Fix PersistentQueue deadlock and add operational visibility by v0d1ch · Pull Request #2742 · cardano-scaling/hydra

v0d1ch · 2026-06-23T11:10:53Z

Fix FIXME deadlock: popPersistentQueue was using value equality to
decide whether to remove the head item. A false Eq result silently
no-oped, leaving the head item stuck. With the queue at capacity (100),
all callers of broadcast would block indefinitely with no log output.
Now pops unconditionally by index — safe because broadcastMessages is
the only consumer.
removeFile safety: catches isDoesNotExistError in pop so a missing
file (e.g. from a prior partial crash) doesn't kill the broadcast loop.
Operational visibility: emits PersistentQueueFull before blocking on
a full queue, and PersistentQueueLoadFailed instead of silently discarding
a startup load error.
Constraint cleanup: Eq msg dropped from withEtcdNetwork and
broadcastMessages.

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

github-actions · 2026-06-23T11:17:56Z

Transaction cost differences

No cost or size differences found

github-actions · 2026-06-23T11:20:47Z

End-to-end benchmark differences

Comparing this PR (new) against master (old). Numbers come from cloud VMs, so changes under 5% are shown as ≈ and are likely run-to-run noise rather than a real regression or improvement. 🟢 = improvement, 🔴 = regression.

Baseline Scenario

Metric	master	PR	Δ
End-to-end TPS (tx/s)	528.85	560.87	🟢 +32.02 (+6.1%)
Per-snapshot TPS P50 (tx/s)	2966.61	4066.45	🟢 +1099.84 (+37.1%)
Per-snapshot TPS P95 (tx/s)	7456.23	7377.20	≈ -79.03 (-1.1%)
Per-snapshot TPS max (tx/s)	7834.64	7707.04	≈ -127.60 (-1.6%)
Avg. Confirmation Time (ms)	546.80	516.70	🟢 -30.10 (-5.5%)
P50 confirmation (ms)	550.50	519.90	🟢 -30.60 (-5.6%)
P95 confirmation (ms)	560.70	530.90	🟢 -29.80 (-5.3%)
P99 confirmation (ms)	561.20	531.40	🟢 -29.80 (-5.3%)
Invalid txs	0.00	0.00	≈ +0.00 (n/a%)

Three local nodes

Metric	master	PR	Δ
End-to-end TPS (tx/s)	290.70	286.02	≈ -4.68 (-1.6%)
Per-snapshot TPS P50 (tx/s)	948.95	1011.08	🟢 +62.13 (+6.5%)
Per-snapshot TPS P95 (tx/s)	2949.35	2664.89	🔴 -284.46 (-9.6%)
Per-snapshot TPS max (tx/s)	3145.39	2817.21	🔴 -328.18 (-10.4%)
Avg. Confirmation Time (ms)	2804.80	2857.70	≈ +52.90 (+1.9%)
P50 confirmation (ms)	2859.90	2904.20	≈ +44.30 (+1.5%)
P95 confirmation (ms)	3075.90	3120.40	≈ +44.50 (+1.4%)
P99 confirmation (ms)	3084.00	3120.90	≈ +36.90 (+1.2%)
Invalid txs	0.00	0.00	≈ +0.00 (n/a%)

github-actions · 2026-06-23T11:35:01Z

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	2026-07-08 08:48:31.282149865 UTC
Max. memory units	14000000
Max. CPU units	10000000000
Max. tx size (kB)	16384

Script summary

Name	Hash	Size (Bytes)
νHead	fd75e24c9ea915ce8e48d3ff1d0c54ad09cc01191c24416ad7dba4a3	11621
μHead	83a964e973c065bbe70588f5e089817f92182ae81743e7a54cf3e29e*	4856
νDeposit	c78e8c9205721eb3ef4410f3db9c6169fa6db497c24641d29c20529c	1615
νCRS	09db7ee6cf7a4b358dd5c8a2f19d2c048336ffc5a01ef35a47ca7072	2736

