Browse code
Test: wait for network changes in TestNetworkDBNodeJoinLeaveIteration
In network node change test, the expected behavior is focused on how many nodes
left in networkDB, besides timing issues, things would also go tricky for a
leave-then-join sequence, if the check (counting the nodes) happened before the
first "leave" event, then the testcase actually miss its target and report PASS
without verifying its final result; if the check happened after the 'leave' event,
but before the 'join' event, the test would report FAIL unnecessary;
This code change would check both the db changes and the node count, it would
report PASS only when networkdb has indeed changed and the node count is expected.
Signed-off-by: David Wang <00107082@163.com>
David Wang authored on 2022/07/21 23:14:07Showing 1 changed files
| ... | ... |
@@ -13,6 +13,7 @@ import ( |
| 13 | 13 |
"github.com/docker/docker/pkg/stringid" |
| 14 | 14 |
"github.com/docker/go-events" |
| 15 | 15 |
"github.com/hashicorp/memberlist" |
| 16 |
+ "github.com/hashicorp/serf/serf" |
|
| 16 | 17 |
"github.com/sirupsen/logrus" |
| 17 | 18 |
"gotest.tools/v3/assert" |
| 18 | 19 |
is "gotest.tools/v3/assert/cmp" |
| ... | ... |
@@ -479,22 +480,47 @@ func TestNetworkDBCRUDMediumCluster(t *testing.T) {
|
| 479 | 479 |
func TestNetworkDBNodeJoinLeaveIteration(t *testing.T) {
|
| 480 | 480 |
dbs := createNetworkDBInstances(t, 2, "node", DefaultConfig()) |
| 481 | 481 |
|
| 482 |
+ var ( |
|
| 483 |
+ dbIndex int32 |
|
| 484 |
+ staleNetworkTime [2]serf.LamportTime |
|
| 485 |
+ expectNodeCount int |
|
| 486 |
+ network = "network1" |
|
| 487 |
+ ) |
|
| 488 |
+ dbChangeWitness := func(t poll.LogT) poll.Result {
|
|
| 489 |
+ db := dbs[dbIndex] |
|
| 490 |
+ networkTime := db.networkClock.Time() |
|
| 491 |
+ if networkTime <= staleNetworkTime[dbIndex] {
|
|
| 492 |
+ return poll.Continue("network time is stale, no change registered yet.")
|
|
| 493 |
+ } |
|
| 494 |
+ count := -1 |
|
| 495 |
+ db.Lock() |
|
| 496 |
+ if nodes, ok := db.networkNodes[network]; ok {
|
|
| 497 |
+ count = len(nodes) |
|
| 498 |
+ } |
|
| 499 |
+ db.Unlock() |
|
| 500 |
+ if count != expectNodeCount {
|
|
| 501 |
+ return poll.Continue("current number of nodes is %d, expect %d.", count, expectNodeCount)
|
|
| 502 |
+ } |
|
| 503 |
+ return poll.Success() |
|
| 504 |
+ } |
|
| 505 |
+ |
|
| 482 | 506 |
// Single node Join/Leave |
| 507 |
+ staleNetworkTime[0], staleNetworkTime[1] = dbs[0].networkClock.Time(), dbs[1].networkClock.Time() |
|
| 483 | 508 |
err := dbs[0].JoinNetwork("network1")
|
| 484 | 509 |
assert.NilError(t, err) |
| 485 | 510 |
|
| 486 |
- if len(dbs[0].networkNodes["network1"]) != 1 {
|
|
| 487 |
- t.Fatalf("The networkNodes list has to have be 1 instead of %d", len(dbs[0].networkNodes["network1"]))
|
|
| 488 |
- } |
|
| 511 |
+ dbIndex, expectNodeCount = 0, 1 |
|
| 512 |
+ poll.WaitOn(t, dbChangeWitness, poll.WithTimeout(3*time.Second), poll.WithDelay(5*time.Millisecond)) |
|
| 489 | 513 |
|
| 514 |
+ staleNetworkTime[0], staleNetworkTime[1] = dbs[0].networkClock.Time(), dbs[1].networkClock.Time() |
|
| 490 | 515 |
err = dbs[0].LeaveNetwork("network1")
|
| 491 | 516 |
assert.NilError(t, err) |
| 492 | 517 |
|
| 493 |
- if len(dbs[0].networkNodes["network1"]) != 0 {
|
|
| 494 |
- t.Fatalf("The networkNodes list has to have be 0 instead of %d", len(dbs[0].networkNodes["network1"]))
|
|
| 495 |
- } |
|
| 518 |
+ dbIndex, expectNodeCount = 0, 0 |
