * Improve tracing in peerset commit dd82e0e Author: Marek Kotewicz <[email protected]> Date: Thu Apr 4 19:40:40 2019 +0200 simplification of peerset api (#2123) * Introduction of PeersetHandle * integrate PeersetHandle with the rest of the codebase * fix compilation errors * more tests for peerset, fixed overwriting bug in add_reserved_peer * Slots data structure and bugfixes for peerset * bend to pressure * updated lru-cache to 0.1.2 and updated linked-hash-map to 0.5.2 * peerset discovered list is now a LinkedHashMap * fix review suggestions * split back Peerset and PeersetHandle * test for Peerset::discovered * applied review suggestions * fixes to peerset::incoming * peerset disconnects are all instantaneous * instantaneous drop in peerset finished * Peerset::set_reserved_only can also reconnect nodes * Peerset scores cache uses lru-cache * remove redundant function call and comment from Peerset::on_set_reserved_only * add_peer returns SlotState enum * apply review suggestions * is_reserved -> is_connected_and_reserved commit d90833d Author: Arkadiy Paronyan <[email protected]> Date: Thu Apr 4 18:01:28 2019 +0200 Disconnect on protocol timeout (#2212) commit c0a46b5 Author: André Silva <[email protected]> Date: Thu Apr 4 15:56:49 2019 +0100 core: grandpa: verify commit target in justification (#2201) commit 3a4901a Author: Bastian Köcher <[email protected]> Date: Thu Apr 4 16:56:16 2019 +0200 Introduce `original_storage` and `original_storage_hash` (#2211) Both functions will ignore any overlayed changes and access the backend directly. commit a7a469f Author: Xiliang Chen <[email protected]> Date: Fri Apr 5 03:55:55 2019 +1300 code cleanup (#2206) commit 26c7b44 Author: Arkadiy Paronyan <[email protected]> Date: Wed Apr 3 15:52:46 2019 +0200 Emberic elm testnet (#2197) * Make telemetry onconnect hoook optional * Merge branch 'master' into 'cmichi-send-high-level-consensus-telemetry-by-default' * Introduce GrandpaParams struct to condense parameters * Remove debug statement * Fix tests * Rename parameter * Fix tests * Rename struct * Do not send verbosity level * Combine imports * Implement comments * Run cargo build --all * Remove noisy telemetry * Add docs for public items * Unbox and support Clone trait * Fix merge * Fix merge * Update core/finality-grandpa/src/lib.rs Co-Authored-By: cmichi <[email protected]>

