Incremental decommits off-chain #1223

v0d1ch · 2023-12-22T15:10:34Z

This PR introduces changes related to off-chain decommit protocol logic.
Now user has two options to request a decommit of some UTxO from the open Head:

Issue a http request to /decommit endpoint
Issue a client input Decommit

Decommit happens in sequence and you can't ask for a decommit while another one is in flight.

It is the user's responsibility to craft a decommit transaction which should be applicable to the local ledger state on L2.

If the transaction applies then the ReqDec network effect is issued and the next Snapshot that is to be signed specifies
which UTxO is to be decommitted from the Head.

On next AckSn all nodes need to agree/sign the snapshot that excludes the decommit UTxO from the ledger state.

Once everybody signs the snapshot decommit tx is issued and upon observing it on L1 it is removed from the local state and another decommit can be processed.

CHANGELOG updated or not needed
Documentation updated or not needed
Haddocks updated or not needed
No new TODOs introduced or explained herafter

ffakenz · 2024-01-02T12:59:44Z

hydra-cardano-api/src/Cardano/Api/UTxO.hs

+
+-- | Infix version of 'difference'.
+(\\) :: UTxO' out -> UTxO' out -> UTxO' out
+a \\ b = difference a b


this is not used.

feels weird to have to define this function twice.

i would keep only the infix version and define difference in the where clause.

wdyt?

We thought infix version might come in handy. We can certainly remove it if you want.

hydra-cluster/src/Hydra/Cluster/Scenarios.hs

hydra-node/src/Hydra/API/ServerOutput.hs

hydra-node/src/Hydra/Chain/Direct/Handlers.hs

hydra-plutus/src/Hydra/Contract/Head.hs

hydra-node/test/Hydra/Chain/Direct/Contract/Close.hs

ffakenz · 2024-01-02T15:12:54Z

hydra-node/test/Hydra/BehaviorSpec.hs

@@ -377,6 +377,42 @@ spec = parallel $ do

                waitUntil [n1] $ GetUTxOResponse testHeadId (utxoRefs [2, 42])

+      it "can request decommit" $


i think this test is redundant as it is being covered by the below.

wdyt?

I think so too so we could remove it.

hydra-node/src/Hydra/Snapshot.hs

hydra-node/test/Hydra/HeadLogicSnapshotSpec.hs

hydra-node/test/Hydra/NodeSpec.hs

ghost

I only did a surface review and did not dive deeply in the decommit logic. It overall makes sense to me but take my comments with a grain of salt.

ghost · 2024-01-12T08:33:32Z

hydra-node/test/Hydra/HeadLogicSnapshotSpec.hs

@@ -66,7 +67,7 @@ spec = do
                  NetworkEffect ReqSn{} -> True
                  _ -> False
              )
-    let snapshot1 = Snapshot testHeadId 1 mempty []
+    let snapshot1 = Snapshot testHeadId 1 mempty [] mempty


Seems like we want a smart constructor here :)

ghost · 2024-01-12T08:34:01Z

hydra-node/test/Hydra/HeadLogicSnapshotSpec.hs

@@ -121,7 +122,7 @@ spec = do

      it "sends ReqSn  when leader and there are seen transactions" $ do
        headState <- runEvents bobEnv simpleLedger (inOpenState threeParties simpleLedger) $ do
-          step (NetworkEvent defaultTTL alice $ ReqSn 1 [])
+          step (NetworkEvent defaultTTL alice $ ReqSn 1 [] Nothing)


and another smart constructor there I think?

hydra-node/test/Hydra/HeadLogicSpec.hs

hydra-node/src/Hydra/API/ServerOutput.hs

hydra-node/src/Hydra/HeadLogic.hs

github-actions · 2024-01-17T18:57:27Z

Transactions Costs

Sizes and execution budgets for Hydra protocol transactions. Note that unlisted parameters are currently using arbitrary values and results are not fully deterministic and comparable to previous runs.

Metadata
Generated at	2024-01-24 12:01:36.355624796 UTC
Max. memory units	14000000
Max. CPU units	10000000000
Max. tx size (kB)	16384

Script summary

Name	Hash	Size (Bytes)
νInitial	985245919fcc6c0c5cd116023cd2c947c43e80dcbb5075fe12433fbb	4072
νCommit	7cb20fa71eb4c563ca283566ebe0aa65859d96c3f8cba35c52c181fd	2043
νHead	c23d4bf0cf9f3bc8c4461ca49bcf2ef46e5fb31d6f13cb2bfba5d207	8815
μHead	c1491c68117a7fa173c91f39ed21e7aa4c979acb3194cec5f308b244*	3851

The minting policy hash is only usable for comparison. As the script is parameterized, the actual script is unique per Head.

