[itests] Bitcoind integration test by halseth · Pull Request #1583 · lightningnetwork/lnd

halseth · 2018-07-18T15:53:19Z

This PR enables the bitcoind backend for integration tests.

wpaulino · 2019-07-31T00:09:24Z

lntest/itest/lnd_test.go

Nit: use harnessNetParams?

wpaulino · 2019-07-31T00:10:09Z

lntest/btcd.go

Nit: would be nice if this was provided as a function parameter.

wpaulino · 2019-07-31T00:22:00Z

lntest/harness.go

Any insights as to why this had to be increased?

Because bitcoind was slower to catch up block hashes IIRC. Since this should be faster now I can try to revert this.

lntest/bitcoind.go

Roasbeef · 2019-08-06T00:27:07Z

lntest/bitcoind.go

IIRC, this is defined elsewhere already?

It is re-defined since it is behind a different build flag.

Roasbeef · 2019-08-06T00:30:10Z

itests look to be failing with an infrequent error?

halseth · 2019-08-07T12:08:37Z

itests look to be failing with an infrequent error?

Should be fixed by d531fd3

halseth · 2019-08-07T13:57:12Z

btw @Roasbeef, any reasons not to switch to regtest also for btcd/neutrino?

wpaulino · 2019-08-08T01:49:44Z

Did a couple more travis runs and found the following flakes:

    --- FAIL: TestLightningNetworkDaemon/test_multi-hop_htlc_local_force_close_immediate_expiry (44.08s)
        lnd_test.go:99: Failed: (test multi-hop htlc local force close immediate expiry): exited with error: 
            *errors.errorString unable to find bob's funding output sweep tx: wanted 1, found 2 txs in mempool: [053fe1382d7e4b376490c2b9d18bf8c2cde6e46e6ff89bf649f1d4dadd8eca61 05b1d7d90eea4cdc8c91b2b4f2da68bab3d8525fd3ff0f5c5f47550c06a2af02]
            /home/travis/gopath/src/github.com/lightningnetwork/lnd/lntest/itest/lnd_test.go:10014 (0xd3ce49)
            	testMultiHopHtlcLocalTimeout: t.Fatalf("unable to find bob's funding output sweep tx: %v", err)
            /home/travis/gopath/src/github.com/lightningnetwork/lnd/lntest/itest/lnd_test.go:123 (0xd03d85)
            	(*harnessTest).RunTestCase: testCase.test(h.lndHarness, h)
            /home/travis/gopath/src/github.com/lightningnetwork/lnd/lntest/itest/lnd_test.go:14318 (0xd6de74)
            	TestLightningNetworkDaemon.func4: ht.RunTestCase(testCase)
            /home/travis/.gimme/versions/go1.12.7.linux.amd64/src/testing/testing.go:865 (0x4fcd20)
            	tRunner: fn(t)
            /home/travis/.gimme/versions/go1.12.7.linux.amd64/src/runtime/asm_amd64.s:1337 (0x45e6d1)
            	goexit: BYTE	$0x90	// NOP

    --- FAIL: TestLightningNetworkDaemon/test_multi-hop_htlc_local_force_close_immediate_expiry (41.93s)
        lnd_test.go:99: Failed: (test multi-hop htlc local force close immediate expiry): exited with error: 
            *errors.errorString bob still has pending channels but shouldn't: (*lnrpc.PendingChannelsResponse)(0xc000d70870)(total_limbo_balance:21713 pending_force_closing_channels:<channel:<remote_node_pub:"02aae00790835d168caa7ae46eab9fdaf317b65cf5bfbda0230e1cf065f08d48c9" channel_point:"5250884a5f380defa16cc768ba6d9a13d841745ca37a863b26473ebca5fd1f43:1" capacity:1000000 local_balance:958700 > closing_txid:"a95012857089bc3e06614a6b11cf850211a4803d10b1c9a0600a463a4c7e5673" limbo_balance:21713 maturity_height:1046 blocks_til_maturity:-6 recovered_balance:958700 pending_htlcs:<amount:21713 outpoint:"cc4fc894a49cacf84b3ca4bb621ae61a4c99b1f888eb7dc9d6252648d13f11b2:0" maturity_height:1051 blocks_til_maturity:-1 stage:2 > > )
            /home/travis/gopath/src/github.com/lightningnetwork/lnd/lntest/itest/lnd_test.go:10115 (0xd3d764)
            	testMultiHopHtlcLocalTimeout: t.Fatalf(predErr.Error())
            /home/travis/gopath/src/github.com/lightningnetwork/lnd/lntest/itest/lnd_test.go:123 (0xd03d85)
            	(*harnessTest).RunTestCase: testCase.test(h.lndHarness, h)
            /home/travis/gopath/src/github.com/lightningnetwork/lnd/lntest/itest/lnd_test.go:14318 (0xd6de74)
            	TestLightningNetworkDaemon.func4: ht.RunTestCase(testCase)
            /home/travis/.gimme/versions/go1.12.7.linux.amd64/src/testing/testing.go:865 (0x4fcd20)
            	tRunner: fn(t)
            /home/travis/.gimme/versions/go1.12.7.linux.amd64/src/runtime/asm_amd64.s:1337 (0x45e6d1)
            	goexit: BYTE	$0x90	// NOP