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	5355	9.03	2.97	0.48
2	5447	9.31	3.05	0.49
3	5543	10.29	3.39	0.50
5	5737	10.87	3.55	0.52
10	6217	13.13	4.27	0.56
50	10058	34.71	11.02	0.95
100	14859	61.69	19.47	1.44
115	16296	69.67	21.97	1.59

Cost of Increment Transaction

Parties	Tx size	% max Mem	% max CPU	Min fee ₳
1	2278	17.76	6.40	0.44
2	2409	18.65	7.30	0.46
3	2540	20.24	8.43	0.49
5	2801	21.50	10.05	0.52
10	3456	26.01	14.56	0.62
50	8697	66.62	51.84	1.44
75	11975	91.37	74.90	1.96

Cost of Decrement Transaction

Parties	Tx size	% max Mem	% max CPU	Min fee ₳
1	603	16.03	5.75	0.35
2	734	16.94	6.65	0.37
3	860	17.83	7.55	0.39
5	1122	19.57	9.33	0.43
10	1778	24.17	13.86	0.52
50	7019	62.20	50.32	1.32
75	10296	86.26	73.18	1.83

`Close` transaction costs

Parties	Tx size	% max Mem	% max CPU	Min fee ₳
1	597	15.49	10.37	0.38
2	724	16.37	11.27	0.40
3	854	17.27	12.17	0.42
5	1117	19.16	14.00	0.46
10	1772	23.70	18.52	0.55
50	7009	61.72	55.07	1.35
75	10288	86.06	78.07	1.86

`Contest` transaction costs

Parties	Tx size	% max Mem	% max CPU	Min fee ₳
1	628	18.79	13.48	0.43
2	755	19.89	14.44	0.45
3	885	20.97	15.40	0.47
5	1149	23.10	17.30	0.51
10	1804	28.56	22.10	0.62
50	7044	74.22	60.90	1.50
72	9927	98.41	81.98	1.97

`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	5529	22.68	41.73	0.88
10	1	57	5563	25.03	44.25	0.92
10	5	284	5698	35.28	54.57	1.08
10	10	569	5869	49.29	67.85	1.30
10	20	1140	6209	81.82	95.83	1.78
10	20	1137	6206	81.82	95.83	1.78

`PartialFanOut` transaction costs

Largest chunk of ada-only outputs that can be distributed in one partial fanout step, computed dynamically. The last row is the maximum total UTxO count where at least one output can still be distributed.

Distributed	UTxO (bytes)	Tx size	% max Mem	% max CPU	Min fee ₳
11	569	986	34.31	65.19	0.94
25	1310	1429	67.64	98.26	1.47
30	1309	1428	67.64	98.26	1.47
40	1310	1429	67.64	98.26	1.47
50	1310	1425	67.64	98.26	1.47
100	1309	1428	67.64	98.26	1.47
150	1309	1428	67.64	98.26	1.47
200	1306	1425	67.64	98.26	1.47
200	1311	1430	67.64	98.26	1.47

`PartialFanOut` transaction costs (with native tokens)

Largest chunk of native-token outputs that can be distributed in one partial fanout step, computed dynamically. The last row is the maximum total UTxO count where at least one output can still be distributed.

Distributed	UTxO (bytes)	Tx size	% max Mem	% max CPU	Min fee ₳
11	1120	1609	41.40	67.74	1.04
25	2478	2742	75.99	98.08	1.59
30	2499	2764	75.97	98.12	1.59
40	2205	2456	75.99	98.03	1.58
50	2079	2321	75.97	97.98	1.57
100	2247	2501	75.99	98.03	1.58
150	2100	2347	75.99	97.98	1.57
200	2016	2259	75.97	97.97	1.57
200	2520	2787	75.99	98.13	1.59

`FinalPartialFanOut` transaction costs (with native tokens)

Terminal partial fanout step (FanoutProgress → Final) with outputs carrying a native token. Burns all head tokens and proves accumulator exhaustion via BLS proof.