Cost of Init Transaction

Parties	Tx size	% max Mem	% max CPU	Min fee ₳
1	4378	10.40	4.01	0.46
2	4576	12.68	4.88	0.49
3	4776	14.94	5.74	0.52
5	5179	19.12	7.31	0.59
10	6182	30.43	11.61	0.75
41	12418	99.02	37.59	1.77

Cost of Commit Transaction

This is using ada-only outputs for better comparability.

UTxO	Tx size	% max Mem	% max CPU	Min fee ₳
1	531	11.37	4.44	0.30
2	721	15.04	6.07	0.35
3	912	18.85	7.75	0.40
5	1285	26.90	11.27	0.51
10	2211	49.55	20.97	0.80
19	3905	99.43	41.75	1.43

Cost of CollectCom Transaction

Parties	UTxO (bytes)	Tx size	% max Mem	% max CPU	Min fee ₳
1	57	480	21.80	8.52	0.41
2	113	590	32.24	12.75	0.53
3	169	700	44.56	17.78	0.67
4	225	814	60.25	24.15	0.85
5	283	920	78.13	31.46	1.05
6	339	1031	93.32	37.86	1.22

Cost of Close Transaction

Parties	Tx size	% max Mem	% max CPU	Min fee ₳
1	568	12.75	5.57	0.32
2	671	13.57	5.94	0.33
3	930	15.34	7.10	0.37
5	1295	17.86	8.59	0.41
10	1520	24.17	12.23	0.50
50	8811	71.09	38.60	1.39

Cost of Contest Transaction

Parties	Tx size	% max Mem	% max CPU	Min fee ₳

Cost of Abort Transaction

Some variation because of random mixture of still initial and already committed outputs.

Parties	Tx size	% max Mem	% max CPU	Min fee ₳
1	4307	18.86	8.11	0.55
2	4395	27.08	11.57	0.65
3	4686	47.94	21.03	0.90
4	4732	59.68	26.07	1.03
5	5018	86.92	38.38	1.35

Cost of FanOut Transaction

Involves spending head output and burning head tokens. Uses ada-only UTxO for better comparability.

Parties	UTxO	UTxO (bytes)	Tx size	% max Mem	% max CPU	Min fee ₳
5	0	0	4206	8.27	3.46	0.43
5	1	57	4239	9.37	4.17	0.44
5	5	285	4376	14.20	7.18	0.51
5	10	567	4542	20.14	10.91	0.59
5	20	1139	4884	32.65	18.62	0.76
5	30	1710	5227	44.95	26.25	0.93
5	40	2275	5562	57.34	33.92	1.10
5	50	2847	5904	69.78	41.61	1.27
5	74	4212	6717	98.93	59.78	1.68

End-To-End Benchmark Results

This page is intended to collect the latest end-to-end benchmarks results produced by Hydra's Continuous Integration system from the latest master code.

Please take those results with a grain of salt as they are currently produced from very limited cloud VMs and not controlled hardware. Instead of focusing on the absolute results, the emphasis should be on relative results, eg. how the timings for a scenario evolve as the code changes.

Generated at 2024-01-24 12:04:18.035711122 UTC

Baseline Scenario

Number of nodes	3
Number of txs	9000
Avg. Confirmation Time (ms)	24.330726389
P99	117.79472145000008ms
P95	33.5762009ms
P50	21.3332945ms
Number of Invalid txs	0

Baseline Scenario

Number of nodes	1
Number of txs	3000
Avg. Confirmation Time (ms)	4.325800727
P99	7.135714899999987ms
P95	5.190308049999995ms
P50	4.1685385ms
Number of Invalid txs	0

github-actions · 2024-01-19T17:41:21Z

Test Results

421 tests 414 ✅ 17m 2s ⏱️
138 suites 7 💤
5 files 0 ❌

Results for commit 73d5576.

♻️ This comment has been updated with latest results.

Remove some fixmes but also add one more in HeadLogic to remind us to change the response to tx instead of UTxO.

Once we figure out exact type we want to return we will resolve FIXME in the api schema

Add HeadParameters to DecrementTx Stub out the decrementTx with appropriate params Rename maybeEmitDecommitApproved to maybeEmitDecrementTx

decrementTx is missing the head redeemer for decrement so this is moving into on-chain territory a bit. TODO: Add Decrement redeemer TODO: How to construct the correct UTxO hash Only emit DecrementTx if leader

Construct the Decrement redeemer using provided signatures and utxoHash using the UTxO from confirmed Snapshot.

We want better observability when something goes wrong

It makes sense to have one type for all possible invalid outcomes.

