Skip to content

kvserver: record follower write bytes in replica load #147449

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

Draft
wants to merge 1 commit into
base: master
Choose a base branch
from
Draft

kvserver: record follower write bytes in replica load #147449

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
16 changes: 12 additions & 4 deletions pkg/kv/kvserver/replica_raft.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1285,10 +1285,18 @@ func (r *Replica) handleRaftReadyRaftMuLocked(
return stats, err
}

if r.store.cfg.KVAdmissionController != nil &&
stats.apply.followerStoreWriteBytes.NumEntries > 0 {
r.store.cfg.KVAdmissionController.FollowerStoreWriteBytes(
r.store.StoreID(), stats.apply.followerStoreWriteBytes)
if stats.apply.followerStoreWriteBytes.NumEntries > 0 {
// NB: The write bytes for non-followers are recorded in replica
// send, the folower bytes are accounted for here on the follower
// replica's store.
r.loadStats.RecordWriteBytes(
float64(stats.apply.followerStoreWriteBytes.IngestedBytes) +
float64(stats.apply.followerStoreWriteBytes.WriteBytes))

if r.store.cfg.KVAdmissionController != nil {
r.store.cfg.KVAdmissionController.FollowerStoreWriteBytes(
r.store.StoreID(), stats.apply.followerStoreWriteBytes)
}
}

// etcd raft occasionally adds a nil entry (our own commands are never
Expand Down
74 changes: 47 additions & 27 deletions pkg/kv/kvserver/replica_rankings_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -218,15 +218,13 @@ func TestWriteLoadStatsAccounting(t *testing.T) {
ReplicationMode: base.ReplicationManual,
}
args.ServerArgs.Knobs.Store = &StoreTestingKnobs{DisableCanAckBeforeApplication: true}
tc := serverutils.StartCluster(t, 1, args)
tc := serverutils.StartCluster(t, 2, args)
defer tc.Stopper().Stop(ctx)

const epsilonAllowed = 5

defer tc.Stopper().Stop(ctx)
ts := tc.Server(0)
db := ts.DB()
conn := tc.ServerConn(0)
sqlDB := sqlutils.MakeSQLRunner(conn)
db := tc.Server(0).DB()
sqlDB := sqlutils.MakeSQLRunner(tc.ServerConn(0))

writeSize := float64(9)

Expand All @@ -246,52 +244,74 @@ func TestWriteLoadStatsAccounting(t *testing.T) {
{1234, 1234, 1234, 0, 1234 * writeSize, 0},
}

store, err := ts.GetStores().(*Stores).GetStore(ts.GetFirstStoreID())
require.NoError(t, err)
tc.AddVotersOrFatal(t, scratchKey, tc.Target(1))

repl := store.LookupReplica(roachpb.RKey(scratchKey))
require.NotNil(t, repl)
s1, err := tc.Server(0).GetStores().(*Stores).GetStore(tc.Server(0).GetFirstStoreID())
require.NoError(t, err)
lhRepl := s1.LookupReplica(roachpb.RKey(scratchKey))
require.NotNil(t, lhRepl)
s2, err := tc.Server(1).GetStores().(*Stores).GetStore(tc.Server(1).GetFirstStoreID())
require.NoError(t, err)
followerRepl := s2.LookupReplica(roachpb.RKey(scratchKey))
require.NotNil(t, followerRepl)

// Disable the consistency checker, to avoid interleaving requests
// artificially inflating measurement due to consistency checking.
// artificially inflating measurement due to consistency checking. Also,
// disable the mvcc gc queue and raft log queue to avoid them issuing
// interleaving requests as well.
sqlDB.Exec(t, `SET CLUSTER SETTING server.consistency_check.interval = '0'`)
sqlDB.Exec(t, `SET CLUSTER SETTING kv.range_split.by_load.enabled = false`)
sqlDB.Exec(t, `SET CLUSTER SETTING kv.mvcc_gc_queue.enabled = false`)
sqlDB.Exec(t, `SET CLUSTER SETTING kv.raft_log_queue.enabled = false`)