Distributed	UTxO (bytes)	Tx size	% max Mem	% max CPU	Min fee ₳
1	112	5417	21.54	43.18	0.87
5	590	5811	35.15	54.72	1.08
10	1190	6305	53.23	69.51	1.36
10	990	6105	53.35	69.48	1.35

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 2026-07-08 08:39:27.775290624 UTC

Baseline Scenario

Number of nodes	1
Number of txs	300
Avg. Confirmation Time (ms)	514.4
P99	527.9ms
P95	527.6ms
P50	518.6ms
End-to-end TPS	563.06 tx/s
Snapshots observed	4
Per-snapshot TPS P50	2999.88 tx/s
Per-snapshot TPS P95	7686.48 tx/s
Per-snapshot TPS max	8072.98 tx/s
Number of Invalid txs	0
Fanout outputs	0

Three local nodes

Number of nodes	3
Number of txs	900
Avg. Confirmation Time (ms)	2618.9
P99	2897.8ms
P95	2880.0ms
P50	2665.1ms
End-to-end TPS	309.69 tx/s
Snapshots observed	10
Per-snapshot TPS P50	1081.86 tx/s
Per-snapshot TPS P95	2354.89 tx/s
Per-snapshot TPS max	2421.04 tx/s
Number of Invalid txs	0
Fanout outputs	0

Scenario benchmark results

This page collects results from the scenario matrix: every combination of cluster size, UTxO shape, and incremental-ops mode is exercised by CI from the latest master code and reported below.

Numbers are approximate. They come from cloud VMs rather than controlled hardware, so the useful signal is the relative change between cells and between commits, not the absolute throughput.

Generated at 2026-07-08 08:49:56.312987086 UTC

Summary across cells

TPS columns are rates (transactions per second); Wall clock (s) is the measured elapsed time from the first tx submission to the last confirmation. Times are rounded to one decimal.

Scenario	Txs	Wall clock (s)	End-to-end TPS (tx/s)	Per-snapshot p50 TPS (tx/s)	Avg conf (ms)	P95 conf (ms)
Nodes=1, Constant, incremental ops off, fire and forget	30	0.1	549.49	2505.87	53.8	54.3
Nodes=1, Constant, incremental ops off, wait for tx valid	30	0.2	171.16	179.17	5.8	7.6
Nodes=1, Growing, incremental ops off, fire and forget	30	0.1	412.60	1035.92	71.6	72.4
Nodes=1, Growing, incremental ops off, wait for tx valid	30	0.3	113.02	116.30	8.7	11.7
Nodes=1, Mixed, incremental ops off, fire and forget	30	0.1	514.32	2316.40	57.6	58.1
Nodes=1, Mixed, incremental ops off, wait for tx valid	30	0.2	140.81	141.73	7.0	8.8
Nodes=2, Constant, incremental ops off, fire and forget	60	0.2	394.44	1729.09	149.8	151.6
Nodes=2, Constant, incremental ops off, wait for tx valid	60	0.6	105.61	119.60	18.6	27.6
Nodes=2, Growing, incremental ops off, fire and forget	60	0.2	323.90	846.61	183.2	184.4
Nodes=2, Growing, incremental ops off, wait for tx valid	60	1.0	62.05	62.32	31.7	44.2
Nodes=2, Mixed, incremental ops off, fire and forget	60	0.2	336.33	1327.81	175.9	177.1
Nodes=2, Mixed, incremental ops off, wait for tx valid	60	0.8	79.06	80.32	25.0	34.6
Nodes=3, Constant, incremental ops off, fire and forget	90	0.3	332.04	1649.90	266.7	269.3
Nodes=3, Constant, incremental ops off, wait for tx valid	90	0.9	95.24	89.19	31.3	39.0
Nodes=3, Growing, incremental ops off, fire and forget	90	0.4	252.41	415.28	350.7	354.4
Nodes=3, Growing, incremental ops off, wait for tx valid	90	1.9	46.37	43.23	61.9	92.8
Nodes=3, Mixed, incremental ops off, fire and forget	90	0.3	310.03	1137.20	286.1	289.8
Nodes=3, Mixed, incremental ops off, wait for tx valid	90	1.5	61.65	55.93	47.5	65.5