|
| 519 |
+ poll.WaitOn(t, dbChangeWitness, poll.WithTimeout(3*time.Second), poll.WithDelay(5*time.Millisecond)) |
|
| 496 | 520 |
|
| 497 | 521 |
// Multiple nodes Join/Leave |
| 522 |
+ staleNetworkTime[0], staleNetworkTime[1] = dbs[0].networkClock.Time(), dbs[1].networkClock.Time() |
|
| 498 | 523 |
err = dbs[0].JoinNetwork("network1")
|
| 499 | 524 |
assert.NilError(t, err) |
| 500 | 525 |
|
| ... | ... |
@@ -503,37 +529,34 @@ func TestNetworkDBNodeJoinLeaveIteration(t *testing.T) {
|
| 503 | 503 |
|
| 504 | 504 |
// Wait for the propagation on db[0] |
| 505 | 505 |
dbs[0].verifyNetworkExistence(t, dbs[1].config.NodeID, "network1", true) |
| 506 |
- if len(dbs[0].networkNodes["network1"]) != 2 {
|
|
| 507 |
- t.Fatalf("The networkNodes list has to have be 2 instead of %d - %v", len(dbs[0].networkNodes["network1"]), dbs[0].networkNodes["network1"])
|
|
| 508 |
- } |
|
| 506 |
+ dbIndex, expectNodeCount = 0, 2 |
|
| 507 |
+ poll.WaitOn(t, dbChangeWitness, poll.WithTimeout(3*time.Second), poll.WithDelay(5*time.Millisecond)) |
|
| 509 | 508 |
if n, ok := dbs[0].networks[dbs[0].config.NodeID]["network1"]; !ok || n.leaving {
|
| 510 | 509 |
t.Fatalf("The network should not be marked as leaving:%t", n.leaving)
|
| 511 | 510 |
} |
| 512 | 511 |
|
| 513 | 512 |
// Wait for the propagation on db[1] |
| 514 | 513 |
dbs[1].verifyNetworkExistence(t, dbs[0].config.NodeID, "network1", true) |
| 515 |
- if len(dbs[1].networkNodes["network1"]) != 2 {
|
|
| 516 |
- t.Fatalf("The networkNodes list has to have be 2 instead of %d - %v", len(dbs[1].networkNodes["network1"]), dbs[1].networkNodes["network1"])
|
|
| 517 |
- } |
|
| 514 |
+ dbIndex, expectNodeCount = 1, 2 |
|
| 515 |
+ poll.WaitOn(t, dbChangeWitness, poll.WithTimeout(3*time.Second), poll.WithDelay(5*time.Millisecond)) |
|
| 518 | 516 |
if n, ok := dbs[1].networks[dbs[1].config.NodeID]["network1"]; !ok || n.leaving {
|
| 519 | 517 |
t.Fatalf("The network should not be marked as leaving:%t", n.leaving)
|
| 520 | 518 |
} |
| 521 | 519 |
|
| 522 | 520 |
// Try a quick leave/join |
| 521 |
+ staleNetworkTime[0], staleNetworkTime[1] = dbs[0].networkClock.Time(), dbs[1].networkClock.Time() |
|
| 523 | 522 |
err = dbs[0].LeaveNetwork("network1")
|
| 524 | 523 |
assert.NilError(t, err) |
| 525 | 524 |
err = dbs[0].JoinNetwork("network1")
|
| 526 | 525 |
assert.NilError(t, err) |
| 527 | 526 |
|
| 528 | 527 |
dbs[0].verifyNetworkExistence(t, dbs[1].config.NodeID, "network1", true) |
| 529 |
- if len(dbs[0].networkNodes["network1"]) != 2 {
|
|
| 530 |
- t.Fatalf("The networkNodes list has to have be 2 instead of %d - %v", len(dbs[0].networkNodes["network1"]), dbs[0].networkNodes["network1"])
|
|
| 531 |
- } |
|
| 528 |
+ dbIndex, expectNodeCount = 0, 2 |
|
| 529 |
+ poll.WaitOn(t, dbChangeWitness, poll.WithTimeout(3*time.Second), poll.WithDelay(5*time.Millisecond)) |
|
| 532 | 530 |
|
| 533 | 531 |
dbs[1].verifyNetworkExistence(t, dbs[0].config.NodeID, "network1", true) |
| 534 |
- if len(dbs[1].networkNodes["network1"]) != 2 {
|
|
| 535 |
- t.Fatalf("The networkNodes list has to have be 2 instead of %d - %v", len(dbs[1].networkNodes["network1"]), dbs[1].networkNodes["network1"])
|
|
| 536 |
- } |
|
| 532 |
+ dbIndex, expectNodeCount = 1, 2 |
|
| 533 |
+ poll.WaitOn(t, dbChangeWitness, poll.WithTimeout(3*time.Second), poll.WithDelay(5*time.Millisecond)) |
|
| 537 | 534 |
|
| 538 | 535 |
closeNetworkDBInstances(t, dbs) |
| 539 | 536 |
} |