We don't need to keep it in DecrementTx since we already have it as part of a Snapshot

We only need to emit ReqSn if we are the leader, other events should be published

When we receive ReqDec it doesn't make sense to see DecommitAlreadyInFlight if we didn't request the decommit in the first place.

Since we hash the utxo to decommit separately we also need another argument to the verification function. Also rename Decrement redeemer field to match the others

ch1bo

I was VERY close to approve this. Only the head logic was slightly off. It's a must change as I think we introduce these changes with this PR and make it behave similar to NewTx / TxInvalid behavior (+ avoid boolean blindness tech debt right away).

hydra-node/src/Hydra/API/ServerOutput.hs

hydra-node/json-schemas/api.yaml

hydra-node/src/Hydra/HeadLogic.hs

On ReqDec if there is a decommit tx present in state just issue RequirementFailure DecommitTxInFlight

hydra-node/src/Hydra/HeadLogic.hs

hydra-node/src/Hydra/Snapshot.hs

hydra-node/src/Hydra/Chain/Direct/Tx.hs

v0d1ch self-assigned this Dec 22, 2023

ch1bo self-requested a review January 2, 2024 10:42

ch1bo self-assigned this Jan 2, 2024

ffakenz reviewed Jan 2, 2024

View reviewed changes

ghost approved these changes Jan 12, 2024

View reviewed changes

v0d1ch force-pushed the incremental-decommits-off-chain branch 2 times, most recently from 6a551d2 to bda2115 Compare January 12, 2024 21:15

v0d1ch force-pushed the feature/incremental-decommit branch from c00dd75 to f04183a Compare January 14, 2024 19:32

v0d1ch force-pushed the incremental-decommits-off-chain branch 2 times, most recently from 97b2a69 to 815c2f1 Compare January 14, 2024 22:12

ghost mentioned this pull request Jan 16, 2024

Incremental decommit #1057

Closed

16 tasks

ch1bo removed their assignment Jan 17, 2024

v0d1ch force-pushed the incremental-decommits-off-chain branch from 43a575f to eefba04 Compare January 17, 2024 18:55

v0d1ch force-pushed the feature/incremental-decommit branch from f04183a to 4381adb Compare January 17, 2024 19:21

v0d1ch force-pushed the incremental-decommits-off-chain branch from eefba04 to 1051875 Compare January 17, 2024 19:22

v0d1ch force-pushed the feature/incremental-decommit branch from 4381adb to b873b97 Compare January 18, 2024 20:45

v0d1ch force-pushed the incremental-decommits-off-chain branch from ce1ce0a to 54d84a8 Compare January 18, 2024 20:46

v0d1ch force-pushed the incremental-decommits-off-chain branch 2 times, most recently from 834735d to 14d950d Compare January 22, 2024 08:03

v0d1ch force-pushed the feature/incremental-decommit branch from b873b97 to cc4aba5 Compare January 22, 2024 08:40

ch1bo and others added 8 commits January 22, 2024 09:41

Draft an end-to-end happy path scenario for decommit

a4c7c70

Use http call in decommit test

60ee172

Remove some fixmes but also add one more in HeadLogic to remind us to change the response to tx instead of UTxO.

Tackle the api schema and ServerOutput encoding

adc8e45

Once we figure out exact type we want to return we will resolve FIXME in the api schema

Implement decrement function

e32dd9b

Add HeadParameters to DecrementTx Stub out the decrementTx with appropriate params Rename maybeEmitDecommitApproved to maybeEmitDecrementTx

Add snapshot UTxO to DecrementTx and mostly implement decrementTx

1d7c593

decrementTx is missing the head redeemer for decrement so this is moving into on-chain territory a bit. TODO: Add Decrement redeemer TODO: How to construct the correct UTxO hash Only emit DecrementTx if leader

Add snapshot and signatures to DecrementTx

34ef93e

Construct the Decrement redeemer using provided signatures and utxoHash using the UTxO from confirmed Snapshot.

Fix the api.yaml schema and golden files

5275c04

Use empty ByteString in case decommit utxo is not there

90dd5c0

v0d1ch added 2 commits January 23, 2024 16:19

Add utxo and validationError to DecommitTxInvalid

73d5576

We want better observability when something goes wrong

Introduce DecommitInvalid to hold invalid options

fb69705

It makes sense to have one type for all possible invalid outcomes.

v0d1ch force-pushed the incremental-decommits-off-chain branch from 643c00b to fb69705 Compare January 23, 2024 17:04

v0d1ch added 6 commits January 23, 2024 18:33

Reuse utxoToDecommit from the Snapshot

56b5915

We don't need to keep it in DecrementTx since we already have it as part of a Snapshot