Nodes=1, Constant, incremental ops off, fire and forget

Number of nodes	1
Number of txs	30
Avg. Confirmation Time (ms)	53.8
P99	54.4ms
P95	54.3ms
P50	54.0ms
End-to-end TPS	549.49 tx/s
Snapshots observed	2
Per-snapshot TPS P50	2505.87 tx/s
Per-snapshot TPS P95	4742.71 tx/s
Per-snapshot TPS max	4941.54 tx/s
Number of Invalid txs	0
Fanout outputs	0

Nodes=1, Constant, incremental ops off, wait for tx valid

Number of nodes	1
Number of txs	30
Avg. Confirmation Time (ms)	5.8
P99	7.8ms
P95	7.6ms
P50	5.5ms
End-to-end TPS	171.16 tx/s
Snapshots observed	30
Per-snapshot TPS P50	179.17 tx/s
Per-snapshot TPS P95	192.62 tx/s
Per-snapshot TPS max	193.75 tx/s
Number of Invalid txs	0
Fanout outputs	0

Nodes=1, Growing, incremental ops off, fire and forget

Number of nodes	1
Number of txs	30
Avg. Confirmation Time (ms)	71.6
P99	72.5ms
P95	72.4ms
P50	72.1ms
End-to-end TPS	412.60 tx/s
Snapshots observed	2
Per-snapshot TPS P50	1035.92 tx/s
Per-snapshot TPS P95	1952.87 tx/s
Per-snapshot TPS max	2034.38 tx/s
Number of Invalid txs	0
Fanout outputs	0

Nodes=1, Growing, incremental ops off, wait for tx valid

Number of nodes	1
Number of txs	30
Avg. Confirmation Time (ms)	8.7
P99	12.4ms
P95	11.7ms
P50	8.5ms
End-to-end TPS	113.02 tx/s
Snapshots observed	30
Per-snapshot TPS P50	116.30 tx/s
Per-snapshot TPS P95	160.44 tx/s
Per-snapshot TPS max	174.38 tx/s
Number of Invalid txs	0
Fanout outputs	0

Nodes=1, Mixed, incremental ops off, fire and forget

Each client first grows its UTxO set (1-in to 2-out) for half of its tx budget, then contracts it back (2-in to 1-out) for the remainder.

Number of nodes	1
Number of txs	30
Avg. Confirmation Time (ms)	57.6
P99	58.1ms
P95	58.1ms
P50	57.8ms
End-to-end TPS	514.32 tx/s
Snapshots observed	2
Per-snapshot TPS P50	2316.40 tx/s
Per-snapshot TPS P95	4383.86 tx/s
Per-snapshot TPS max	4567.63 tx/s
Number of Invalid txs	0
Fanout outputs	0

Nodes=1, Mixed, incremental ops off, wait for tx valid

Each client first grows its UTxO set (1-in to 2-out) for half of its tx budget, then contracts it back (2-in to 1-out) for the remainder.

Number of nodes	1
Number of txs	30
Avg. Confirmation Time (ms)	7.0
P99	11.3ms
P95	8.8ms
P50	7.0ms
End-to-end TPS	140.81 tx/s
Snapshots observed	30
Per-snapshot TPS P50	141.73 tx/s
Per-snapshot TPS P95	178.99 tx/s
Per-snapshot TPS max	187.83 tx/s
Number of Invalid txs	0
Fanout outputs	0

Nodes=2, Constant, incremental ops off, fire and forget

Number of nodes	2
Number of txs	60
Avg. Confirmation Time (ms)	149.8
P99	151.7ms
P95	151.6ms
P50	150.1ms
End-to-end TPS	394.44 tx/s
Snapshots observed	2
Per-snapshot TPS P50	1729.09 tx/s
Per-snapshot TPS P95	3278.57 tx/s
Per-snapshot TPS max	3416.30 tx/s
Number of Invalid txs	0
Fanout outputs	0

Nodes=2, Constant, incremental ops off, wait for tx valid

Number of nodes	2
Number of txs	60
Avg. Confirmation Time (ms)	18.6
P99	28.3ms
P95	27.6ms
P50	18.7ms
End-to-end TPS	105.61 tx/s
Snapshots observed	60
Per-snapshot TPS P50	119.60 tx/s
Per-snapshot TPS P95	141.94 tx/s
Per-snapshot TPS max	148.95 tx/s
Number of Invalid txs	0
Fanout outputs	0