wpaulino · 2019-08-13T01:06:28Z

Tracked down the failures above and they should be resolved with the following patch:

diff --git a/lntest/bitcoind.go b/lntest/bitcoind.go
index dd0cc5e9..3e7c1d35 100644
--- a/lntest/bitcoind.go
+++ b/lntest/bitcoind.go
@@ -102,6 +102,7 @@ func NewBackend(miner string, netParams *chaincfg.Params) (
 		"-datadir="+tempBitcoindDir,
 		"-regtest",
 		"-txindex",
+		"-debug",
 		"-whitelist=127.0.0.1", // whitelist localhost to speed up relay
 		"-rpcauth=weks:469e9bb14ab2360f8e226efed5ca6f"+
 			"d$507c670e800a95284294edb5773b05544b"+
diff --git a/lntest/itest/lnd_test.go b/lntest/itest/lnd_test.go
index 85212435..c2f9e5ab 100644
--- a/lntest/itest/lnd_test.go
+++ b/lntest/itest/lnd_test.go
@@ -10004,19 +10004,13 @@ func testMultiHopHtlcLocalTimeout(net *lntest.NetworkHarness, t *harnessTest) {
 
 	// We'll mine defaultCSV blocks in order to generate the sweep
 	// transaction of Bob's funding output. This will also bring us to the
-	// maturity height of the htlc tx output.
+	// maturity height of the htlc tx output, so we should expect to see
+	// both transactions in the mempool.
 	if _, err := net.Miner.Node.Generate(defaultCSV); err != nil {
 		t.Fatalf("unable to generate blocks: %v", err)
 	}
 
-	_, err = waitForTxInMempool(net.Miner.Node, minerMempoolTimeout)
-	if err != nil {
-		t.Fatalf("unable to find bob's funding output sweep tx: %v", err)
-	}
-
-	// The second layer HTLC timeout transaction should now have been
-	// broadcast on-chain.
-	secondLayerHash, err := waitForTxInMempool(net.Miner.Node, minerMempoolTimeout)
+	txids, err := waitForNTxsInMempool(net.Miner.Node, 2, minerMempoolTimeout)
 	if err != nil {
 		t.Fatalf("unable to find bob's second layer transaction")
 	}
@@ -10045,12 +10039,14 @@ func testMultiHopHtlcLocalTimeout(net *lntest.NetworkHarness, t *harnessTest) {
 
 	// Now we'll mine an additional block, which should include the second
 	// layer sweep tx.
-	block := mineBlocks(t, net, 1, 1)[0]
+	block := mineBlocks(t, net, 1, 2)[0]
 
 	// The block should have confirmed Bob's second layer sweeping
 	// transaction. Therefore, at this point, there should be no active
 	// HTLC's on the commitment transaction from Alice -> Bob.
-	assertTxInBlock(t, block, secondLayerHash)
+	for _, txid := range txids {
+		assertTxInBlock(t, block, txid)
+	}
 	nodes = []*lntest.HarnessNode{net.Alice}
 	err = lntest.WaitPredicate(func() bool {
 		predErr = assertNumActiveHtlcs(nodes, 0)

Roasbeef · 2019-08-13T01:38:36Z

@wpaulino so the flakes weren't fundamental to bitcoind, but just showed up more often, making the patch above a general flake fix for all backends?

wpaulino · 2019-08-13T01:43:17Z

Correct. Things seem to be slower with bitcoind, likely due to the several HTTP requests made. The patch above solves an incorrect assumption within the test itself. We should expect both bob's commit sweep and HTLC timeout transactions in the mempool before proceeding. Without the patch we'd only wait for one of these twice, causing us to proceed with further assumptions that rely on this one.

wpaulino · 2019-08-27T22:32:18Z

This PR should now be rebased on top of #3434 in order to address the flake pointed out above.

