[Bug]: flaky simulation test: unique constraint violation

[mark-rushakoff]

Summary of Bug

https://github.com/cosmos/cosmos-sdk/actions/runs/4638318575/jobs/8208005708?pr=15727 failed with stack:

--- FAIL: TestAppStateDeterminism (0.27s)
panic: group policies: unique constraint violation [recovered]
	panic: group policies: unique constraint violation

goroutine 230 [running]:
testing.tRunner.func1.2({0x25139e0, 0xc0002e4b60})
	/opt/hostedtoolcache/go/1.20.3/x64/src/testing/testing.go:1526 +0x24e
testing.tRunner.func1()
	/opt/hostedtoolcache/go/1.20.3/x64/src/testing/testing.go:1529 +0x39f
panic({0x25139e0, 0xc0002e4b60})
	/opt/hostedtoolcache/go/1.20.3/x64/src/runtime/panic.go:884 +0x213
github.com/cosmos/cosmos-sdk/x/group/keeper.Keeper.InitGenesis({{0x3353b40, 0xc000c9bda0}, {0x7fcdd4386590, 0xc000d21560}, {0xc001827500, {0x1}}, {0x3374660, 0xc000981380}, {0xc001827560}, {0x3374660, ...}, ...}, ...)
	/home/runner/work/cosmos-sdk/cosmos-sdk/x/group/keeper/genesis.go:29 +0x645
github.com/cosmos/cosmos-sdk/x/group/module.AppModule.InitGenesis(...)
	/home/runner/work/cosmos-sdk/cosmos-sdk/x/group/module/module.go:133
github.com/cosmos/cosmos-sdk/types/module.(*Manager).InitGenesis(_, {{0x336f7d8, 0xc0000dc000}, {0x3382c10, 0xc0023fe640}, {{0x0, 0x0}, {0x27839ae, 0xe}, 0x0, ...}, ...}, ...)
	/home/runner/work/cosmos-sdk/cosmos-sdk/types/module/module.go:419 +0x3b3
github.com/cosmos/cosmos-sdk/runtime.(*App).InitChainer(_, {{0x336f7d8, 0xc0000dc000}, {0x3382c10, 0xc0023fe640}, {{0x0, 0x0}, {0x27839ae, 0xe}, 0x0, ...}, ...}, ...)
	/home/runner/work/cosmos-sdk/cosmos-sdk/runtime/app.go:134 +0x12e
github.com/cosmos/cosmos-sdk/baseapp.(*BaseApp).InitChain(0xc0003d4e00, {{0x0, 0xf3e97aec7, 0x4876d80}, {0x27839ae, 0xe}, 0xc000149e00, {0x0, 0x0, 0x0}, ...})
	/home/runner/work/cosmos-sdk/cosmos-sdk/baseapp/abci.go:83 +0x635
github.com/cosmos/cosmos-sdk/x/simulation.initChain(0x0?, {0x3fda9e03ce8b6356, 0x3, _, {_, _, _}, {_, _}, {_, ...}}, ...)
	/home/runner/work/cosmos-sdk/cosmos-sdk/x/simulation/simulate.go:42 +0x265
github.com/cosmos/cosmos-sdk/x/simulation.SimulateFromSeed({0x3398660?, 0xc000728340?}, {0x334fb60?, _}, _, _, _, {0xc003085500, 0x2e, 0x38}, ...)
	/home/runner/work/cosmos-sdk/cosmos-sdk/x/simulation/simulate.go:76 +0x4b5
cosmossdk.io/simapp.TestAppStateDeterminism(0xc000728340)
	/home/runner/work/cosmos-sdk/cosmos-sdk/simapp/sim_test.go:371 +0xcc8
testing.tRunner(0xc000728340, 0x2efc168)
	/opt/hostedtoolcache/go/1.20.3/x64/src/testing/testing.go:1576 +0x10b
created by testing.(*T).Run
	/opt/hostedtoolcache/go/1.20.3/x64/src/testing/testing.go:1629 +0x3ea
exit status 2
FAIL	cosmossdk.io/simapp	0.357s

Version

main branch at commit df161c2

Steps to Reproduce

Apply this patch to use the fixed seed from the failure:

diff --git a/simapp/sim_test.go b/simapp/sim_test.go
index b30f1771c..d89088a13 100644
--- a/simapp/sim_test.go
+++ b/simapp/sim_test.go
@@ -351,6 +351,7 @@ func TestAppStateDeterminism(t *testing.T) {

        for i := 0; i < numSeeds; i++ {
                config.Seed = rand.Int63()
+               config.Seed = 7748983398051311403

                for j := 0; j < numTimesToRunPerSeed; j++ {
                        var logger log.Logger

(the -Seed flag isn't respected in this test)

The failure reproduces immediately.

Upon investigating the state, the group policies look like:

[
  {
    "address": "cosmos138ya066vjg82r7ldrhnw9z99auvk24ksqu4ze9",
    "group_id": "1",
    "admin": "cosmos138ya066vjg82r7ldrhnw9z99auvk24ksqu4ze9",
    "metadata": "MGZkMAdSAQ",
    "version": "1",
    "decision_policy": {
      "@type": "/cosmos.group.v1.ThresholdDecisionPolicy",
      "threshold": "10",
      "windows": {
        "voting_period": "1s",
        "min_execution_period": "0s"
      }
    },
    "created_at": "0001-01-01T00:00:00Z"
  },
  {
    "address": "cosmos1h85jr2t5n0ma8cl0s5v3zg9xdufkjm6a9h2sv7",
    "group_id": "2",
    "admin": "cosmos1h85jr2t5n0ma8cl0s5v3zg9xdufkjm6a9h2sv7",
    "metadata": "tSdwuNguAr",
    "version": "1",
    "decision_policy": {
      "@type": "/cosmos.group.v1.ThresholdDecisionPolicy",
      "threshold": "10",
      "windows": {
        "voting_period": "1s",
        "min_execution_period": "0s"
      }
    },
    "created_at": "0001-01-01T00:00:00Z"
  },
  {
    "address": "cosmos138ya066vjg82r7ldrhnw9z99auvk24ksqu4ze9",
    "group_id": "3",
    "admin": "cosmos138ya066vjg82r7ldrhnw9z99auvk24ksqu4ze9",
    "metadata": "nbQqkXjdmH",
    "version": "1",
    "decision_policy": {
      "@type": "/cosmos.group.v1.ThresholdDecisionPolicy",
      "threshold": "10",
      "windows": {
        "voting_period": "1s",
        "min_execution_period": "0s"
      }
    },
    "created_at": "0001-01-01T00:00:00Z"
  }
]

The address and admin are identical on the first and last entry. Since the failure mentions an invalid unique constraint, maybe we need to prevent generating that state?

[mark-rushakoff]

This happened again at https://github.com/cosmos/cosmos-sdk/actions/runs/4679650341/jobs/8289964158#step:5:4980 for #15819 (changing unrelated CI scripts), with stack trace:

--- FAIL: TestAppStateDeterminism (1430.48s)
panic: group policies: unique constraint violation [recovered]
	panic: group policies: unique constraint violation

goroutine 227 [running]:
testing.tRunner.func1.2({0x254b3e0, 0xc00576d4c0})
	/opt/hostedtoolcache/go/1.20.3/x64/src/testing/testing.go:1526 +0x24e
testing.tRunner.func1()
	/opt/hostedtoolcache/go/1.20.3/x64/src/testing/testing.go:1529 +0x39f
panic({0x254b3e0, 0xc00576d4c0})
	/opt/hostedtoolcache/go/1.20.3/x64/src/runtime/panic.go:884 +0x213
github.com/cosmos/cosmos-sdk/x/group/keeper.Keeper.InitGenesis({{0x3391fc8, 0xc041c68c00}, {0x7f4f6c356dd8, 0xc05af708f0}, {0xc057d997a0, {0x1}}, {0x33b29e0, 0xc05bbfb940}, {0xc057d99800}, {0x33b29e0, ...}, ...}, ...)
	/home/runner/work/cosmos-sdk/cosmos-sdk/x/group/keeper/genesis.go:29 +0x645
github.com/cosmos/cosmos-sdk/x/group/module.AppModule.InitGenesis(...)
	/home/runner/work/cosmos-sdk/cosmos-sdk/x/group/module/module.go:133
github.com/cosmos/cosmos-sdk/types/module.(*Manager).InitGenesis(_, {{0x33add58, 0xc000058128}, {0x33c1530, 0xc006b6af40}, {{0x0, 0x0}, {0x27bfc3a, 0xe}, 0x0, ...}, ...}, ...)
	/home/runner/work/cosmos-sdk/cosmos-sdk/types/module/module.go:419 +0x3b3
github.com/cosmos/cosmos-sdk/runtime.(*App).InitChainer(_, {{0x33add58, 0xc000058128}, {0x33c1530, 0xc006b6af40}, {{0x0, 0x0}, {0x27bfc3a, 0xe}, 0x0, ...}, ...}, ...)
	/home/runner/work/cosmos-sdk/cosmos-sdk/runtime/app.go:134 +0x12e
github.com/cosmos/cosmos-sdk/baseapp.(*BaseApp).InitChain(0xc05bb28400, {{0x0, 0x1054c3d9b2, 0x48cef60}, {0x27bfc3a, 0xe}, 0xc0036a62a0, {0x0, 0x0, 0x0}, ...})
	/home/runner/work/cosmos-sdk/cosmos-sdk/baseapp/abci.go:84 +0x5b5
github.com/cosmos/cosmos-sdk/x/simulation.initChain(0x0?, {0x3f92f1707cee6af8, 0x19, _, {_, _, _}, {_, _}, {_, ...}}, ...)
	/home/runner/work/cosmos-sdk/cosmos-sdk/x/simulation/simulate.go:42 +0x265
github.com/cosmos/cosmos-sdk/x/simulation.SimulateFromSeed({0x33d70c0?, 0xc000647d40?}, {0x338dd80?, _}, _, _, _, {0xc00d05ae00, 0x2e, 0x38}, ...)
	/home/runner/work/cosmos-sdk/cosmos-sdk/x/simulation/simulate.go:76 +0x4b5
cosmossdk.io/simapp.TestAppStateDeterminism(0xc000647d40)
	/home/runner/work/cosmos-sdk/cosmos-sdk/simapp/sim_test.go:371 +0xcc8
testing.tRunner(0xc000647d40, 0x2f38500)
	/opt/hostedtoolcache/go/1.20.3/x64/src/testing/testing.go:1576 +0x10b
created by testing.(*T).Run
	/opt/hostedtoolcache/go/1.20.3/x64/src/testing/testing.go:1629 +0x3ea
exit status 2
FAIL	cosmossdk.io/simapp	1430.635s
make: *** [Makefile:258: test-sim-nondeterminism] Error 1

and with seed 8979509998758717859.

[gjermundgaraba]

We're seeing the same issue: EmpowerPlastic/empowerchain#633

And: https://github.com/EmpowerPlastic/empowerchain/actions/runs/4783448308/jobs/8503754443?pr=628