for _, testCase := range testCases {
// Reset the request counts to 0 before sending to clear previous requests.
repl.loadStats.Reset()
lhRepl.loadStats.Reset()
followerRepl.loadStats.Reset()

requestsBefore := repl.loadStats.TestingGetSum(load.Requests)
writesBefore := repl.loadStats.TestingGetSum(load.WriteKeys)
readsBefore := repl.loadStats.TestingGetSum(load.ReadKeys)
readBytesBefore := repl.loadStats.TestingGetSum(load.ReadBytes)
writeBytesBefore := repl.loadStats.TestingGetSum(load.WriteBytes)
requestsBefore := lhRepl.loadStats.TestingGetSum(load.Requests)
readsBefore := lhRepl.loadStats.TestingGetSum(load.ReadKeys)
lhWritesBefore := lhRepl.loadStats.TestingGetSum(load.WriteKeys)
readBytesBefore := lhRepl.loadStats.TestingGetSum(load.ReadBytes)
lhWriteBytesBefore := lhRepl.loadStats.TestingGetSum(load.WriteBytes)
followerWriteBytesBefore := followerRepl.loadStats.TestingGetSum(load.WriteBytes)

for i := 0; i < testCase.writes; i++ {
_, pErr := db.Inc(ctx, scratchKey, 1)
require.Nil(t, pErr)
}
require.Equal(t, 0.0, requestsBefore)
require.Equal(t, 0.0, writesBefore)
require.Equal(t, 0.0, lhWritesBefore)
require.Equal(t, 0.0, readsBefore)
require.Equal(t, 0.0, writeBytesBefore)
require.Equal(t, 0.0, lhWriteBytesBefore)
require.Equal(t, 0.0, readBytesBefore)

requestsAfter := repl.loadStats.TestingGetSum(load.Requests)
writesAfter := repl.loadStats.TestingGetSum(load.WriteKeys)
readsAfter := repl.loadStats.TestingGetSum(load.ReadKeys)
readBytesAfter := repl.loadStats.TestingGetSum(load.ReadBytes)
writeBytesAfter := repl.loadStats.TestingGetSum(load.WriteBytes)
require.Equal(t, 0.0, followerWriteBytesBefore)

require.NoError(t, waitForApplication(
ctx,
s1.cfg.NodeDialer,
lhRepl.GetRangeID(),
lhRepl.Desc().Replicas().Descriptors(),
lhRepl.GetLeaseAppliedIndex(),
))

requestsAfter := lhRepl.loadStats.TestingGetSum(load.Requests)
lhWritesAfter := lhRepl.loadStats.TestingGetSum(load.WriteKeys)
readsAfter := lhRepl.loadStats.TestingGetSum(load.ReadKeys)
readBytesAfter := lhRepl.loadStats.TestingGetSum(load.ReadBytes)
lhWriteBytesAfter := lhRepl.loadStats.TestingGetSum(load.WriteBytes)
followerWriteBytesAfter := followerRepl.loadStats.TestingGetSum(load.WriteBytes)

assertGreaterThanInDelta(t, testCase.expectedRQPS, requestsAfter, epsilonAllowed)
assertGreaterThanInDelta(t, testCase.expectedWPS, writesAfter, epsilonAllowed)
assertGreaterThanInDelta(t, testCase.expectedWPS, lhWritesAfter, epsilonAllowed)
assertGreaterThanInDelta(t, testCase.expectedRPS, readsAfter, epsilonAllowed)
assertGreaterThanInDelta(t, testCase.expectedRBPS, readBytesAfter, epsilonAllowed)
// NB: We assert that the written bytes is greater than the write
// batch request size. However the size multiplication factor,
// varies between 3 and 5 so we instead assert that it is greater
// than the logical bytes.
require.GreaterOrEqual(t, writeBytesAfter, testCase.expectedWBPS)
require.GreaterOrEqual(t, lhWriteBytesAfter, testCase.expectedWBPS)
require.GreaterOrEqual(t, followerWriteBytesAfter, testCase.expectedWBPS)
}
}

Expand Down