Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

checker: priority to fix orphan peers (#3522) #3529

Merged
merged 5 commits into from
May 10, 2021
Merged

checker: priority to fix orphan peers (#3522) #3529

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
8359cbb
cherry pick #3522 to release-4.0
nolouch Mar 25, 2021
94b0c53
Merge branch 'release-4.0' into release-4.0-60508bde0819
nolouch Apr 6, 2021
fd25878
fix
nolouch Apr 23, 2021
15e9a07
Merge branch 'release-4.0' into release-4.0-60508bde0819
ti-chi-bot May 10, 2021
efe8dcf
Merge branch 'release-4.0' into release-4.0-60508bde0819
ti-chi-bot May 10, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 6 additions & 6 deletions server/schedule/checker/rule_checker.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -61,6 +61,11 @@ func (c *RuleChecker) Check(region *core.RegionInfo) *operator.Operator {
// multiple rules.
return c.fixRange(region)
}
op, err := c.fixOrphanPeers(region, fit)
if err == nil && op != nil {
return op
}
log.Debug("fail to fix orphan peer", errs.ZapError(err))
for _, rf := range fit.RuleFits {
op, err := c.fixRulePeer(region, fit, rf)
if err != nil {
Expand All @@ -71,12 +76,7 @@ func (c *RuleChecker) Check(region *core.RegionInfo) *operator.Operator {
return op
}
}
op, err := c.fixOrphanPeers(region, fit)
if err != nil {
log.Debug("fail to fix orphan peer", errs.ZapError(err))
return nil
}
return op
return nil
}

func (c *RuleChecker) fixRange(region *core.RegionInfo) *operator.Operator {
Expand Down
164 changes: 164 additions & 0 deletions server/schedule/checker/rule_checker_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -299,3 +299,167 @@ func (s *testRuleCheckerSuite) TestIssue2419(c *C) {
c.Assert(op.Step(1).(operator.PromoteLearner).ToStore, Equals, uint64(4))
c.Assert(op.Step(2).(operator.RemovePeer).FromStore, Equals, uint64(3))
}
<<<<<<< HEAD
=======

// Ref https://github.com/tikv/pd/issues/3521
// The problem is when offline a store, we may add learner multiple times if
// the operator is timeout.
func (s *testRuleCheckerSuite) TestIssue3521_PriorityFixOrphanPeer(c *C) {
s.cluster.AddLabelsStore(1, 1, map[string]string{"host": "host1"})
s.cluster.AddLabelsStore(2, 1, map[string]string{"host": "host1"})
s.cluster.AddLabelsStore(3, 1, map[string]string{"host": "host2"})
s.cluster.AddLabelsStore(4, 1, map[string]string{"host": "host4"})
s.cluster.AddLabelsStore(5, 1, map[string]string{"host": "host5"})
s.cluster.AddLeaderRegionWithRange(1, "", "", 1, 2, 3)
op := s.rc.Check(s.cluster.GetRegion(1))
c.Assert(op, IsNil)
var add operator.AddLearner
var remove operator.RemovePeer
s.cluster.SetStoreOffline(2)
op = s.rc.Check(s.cluster.GetRegion(1))
c.Assert(op, NotNil)
c.Assert(op.Step(0), FitsTypeOf, add)
c.Assert(op.Desc(), Equals, "replace-rule-offline-peer")
r := s.cluster.GetRegion(1).Clone(core.WithAddPeer(
&metapb.Peer{
Id: 5,
StoreId: 4,
Role: metapb.PeerRole_Learner,
}))
s.cluster.PutRegion(r)
op = s.rc.Check(s.cluster.GetRegion(1))
c.Assert(op.Step(0), FitsTypeOf, remove)
c.Assert(op.Desc(), Equals, "remove-orphan-peer")
}

func (s *testRuleCheckerSuite) TestIssue3293(c *C) {
s.cluster.AddLabelsStore(1, 1, map[string]string{"host": "host1"})
s.cluster.AddLabelsStore(2, 1, map[string]string{"host": "host1"})
s.cluster.AddLabelsStore(3, 1, map[string]string{"host": "host2"})
s.cluster.AddLabelsStore(4, 1, map[string]string{"host": "host4"})
s.cluster.AddLabelsStore(5, 1, map[string]string{"host": "host5"})
s.cluster.AddLeaderRegionWithRange(1, "", "", 1, 2)
err := s.ruleManager.SetRule(&placement.Rule{
GroupID: "TiDB_DDL_51",
ID: "0",
Role: placement.Follower,
Count: 1,
LabelConstraints: []placement.LabelConstraint{
{
Key: "host",
Values: []string{
"host5",
},
Op: placement.In,
},
},
})
c.Assert(err, IsNil)
s.cluster.DeleteStore(s.cluster.TakeStore(5))
err = s.ruleManager.SetRule(&placement.Rule{
GroupID: "TiDB_DDL_51",
ID: "default",
Role: placement.Voter,
Count: 3,
})
c.Assert(err, IsNil)
err = s.ruleManager.DeleteRule("pd", "default")
c.Assert(err, IsNil)
op := s.rc.Check(s.cluster.GetRegion(1))
c.Assert(op, NotNil)
c.Assert(op.Desc(), Equals, "add-rule-peer")
}

func (s *testRuleCheckerSuite) TestIssue3299(c *C) {
s.cluster.AddLabelsStore(1, 1, map[string]string{"host": "host1"})
s.cluster.AddLabelsStore(2, 1, map[string]string{"dc": "sh"})
s.cluster.AddLeaderRegionWithRange(1, "", "", 1, 2)

testCases := []struct {
constraints []placement.LabelConstraint
err string
}{
{
constraints: []placement.LabelConstraint{
{
Key: "host",
Values: []string{"host5"},
Op: placement.In,
},
},
err: ".*can not match any store",
},
{
constraints: []placement.LabelConstraint{
{
Key: "ho",
Values: []string{"sh"},
Op: placement.In,
},
},
err: ".*can not match any store",
},
{
constraints: []placement.LabelConstraint{
{
Key: "host",
Values: []string{"host1"},
Op: placement.In,
},
{
Key: "host",
Values: []string{"host1"},
Op: placement.NotIn,
},
},
err: ".*can not match any store",
},
{
constraints: []placement.LabelConstraint{
{
Key: "host",
Values: []string{"host1"},
Op: placement.In,
},
{
Key: "host",
Values: []string{"host3"},
Op: placement.In,
},
},
err: ".*can not match any store",
},
{
constraints: []placement.LabelConstraint{
{
Key: "host",
Values: []string{"host1"},
Op: placement.In,
},
{
Key: "host",
Values: []string{"host1"},
Op: placement.In,
},
},
err: "",
},
}

for _, t := range testCases {
err := s.ruleManager.SetRule(&placement.Rule{
GroupID: "p",
ID: "0",
Role: placement.Follower,
Count: 1,
LabelConstraints: t.constraints,
})
if t.err != "" {
c.Assert(err, ErrorMatches, t.err)
} else {
c.Assert(err, IsNil)
}
}
}
>>>>>>> 60508bde... checker: priority to fix orphan peers (#3522)