Nodes=2, Growing, incremental ops off, fire and forget

Number of nodes	2
Number of txs	60
Avg. Confirmation Time (ms)	183.2
P99	184.5ms
P95	184.4ms
P50	183.8ms
End-to-end TPS	323.90 tx/s
Snapshots observed	2
Per-snapshot TPS P50	846.61 tx/s
Per-snapshot TPS P95	1602.53 tx/s
Per-snapshot TPS max	1669.73 tx/s
Number of Invalid txs	0
Fanout outputs	0

Nodes=2, Growing, incremental ops off, wait for tx valid

Number of nodes	2
Number of txs	60
Avg. Confirmation Time (ms)	31.7
P99	48.8ms
P95	44.2ms
P50	32.1ms
End-to-end TPS	62.05 tx/s
Snapshots observed	60
Per-snapshot TPS P50	62.32 tx/s
Per-snapshot TPS P95	108.42 tx/s
Per-snapshot TPS max	115.34 tx/s
Number of Invalid txs	0
Fanout outputs	0

Nodes=2, Mixed, incremental ops off, fire and forget

Each client first grows its UTxO set (1-in to 2-out) for half of its tx budget, then contracts it back (2-in to 1-out) for the remainder.

Number of nodes	2
Number of txs	60
Avg. Confirmation Time (ms)	175.9
P99	177.7ms
P95	177.1ms
P50	176.2ms
End-to-end TPS	336.33 tx/s
Snapshots observed	2
Per-snapshot TPS P50	1327.81 tx/s
Per-snapshot TPS P95	2516.99 tx/s
Per-snapshot TPS max	2622.70 tx/s
Number of Invalid txs	0
Fanout outputs	0

Nodes=2, Mixed, incremental ops off, wait for tx valid

Each client first grows its UTxO set (1-in to 2-out) for half of its tx budget, then contracts it back (2-in to 1-out) for the remainder.

Number of nodes	2
Number of txs	60
Avg. Confirmation Time (ms)	25.0
P99	36.2ms
P95	34.6ms
P50	25.6ms
End-to-end TPS	79.06 tx/s
Snapshots observed	60
Per-snapshot TPS P50	80.32 tx/s
Per-snapshot TPS P95	120.21 tx/s
Per-snapshot TPS max	138.94 tx/s
Number of Invalid txs	0
Fanout outputs	0

Nodes=3, Constant, incremental ops off, fire and forget

Number of nodes	3
Number of txs	90
Avg. Confirmation Time (ms)	266.7
P99	269.4ms
P95	269.3ms
P50	268.6ms
End-to-end TPS	332.04 tx/s
Snapshots observed	2
Per-snapshot TPS P50	1649.90 tx/s
Per-snapshot TPS P95	3131.09 tx/s
Per-snapshot TPS max	3262.75 tx/s
Number of Invalid txs	0
Fanout outputs	0

Nodes=3, Constant, incremental ops off, wait for tx valid

Number of nodes	3
Number of txs	90
Avg. Confirmation Time (ms)	31.3
P99	39.9ms
P95	39.0ms
P50	30.5ms
End-to-end TPS	95.24 tx/s
Snapshots observed	60
Per-snapshot TPS P50	89.19 tx/s
Per-snapshot TPS P95	173.00 tx/s
Per-snapshot TPS max	186.35 tx/s
Number of Invalid txs	0
Fanout outputs	0

Nodes=3, Growing, incremental ops off, fire and forget

Number of nodes	3
Number of txs	90
Avg. Confirmation Time (ms)	350.7
P99	354.7ms
P95	354.4ms
P50	352.4ms
End-to-end TPS	252.41 tx/s
Snapshots observed	2
Per-snapshot TPS P50	415.28 tx/s
Per-snapshot TPS P95	785.33 tx/s
Per-snapshot TPS max	818.23 tx/s
Number of Invalid txs	0
Fanout outputs	0