* Send high-level consensus telemetry by default * Notify telemetry on finalized * Send used authority set to telemetry * Do not send commit message telemetry by default * Fix typo * Allow for notifications on telemetry connect ...and send the current authority set on each connect. * Send authority set to telemetry on change * Revert "Send used authority set to telemetry" This reverts commit 1deceea. * Merge branch 'master' into 'cmichi-send-high-level-consensus-telemetry-by-default' Squashed commit of the following: commit 6de583a Author: Xiliang Chen <[email protected]> Date: Wed Apr 10 20:26:29 2019 +1200 update authers for rest of the node-template cargo.toml files (paritytech#2242) commit 5240bc1 Author: Bastian Köcher <[email protected]> Date: Tue Apr 9 10:31:18 2019 +0200 Throw a compile error for `on_finalise` and `on_initialise` (paritytech#2236) commit 67d2e71 Author: Pierre Krieger <[email protected]> Date: Tue Apr 9 05:30:43 2019 -0300 Add warning when using default protocol ID (paritytech#2234) * Add warning when using default protocol ID * Update core/service/src/lib.rs commit 1421fed Author: Xiliang Chen <[email protected]> Date: Tue Apr 9 17:22:20 2019 +1200 update name and authors to placeholder text for node-template (paritytech#2222) * update name and authors to placeholder text * revert package name change commit 6617f23 Author: André Silva <[email protected]> Date: Mon Apr 8 12:50:34 2019 +0100 grandpa: Voter persistence and upgrade to finality-grandpa v0.7 (paritytech#2139) * core: grandpa: migrate to grandpa 0.7 * core: grandpa: store current round votes and load them on startup * core: grandpa: resend old persisted votes for the current round * core: grandpa: store base and votes for last completed round * core: grandpa: fix latest grandpa 0.7 changes * core: grandpa: update to grandpa 0.7.1 * core: grandpa: persist votes for last two completed rounds * core: grandpa: simplify VoterSetState usage * core: grandpa: use Environment::update_voter_set_state * core: grandpa: fix aux_schema test * core: grandpa: add docs * core: grandpa: add note about environment assumption * core: grandpa: don't update voter set state on ignored votes * core: grandpa: add test for v1 -> v2 aux_schema migration * core: grandpa: add test for voter vote persistence * core: grandpa: use grandpa 0.7.1 from crates.io * core: grandpa: use try_init in test * core: grandpa: add comment about block_import in test * core: grandpa: avoid cloning HasVoted * core: grandpa: add missing docs * core: grandpa: cleanup up can_propose/prevote/precommit commit 21e0877 Author: Gregory Terzian <[email protected]> Date: Mon Apr 8 13:17:00 2019 +0200 remove clone bound on specialization in testnet factory (paritytech#2157) commit 7c64746 Author: Andrew Jones <[email protected]> Date: Sat Apr 6 12:23:56 2019 +0100 Contract import/export validation (paritytech#2203) * Reject validation of contract with unknown exports * Validate imports eagerly * Increment spec version commit 12718fa Author: Pierre Krieger <[email protected]> Date: Fri Apr 5 14:07:09 2019 -0300 Fix state inconsistency between handler and behaviour (paritytech#2220) * Fix state inconsistency between handler and behaviour * Fix the error! being in the wrong place commit f917d12 Author: Bastian Köcher <[email protected]> Date: Fri Apr 5 18:50:38 2019 +0200 Use `storage_root` of newly calculated header (paritytech#2216) Instead of calculating the `storage_root` a second time, we just can take the `storage_root` from the new header. commit 3359ce0 Author: Marek Kotewicz <[email protected]> Date: Fri Apr 5 14:44:46 2019 +0200 Peerset::discovered accepts many peer ids (paritytech#2213) * Peerset::discovered accepts many peer ids * Improve tracing in peerset commit dd82e0e Author: Marek Kotewicz <[email protected]> Date: Thu Apr 4 19:40:40 2019 +0200 simplification of peerset api (paritytech#2123) * Introduction of PeersetHandle * integrate PeersetHandle with the rest of the codebase * fix compilation errors * more tests for peerset, fixed overwriting bug in add_reserved_peer * Slots data structure and bugfixes for peerset * bend to pressure * updated lru-cache to 0.1.2 and updated linked-hash-map to 0.5.2 * peerset discovered list is now a LinkedHashMap * fix review suggestions * split back Peerset and PeersetHandle * test for Peerset::discovered * applied review suggestions * fixes to peerset::incoming * peerset disconnects are all instantaneous * instantaneous drop in peerset finished * Peerset::set_reserved_only can also reconnect nodes * Peerset scores cache uses lru-cache * remove redundant function call and comment from Peerset::on_set_reserved_only * add_peer returns SlotState enum * apply review suggestions * is_reserved -> is_connected_and_reserved commit d90833d Author: Arkadiy Paronyan <[email protected]> Date: Thu Apr 4 18:01:28 2019 +0200 Disconnect on protocol timeout (paritytech#2212) commit c0a46b5 Author: André Silva <[email protected]> Date: Thu Apr 4 15:56:49 2019 +0100 core: grandpa: verify commit target in justification (paritytech#2201) commit 3a4901a Author: Bastian Köcher <[email protected]> Date: Thu Apr 4 16:56:16 2019 +0200 Introduce `original_storage` and `original_storage_hash` (paritytech#2211) Both functions will ignore any overlayed changes and access the backend directly. commit a7a469f Author: Xiliang Chen <[email protected]> Date: Fri Apr 5 03:55:55 2019 +1300 code cleanup (paritytech#2206) commit 26c7b44 Author: Arkadiy Paronyan <[email protected]> Date: Wed Apr 3 15:52:46 2019 +0200 Emberic elm testnet (paritytech#2197) * Make telemetry onconnect hoook optional * Merge branch 'master' into 'cmichi-send-high-level-consensus-telemetry-by-default' * Introduce GrandpaParams struct to condense parameters * Remove debug statement * Fix tests * Rename parameter * Fix tests * Rename struct * Do not send verbosity level * Combine imports * Implement comments * Run cargo build --all * Remove noisy telemetry * Add docs for public items * Unbox and support Clone trait * Fix merge * Fix merge * Update core/finality-grandpa/src/lib.rs Co-Authored-By: cmichi <[email protected]>

