feat: add priority queue for protocol messages to prevent signature loss under high load

[awcjack]

Summary

Under high transaction load, snapshot signatures were being lost because all messages shared a single FIFO queue. ReqSn/AckSn protocol messages got buried behind ReqTx transactions, causing signature collection to fail.

Solution

Dual-queue system that processes protocol messages before transactions:

• HighPriority: ReqSn, AckSn, ChainInput, ClientInput, ConnectivityEvent
• LowPriority: ReqTx, ReqDec

This ensures protocol state machine messages are never starved by transaction load.

Changes

• Added MessagePriority type and classifyPriority function in Hydra.HeadLogic.Input
• Updated InputQueue to use dual queues (high/low priority)
• High priority queue is always drained before processing low priority messages
• Updated tests to verify priority queue behavior

Testing

• All HeadLogic tests pass (69 tests)
• All Behavior tests pass (45 tests)
• InputQueue tests updated and pass

[github-actions]

Transaction 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	2026-01-06 00:13:17.641452035 UTC
Max. memory units	14000000
Max. CPU units	10000000000
Max. tx size (kB)	16384

Script summary

Name	Hash	Size (Bytes)
νInitial	c8a101a5c8ac4816b0dceb59ce31fc2258e387de828f02961d2f2045	2652
νCommit	61458bc2f297fff3cc5df6ac7ab57cefd87763b0b7bd722146a1035c	685
νHead	a1442faf26d4ec409e2f62a685c1d4893f8d6bcbaf7bcb59d6fa1340	14599
μHead	fd173b993e12103cd734ca6710d364e17120a5eb37a224c64ab2b188*	5284
νDeposit	ae01dade3a9c346d5c93ae3ce339412b90a0b8f83f94ec6baa24e30c	1102

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

Init transaction costs

Parties	Tx size	% max Mem	% max CPU	Min fee ₳
1	5840	10.57	3.36	0.52
2	6038	12.99	4.13	0.55
3	6238	14.71	4.65	0.58
5	6640	18.43	5.81	0.63
10	7647	29.31	9.25	0.79
43	14282	98.94	30.92	1.80

Commit transaction costs

This uses ada-only outputs for better comparability.

UTxO	Tx size	% max Mem	% max CPU	Min fee ₳
1	561	2.44	1.16	0.20
2	739	3.38	1.73	0.22
3	924	4.36	2.33	0.24
5	1282	6.41	3.60	0.28
10	2173	12.13	7.25	0.40
54	10048	98.61	68.52	1.88

CollectCom transaction costs

Parties	UTxO (bytes)	Tx size	% max Mem	% max CPU	Min fee ₳
1	57	525	25.24	7.32	0.43
2	114	636	34.23	9.85	0.53
3	170	747	42.25	12.15	0.61
4	226	858	49.83	14.41	0.69
5	280	969	61.05	17.49	0.81
6	340	1081	71.10	20.29	0.92
7	393	1192	85.98	24.25	1.07
8	451	1303	86.83	24.81	1.09
9	504	1414	94.26	27.21	1.17

Cost of Increment Transaction

Parties	Tx size	% max Mem	% max CPU	Min fee ₳
1	1807	24.37	7.71	0.48
2	1922	25.88	8.79	0.51
3	2132	28.47	10.18	0.55
5	2327	29.88	11.93	0.58
10	3230	43.14	18.95	0.78
39	7622	97.62	53.47	1.66

Cost of Decrement Transaction

Parties	Tx size	% max Mem	% max CPU	Min fee ₳
1	638	22.54	7.30	0.41
2	700	22.55	7.95	0.42
3	884	25.47	9.47	0.46
5	1193	29.97	12.05	0.53
10	2049	39.63	18.05	0.69
43	6783	98.65	56.48	1.65

Close transaction costs

Parties	Tx size	% max Mem	% max CPU	Min fee ₳
1	678	29.13	8.90	0.48
2	824	31.54	10.25	0.52
3	952	33.47	11.46	0.55
5	1201	36.35	13.57	0.59
10	2228	47.15	20.05	0.77
37	6000	97.81	52.18	1.58

Contest transaction costs

Parties	Tx size	% max Mem	% max CPU	Min fee ₳
1	700	33.87	10.16	0.53
2	804	35.88	11.39	0.56
3	896	37.20	12.40	0.58
5	1223	41.90	15.05	0.65
10	2136	55.43	22.23	0.85
29	4779	97.69	46.61	1.49

Abort transaction costs

There is some variation due to the random mixture of initial and already committed outputs.

Parties	Tx size	% max Mem	% max CPU	Min fee ₳
1	5789	27.16	9.11	0.69
2	5966	35.89	12.06	0.79
3	6084	44.77	15.03	0.89
4	6213	51.68	17.35	0.97
5	6303	58.48	19.61	1.04
6	6606	74.66	25.13	1.23
7	6837	84.86	28.60	1.34
8	6838	88.46	29.77	1.38

FanOut transaction costs

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 ₳
10	0	0	5835	19.63	6.56	0.61
10	5	285	6004	30.67	10.88	0.74
10	10	570	6175	39.69	14.52	0.85
10	20	1139	6513	59.98	22.53	1.08
10	30	1706	6852	79.15	30.16	1.31
10	39	2218	7157	98.05	37.58	1.53

End-to-end benchmark results

This page is intended to collect the latest end-to-end benchmark results produced by Hydra's continuous integration (CI) system from the latest master code.

Please note that these results are approximate as they are currently produced from limited cloud VMs and not controlled hardware. Rather than focusing on the absolute results, the emphasis should be on relative results, such as how the timings for a scenario evolve as the code changes.

Generated at 2026-01-06 00:16:42.165077015 UTC

Baseline Scenario

Number of nodes	1
Number of txs	300
Avg. Confirmation Time (ms)	5.473456956
P99	7.609304379999963ms
P95	6.68419725ms
P50	5.2490615ms
Number of Invalid txs	0

Three local nodes

Number of nodes	3
Number of txs	900
Avg. Confirmation Time (ms)	38.691192806
P99	70.48709549999998ms
P95	57.64521705ms
P50	35.9569545ms
Number of Invalid txs	0

[github-actions]

Transaction cost differences

No cost or size differences found