Nodes=3, Growing, incremental ops off, wait for tx valid

Number of nodes	3
Number of txs	90
Avg. Confirmation Time (ms)	61.9
P99	112.1ms
P95	92.8ms
P50	59.9ms
End-to-end TPS	46.37 tx/s
Snapshots observed	62
Per-snapshot TPS P50	43.23 tx/s
Per-snapshot TPS P95	109.99 tx/s
Per-snapshot TPS max	160.78 tx/s
Number of Invalid txs	0
Fanout outputs	0

Nodes=3, Mixed, incremental ops off, fire and forget

Each client first grows its UTxO set (1-in to 2-out) for half of its tx budget, then contracts it back (2-in to 1-out) for the remainder.

Number of nodes	3
Number of txs	90
Avg. Confirmation Time (ms)	286.1
P99	290.2ms
P95	289.8ms
P50	287.5ms
End-to-end TPS	310.03 tx/s
Snapshots observed	2
Per-snapshot TPS P50	1137.20 tx/s
Per-snapshot TPS P95	2157.06 tx/s
Per-snapshot TPS max	2247.72 tx/s
Number of Invalid txs	0
Fanout outputs	0

Nodes=3, Mixed, incremental ops off, wait for tx valid

Each client first grows its UTxO set (1-in to 2-out) for half of its tx budget, then contracts it back (2-in to 1-out) for the remainder.

Number of nodes	3
Number of txs	90
Avg. Confirmation Time (ms)	47.5
P99	84.9ms
P95	65.5ms
P50	46.2ms
End-to-end TPS	61.65 tx/s
Snapshots observed	63
Per-snapshot TPS P50	55.93 tx/s
Per-snapshot TPS P95	120.80 tx/s
Per-snapshot TPS max	153.12 tx/s
Number of Invalid txs	0
Fanout outputs	0

vrom911

I have a couple of comments 🙌🏼

popPersistentQueue previously matched the dequeued item by value equality, which could silently no-op if Eq returned False, leaving the head item stuck and blocking all subsequent writes once capacity was reached. It now pops unconditionally by index — safe because broadcastMessages is the sole consumer. Also: guard against removeFile throwing isDoesNotExistError; roll back nextIx if writeFileBS fails; emit PersistentQueueFull and PersistentQueueLoadFailed traces so operators can see when the queue is under pressure or failed to load. Drops the Eq msg constraint from withEtcdNetwork and broadcastMessages. Signed-off-by: Sasha Bogicevic <sasha.bogicevic@iohk.io>

The onException guard was unnecessary: index gaps are harmless (files are loaded by sorted listing, not assumed contiguous), and the rollback has no effect on restart since nextIx is always derived fresh from disk. Signed-off-by: Sasha Bogicevic <sasha.bogicevic@iohk.io>

A removeFile error other than "does not exist" previously propagated out of broadcastMessages' forever loop (its catch only handles GrpcException) and tore down the whole network component. By the time the delete runs the message was already broadcast and dequeued, so the failure is now traced as PersistentQueueDeleteFailed and the loop keeps going; the leftover file only means the message may be re-broadcast after a restart, same as the existing crash-recovery path. The handler stays pinned to IOException so async cancellation still propagates. Genuinely broken disks still fail fast via the unguarded write in writePersistentQueue. Also add the previously missing popPersistentQueue coverage, pinning the liveness contract behind the deadlock fixed in 553cd0b: a writer blocked at capacity must be released by a pop (using the PersistentQueueFull trace as the block signal), a 100-item soak through a capacity-10 queue asserts FIFO exactly-once delivery and full disk drainage, popped items are not reloaded on restart, a missing backing file is tolerated, a failing deletion (EISDIR) is traced without crashing, and corrupt items on load are traced as PersistentQueueLoadFailed. Signed-off-by: Sasha Bogicevic <sasha.bogicevic@iohk.io>