…annel Otherwise following tests would be flaky because of unexpected sweep transaction in the mempool.

Co-authored-by: Wilmer Paulino <wilmer.paulino@gmail.com>

The local timeout could be too short for certain backends.

halseth · 2019-09-03T11:25:30Z

Looks good after rebase on #3434 👍

halseth · 2019-09-03T11:25:43Z

PTAL @wpaulino @Roasbeef

wpaulino

LGTM ⚡️

wpaulino · 2019-09-04T21:09:11Z

lntest/bitcoind.go

+		"bitcoind",
+		"-datadir="+tempBitcoindDir,
+		"-regtest",
+		"-txindex",


Should we run without this? Would ensure that the fallback path (no txindex, scan manually) is tested instead.

We could run it with both options, though this would significantly increase the running time of the current integration tests. On the other hand, we properly test both variants in the unit tests of the notifier.

Could randomize it lol

Roasbeef

LGTM 🖌

halseth force-pushed the multibackend-integration-test branch 24 times, most recently from d524723 to 19bfc14 Compare July 25, 2018 09:54

halseth force-pushed the multibackend-integration-test branch from 19bfc14 to 0e292f4 Compare August 6, 2018 11:43

halseth force-pushed the multibackend-integration-test branch 5 times, most recently from 8c78286 to 5b4296a Compare September 10, 2018 12:42

wpaulino added v0.8.0 P2 should be fixed if one has time labels Jul 30, 2019

wpaulino reviewed Jul 31, 2019

View reviewed changes

Roasbeef reviewed Aug 6, 2019

View reviewed changes

halseth force-pushed the multibackend-integration-test branch 2 times, most recently from 2ac4811 to 190a40c Compare August 7, 2019 12:07

halseth force-pushed the multibackend-integration-test branch from 190a40c to f5b5fd4 Compare August 28, 2019 13:42

halseth and others added 6 commits September 2, 2019 09:37

lnd_test: use RegTest instead of SimNet during integration tests

35f3544

lntest: add BitcoindBackendCfg

45d41dc

lntest: compile time check btcd and neutrino BackendCfg interface

a57b3de

make+travis: enable bitcoind itests on travis

857a2c3

lntest/itest: cleanup force closed channel in testSendUpdateDisableCh…

168fc4e

…annel Otherwise following tests would be flaky because of unexpected sweep transaction in the mempool.

lntest/bitcoind: set debug loglevel

c3480c6

Co-authored-by: Wilmer Paulino <wilmer.paulino@gmail.com>

halseth force-pushed the multibackend-integration-test branch from f5b5fd4 to c3480c6 Compare September 2, 2019 07:40

lntest/itest: use timeout constants in rejectHTLC test

d7f0372

The local timeout could be too short for certain backends.

wpaulino requested review from Roasbeef and wpaulino September 4, 2019 02:50

wpaulino approved these changes Sep 4, 2019

View reviewed changes

Roasbeef approved these changes Sep 5, 2019

View reviewed changes

Roasbeef merged commit b43a9f2 into lightningnetwork:master Sep 5, 2019

Conversation

halseth commented Jul 18, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Roasbeef commented Aug 6, 2019

Uh oh!

halseth commented Aug 7, 2019

Uh oh!

halseth commented Aug 7, 2019

Uh oh!

wpaulino commented Aug 8, 2019

Uh oh!

wpaulino commented Aug 13, 2019

Uh oh!

Roasbeef commented Aug 13, 2019

Uh oh!

wpaulino commented Aug 13, 2019

Uh oh!

wpaulino commented Aug 27, 2019

Uh oh!

halseth commented Sep 3, 2019

Uh oh!

halseth commented Sep 3, 2019

Uh oh!

wpaulino left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Roasbeef left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

halseth commented Jul 18, 2018 •

edited

Loading