debris added 3 commits March 26, 2019 18:45

Introduction of PeersetHandle

c86b760

integrate PeersetHandle with the rest of the codebase

97430fb

Merge branch 'master' into peerset-changes

12788ca

debris added A0-please_review Pull request needs code review. M4-core labels Mar 26, 2019

debris requested a review from tomaka March 26, 2019 21:03

fix compilation errors

939cc30

gterzian reviewed Mar 27, 2019

View reviewed changes

tomusdrw approved these changes Mar 28, 2019

View reviewed changes

Merge branch 'master' into peerset-changes

284fc7e

bkchr reviewed Mar 28, 2019

View reviewed changes

more tests for peerset, fixed overwriting bug in add_reserved_peer

5a04f53

debris force-pushed the peerset-changes branch from 71f5e90 to 5a04f53 Compare March 28, 2019 12:00

debris commented Mar 28, 2019

View reviewed changes

tomaka suggested changes Mar 28, 2019

View reviewed changes

gavofyork added this to the 1.0 milestone Mar 29, 2019

gavofyork added A0-please_review Pull request needs code review. and removed A0-please_review Pull request needs code review. labels Mar 29, 2019

gavofyork modified the milestones: 1.0, 1.1 Mar 29, 2019

debris added 4 commits March 29, 2019 14:44

Merge branch 'master' into peerset-changes

5037c8e

Merge branch 'master' into peerset-changes

119bff4

Slots data structure and bugfixes for peerset

141284e

bend to pressure

e849837

gavofyork reviewed Apr 2, 2019

View reviewed changes

gavofyork approved these changes Apr 2, 2019

View reviewed changes

debris added 3 commits April 2, 2019 11:04

Merge branch 'master' into peerset-changes

9ecb800

updated lru-cache to 0.1.2 and updated linked-hash-map to 0.5.2

a74b69c

peerset discovered list is now a LinkedHashMap

94dc0f7

peerset disconnects are all instantaneous

6bc7e14

debris force-pushed the peerset-changes branch from 729f1a7 to 6bc7e14 Compare April 4, 2019 09:16

instantaneous drop in peerset finished

a92825f

Peerset::set_reserved_only can also reconnect nodes

ce5b4b2

tomaka suggested changes Apr 4, 2019

View reviewed changes

debris added 5 commits April 4, 2019 15:45

Peerset scores cache uses lru-cache

a1fb988

remove redundant function call and comment from Peerset::on_set_reser…

8b20252

…ved_only

add_peer returns SlotState enum

310cef2

apply review suggestions

3e2340d

is_reserved -> is_connected_and_reserved

de2525b

debris requested a review from tomaka April 4, 2019 14:37

tomaka approved these changes Apr 4, 2019

View reviewed changes

tomaka added A8-looksgood and removed A5-grumble labels Apr 4, 2019

Merge branch 'master' into peerset-changes

494cca9

debris merged commit dd82e0e into master Apr 4, 2019

debris deleted the peerset-changes branch April 4, 2019 17:40

	/// Reports an adjustement to the reputation of the given peer.
	/// Reports an adjustment to the reputation of the given peer.

simplification of peerset api #2123

simplification of peerset api #2123

Uh oh!

Conversation

debris commented Mar 26, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

gterzian Mar 28, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

gterzian Mar 28, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

tomusdrw left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

tomaka Mar 28, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

debris commented Apr 4, 2019

Uh oh!

arkpar commented Apr 4, 2019

Uh oh!

debris commented Mar 26, 2019 •

edited

Loading

gterzian Mar 28, 2019 •

edited

Loading

gterzian Mar 28, 2019 •

edited

Loading

tomaka Mar 28, 2019 •

edited

Loading