v0d1ch self-assigned this Jun 23, 2026

v0d1ch requested a review from a team June 23, 2026 11:13

v0d1ch added this to ☕ Hydra Team Work Jun 23, 2026

v0d1ch moved this to In review 👀 in ☕ Hydra Team Work Jun 23, 2026

vrom911 reviewed Jul 1, 2026

View reviewed changes

Comment thread hydra-node/src/Hydra/Network/Etcd.hs Outdated

Comment thread hydra-node/test/Hydra/PersistentQueueSpec.hs

noonio added this to the 2.2.1 milestone Jul 7, 2026

v0d1ch force-pushed the queue-improvements branch 2 times, most recently from d31a30f to ec6d0f7 Compare July 7, 2026 15:41

v0d1ch added 3 commits July 8, 2026 10:19

v0d1ch force-pushed the queue-improvements branch from 7f19373 to ca08cb7 Compare July 8, 2026 08:19

v0d1ch requested a review from vrom911 July 8, 2026 08:19

noonio approved these changes Jul 8, 2026

View reviewed changes

vrom911 approved these changes Jul 8, 2026

View reviewed changes

vrom911 merged commit 81afcce into master Jul 8, 2026
35 checks passed

vrom911 deleted the queue-improvements branch July 8, 2026 10:16

github-project-automation Bot moved this from In review 👀 to Done ✔ in ☕ Hydra Team Work Jul 8, 2026

This was referenced Aug 4, 2026

PersistentQueue blocks on full queue with no visibility #2688

Closed

popPersistentQueue only pops when next == item #2689

Closed

Navigation Menu

Uh oh!

Fix PersistentQueue deadlock and add operational visibility - #2742

Fix PersistentQueue deadlock and add operational visibility#2742
vrom911 merged 3 commits into
masterfrom
queue-improvements

v0d1ch commented Jun 23, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Jun 23, 2026

Uh oh!

github-actions Bot commented Jun 23, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Jun 23, 2026 •

edited

Loading

Uh oh!

vrom911 left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Conversation

v0d1ch commented Jun 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented Jun 23, 2026

Transaction cost differences

Uh oh!

github-actions Bot commented Jun 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

End-to-end benchmark differences

Baseline Scenario

Three local nodes

Uh oh!

github-actions Bot commented Jun 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Transaction costs

Script summary

Init transaction costs

Cost of Increment Transaction

Cost of Decrement Transaction

Close transaction costs

Contest transaction costs

FanOut transaction costs

PartialFanOut transaction costs

PartialFanOut transaction costs (with native tokens)

FinalPartialFanOut transaction costs (with native tokens)

End-to-end benchmark results

Baseline Scenario

Three local nodes

Scenario benchmark results

Summary across cells

Nodes=1, Constant, incremental ops off, fire and forget

Nodes=1, Constant, incremental ops off, wait for tx valid

Nodes=1, Growing, incremental ops off, fire and forget

Nodes=1, Growing, incremental ops off, wait for tx valid

Nodes=1, Mixed, incremental ops off, fire and forget

Nodes=1, Mixed, incremental ops off, wait for tx valid

Nodes=2, Constant, incremental ops off, fire and forget

Nodes=2, Constant, incremental ops off, wait for tx valid

Nodes=2, Growing, incremental ops off, fire and forget

Nodes=2, Growing, incremental ops off, wait for tx valid

Nodes=2, Mixed, incremental ops off, fire and forget

Nodes=2, Mixed, incremental ops off, wait for tx valid

Nodes=3, Constant, incremental ops off, fire and forget

Nodes=3, Constant, incremental ops off, wait for tx valid

Nodes=3, Growing, incremental ops off, fire and forget

Nodes=3, Growing, incremental ops off, wait for tx valid

Nodes=3, Mixed, incremental ops off, fire and forget

Nodes=3, Mixed, incremental ops off, wait for tx valid

Uh oh!

vrom911 left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

v0d1ch commented Jun 23, 2026 •

edited

Loading

github-actions Bot commented Jun 23, 2026 •

edited

Loading

github-actions Bot commented Jun 23, 2026 •

edited

Loading

`Init` transaction costs

`Close` transaction costs

`Contest` transaction costs

`FanOut` transaction costs

`PartialFanOut` transaction costs

`PartialFanOut` transaction costs (with native tokens)

`FinalPartialFanOut` transaction costs (with native tokens)