markets: OnDealExpiredOrSlashed - get deal by proposal instead of deal ID

[dirkmc]

Fixes #5414
Depends on #5309 filecoin-project/go-fil-markets#476
Also move common code for OnDealExpiredOrSlashed from storage adapter client and provider into one place.

[magik6k]

Technically we can reorg after getting this info

[dirkmc]

We can, but in order to fix that we would need to change the way the state change API works.
I suggest we merge this improvement first and then tackle the state change API changes.

[arajasek]

@dirkmc What's up with this PR?

[dirkmc]

I think it was just waiting for approval - if we want to get this in let me know and I'll rebase

[jennijuju]

@dirkmc we opened it for review - could you please resolve the conflict?

[dirkmc]

@jennijuju I've resolved the conflict now, please review when ready

[jennijuju]

still waiting filecoin-project/go-fil-markets#476 and will be reviewed by the ignite team soon.

[raulk]

Here's a start for a test. Unfortunately the sector start and end are set too far out in the future (epoch 23k), so it takes a long while to run. What worried me is that despite the expiry epoch elapsing, the deal did not appear to be set as "expired" on the client side. I might have a bug in my test, or there's another problem here. Abstaining from merging until we get to the bottom of this.

itests/deal_expiry_test.go:

func TestDealExpiry(t *testing.T) {
	kit.QuietMiningLogs()

	// reset minimum deal duration to 0, so we can make very short-lived deals.
	// NOTE: this will need updating with every new specs-actors version.
	market2.DealMinDuration = 0
	market3.DealMinDuration = 0
	market4.DealMinDuration = 0
	market5.DealMinDuration = 0

	ctx := context.Background()

	var (
		client kit.TestFullNode
		miner1 kit.TestMiner
	)

	ens := kit.NewEnsemble(t, kit.MockProofs())
	ens.FullNode(&client)
	ens.Miner(&miner1, &client, kit.WithAllSubsystems())
	bm := ens.Start().InterconnectAll().BeginMining(50 * time.Millisecond)

	dh := kit.NewDealHarness(t, &client, &miner1, &miner1)

	client.WaitTillChain(ctx, kit.HeightAtLeast(5))

	_, _, _ = dh.MakeOnlineDeal(ctx, kit.MakeFullDealParams{
		Rseed:             0,
		FastRet:           true,
		StartEpoch:        0,
		MinBlocksDuration: 50,
	})

	go func() {
		ch, _ := client.ChainNotify(ctx)
		for range ch {
			bm[0].InjectNulls(10)
		}
	}()

	for {
		// sectors, err := miner.SectorsList(ctx)
		// require.NoError(t, err)

		minerDeals, err := miner1.DealsList(ctx)
		require.NoError(t, err)

		for _, d := range minerDeals {
			spew.Dump(d)
		}

		ts, err := client.ChainHead(ctx)
		require.NoError(t, err)

		fmt.Println(ts.Height())

		deals, err := client.ClientListDeals(ctx)
		require.NoError(t, err)

		fmt.Println(storagemarket.DealStates[deals[0].State])
		time.Sleep(1 * time.Second)

		fmt.Printf("%+v\n", deals[0])
	}
}

[raulk]

See comment above.

[raulk]

We would benefit from two more tests to verify:

• behaviour on miner-initiated sector termination
• behaviour on protocol-initiated sector termination (aka. slashing)

But happy merging this as-is, as it's already strictly better than what we have in master. @dirkmc should we open issues to track the remaining tests?

[dirkmc]

Created issues to track

• behaviour on miner-initiated sector termination create test: miner-initiated sector termination #7198
• behaviour on protocol-initiated sector termination (aka. slashing) create test: protocol-initiated sector termination (slashing) #7199

[codecov]

Codecov Report

Merging #5431 (ac51b8e) into master (f33c02e) will increase coverage by 0.04%.
The diff coverage is 54.79%.

@@            Coverage Diff             @@
##           master    #5431      +/-   ##
==========================================
+ Coverage   39.04%   39.09%   +0.04%     
==========================================
  Files         607      608       +1     
  Lines       64565    64542      -23     
==========================================
+ Hits        25212    25230      +18     
+ Misses      34963    34929      -34     
+ Partials     4390     4383       -7     

Impacted Files	Coverage Δ
node/modules/storageminer.go	62.13% <0.00%> (ø)
markets/storageadapter/ondealexpired.go	43.39% <43.39%> (ø)
chain/events/events_called.go	81.32% <71.42%> (ø)
itests/kit/deals.go	87.95% <100.00%> (+0.14%)	⬆️
markets/storageadapter/client.go	49.68% <100.00%> (+7.18%)	⬆️
markets/storageadapter/provider.go	49.10% <100.00%> (+6.86%)	⬆️
extern/sector-storage/sched_worker.go	76.53% <0.00%> (-1.45%)	⬇️
chain/vm/vm.go	59.47% <0.00%> (-1.12%)	⬇️
storage/wdpost_changehandler.go	97.64% <0.00%> (-0.95%)	⬇️
chain/sync_manager.go	66.77% <0.00%> (-0.63%)	⬇️
... and 9 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update f33c02e...ac51b8e. Read the comment docs.