Do not require networkId in decrementTx

5b57fc1

Remove decommit utxo check and add a TODO about it

e0d5d4a

Check if leader before outputting ReqDec

c532d02

Revert signable rep for Snapshot

23ff08a

Remove DecomitRequestReceived

0dc0e7b

v0d1ch force-pushed the incremental-decommits-off-chain branch from 56e6d40 to 0dc0e7b Compare January 23, 2024 19:31

v0d1ch added 7 commits January 23, 2024 20:36

Remove orphans ignoring and some minor test changes

fb99a1e

Output DecommitRecorded and DecommitRequested on each ReqDec

f8d74c7

We only need to emit ReqSn if we are the leader, other events should be published

Formatting changes

7131193

Reorder HeadLogic events to natural flow of decommit

8b7069c

Only emit DecommitAlreadyInFlight if we requested it

6b1d55a

When we receive ReqDec it doesn't make sense to see DecommitAlreadyInFlight if we didn't request the decommit in the first place.

Fix the snapshot verification

0070b7e

Since we hash the utxo to decommit separately we also need another argument to the verification function. Also rename Decrement redeemer field to match the others

Simplify e2e test a bit

5a8b026

v0d1ch force-pushed the incremental-decommits-off-chain branch from fc080ff to 7dba1f8 Compare January 24, 2024 12:06

ch1bo requested changes Jan 24, 2024

View reviewed changes

v0d1ch added 3 commits January 24, 2024 15:29

Fix plutus golden tests

10772f1

Inline requireNoDecommitInFlight

aff14fb

On ReqDec if there is a decommit tx present in state just issue RequirementFailure DecommitTxInFlight

Rename DecommitProcessed to DecommitFinalized

32c8fb5

v0d1ch force-pushed the incremental-decommits-off-chain branch from a831ed1 to 32c8fb5 Compare January 24, 2024 14:29

v0d1ch added 3 commits January 24, 2024 15:36

Inline DecommitAlreadyInFlight and DecommitTxInvalid schemas

1f7d4cc

Don't reuse TransactionSubmitted for decommit

a5e22f9

Use less demanding constraints for DecommitInvalidReason

3d56b9f

ch1bo approved these changes Jan 25, 2024

View reviewed changes

hydra-node/src/Hydra/HeadLogic.hs Outdated Show resolved Hide resolved

hydra-node/src/Hydra/Snapshot.hs Outdated Show resolved Hide resolved

hydra-node/src/Hydra/Chain/Direct/Tx.hs Show resolved Hide resolved

Final changes

d2d9fe0

v0d1ch merged commit 8f502fc into feature/incremental-decommit Jan 26, 2024
17 of 20 checks passed

v0d1ch deleted the incremental-decommits-off-chain branch January 26, 2024 09:53

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Incremental decommits off-chain #1223

Incremental decommits off-chain #1223

v0d1ch commented Dec 22, 2023 •

edited

Loading

ffakenz Jan 2, 2024

v0d1ch Jan 14, 2024

ffakenz Jan 2, 2024

v0d1ch Jan 14, 2024

ghost left a comment

ghost Jan 12, 2024

ghost Jan 12, 2024

github-actions bot commented Jan 17, 2024 •

edited

Loading

github-actions bot commented Jan 19, 2024 •

edited

Loading

ch1bo left a comment

		@@ -377,6 +377,42 @@ spec = parallel $ do

		waitUntil [n1] $ GetUTxOResponse testHeadId (utxoRefs [2, 42])

		it "can request decommit" $

Incremental decommits off-chain #1223

Incremental decommits off-chain #1223

Conversation

v0d1ch commented Dec 22, 2023 • edited Loading

ffakenz Jan 2, 2024

Choose a reason for hiding this comment

v0d1ch Jan 14, 2024

Choose a reason for hiding this comment

ffakenz Jan 2, 2024

Choose a reason for hiding this comment

v0d1ch Jan 14, 2024

Choose a reason for hiding this comment

ghost left a comment

Choose a reason for hiding this comment

ghost Jan 12, 2024

Choose a reason for hiding this comment

ghost Jan 12, 2024

Choose a reason for hiding this comment

github-actions bot commented Jan 17, 2024 • edited Loading

Transactions Costs

Script summary

Cost of Init Transaction

Cost of Commit Transaction

Cost of CollectCom Transaction

Cost of Close Transaction

Cost of Contest Transaction

Cost of Abort Transaction

Cost of FanOut Transaction

End-To-End Benchmark Results

Baseline Scenario

Baseline Scenario

github-actions bot commented Jan 19, 2024 • edited Loading

Test Results

ch1bo left a comment

Choose a reason for hiding this comment

v0d1ch commented Dec 22, 2023 •

edited

Loading

github-actions bot commented Jan 17, 2024 •

edited

Loading

github-actions bot commented Jan 19, 2024 •

edited

Loading