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.

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

kvcoord: add proxy tracing test #124043

Merged
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
109 changes: 109 additions & 0 deletions pkg/kv/kvclient/kvcoord/dist_sender_server_test.go
Original file line number Diff line number Diff line change
Expand Up @@ -4803,3 +4803,112 @@ func TestPartialPartition(t *testing.T) {
})
}
}

// TestProxyTracing asserts when enabling a partial partition between two
// nodes, the request is proxied via a third node and that tracing captures
// the relevant event.
func TestProxyTracing(t *testing.T) {
defer leaktest.AfterTest(t)()
defer log.Scope(t).Close(t)
ctx := context.Background()

const numServers = 3
const numRanges = 3
st := cluster.MakeTestingClusterSettings()
kvserver.ExpirationLeasesOnly.Override(ctx, &st.SV, true)
kvserver.RangefeedEnabled.Override(ctx, &st.SV, true)
kvserver.RangeFeedRefreshInterval.Override(ctx, &st.SV, 10*time.Millisecond)
closedts.TargetDuration.Override(ctx, &st.SV, 10*time.Millisecond)
closedts.SideTransportCloseInterval.Override(ctx, &st.SV, 10*time.Millisecond)

var p rpc.Partitioner
tc := testcluster.StartTestCluster(t, numServers, base.TestClusterArgs{
ServerArgsPerNode: func() map[int]base.TestServerArgs {
perNode := make(map[int]base.TestServerArgs)
for i := 0; i < numServers; i++ {
ctk := rpc.ContextTestingKnobs{}
// Partition between n1 and n3.
p.RegisterTestingKnobs(roachpb.NodeID(i+1), [][2]roachpb.NodeID{{1, 3}}, &ctk)
perNode[i] = base.TestServerArgs{
Settings: st,
Knobs: base.TestingKnobs{
Server: &server.TestingKnobs{
ContextTestingKnobs: ctk,
},
},
}
}
return perNode
}(),
})
defer tc.Stopper().Stop(ctx)

// Set up the mapping after the nodes have started and we have their
// addresses.
for i := 0; i < numServers; i++ {
g := tc.Servers[i].StorageLayer().GossipI().(*gossip.Gossip)
addr := g.GetNodeAddr().String()
nodeID := g.NodeID.Get()
p.RegisterNodeAddr(addr, nodeID)
}

conn := tc.Conns[0]

// Create a table and pin the leaseholder replicas to n3. The partition
// between n1 and n3 will lead to re-routing via n2, which we expect captured
// in the trace.
_, err := conn.Exec("CREATE TABLE t (i INT)")
require.NoError(t, err)
_, err = conn.Exec("ALTER TABLE t CONFIGURE ZONE USING num_replicas=3, lease_preferences='[[+dc=dc3]]', constraints='[]'")
require.NoError(t, err)
_, err = conn.Exec(
fmt.Sprintf("INSERT INTO t(i) select generate_series(1,%d)", numRanges-1))
require.NoError(t, err)
_, err = conn.Exec("ALTER TABLE t SPLIT AT SELECT i FROM t")
require.NoError(t, err)
require.NoError(t, tc.WaitForFullReplication())

leaseCount := func(node int) int {
var count int
err := conn.QueryRow(fmt.Sprintf(
"SELECT count(*) FROM [SHOW RANGES FROM TABLE t WITH DETAILS] WHERE lease_holder = %d", node),
).Scan(&count)
require.NoError(t, err)
return count
}

checkLeaseCount := func(node, expectedLeaseCount int) error {
if count := leaseCount(node); count != expectedLeaseCount {
require.NoError(t, tc.GetFirstStoreFromServer(t, 0).
ForceLeaseQueueProcess())
return errors.Errorf("expected %d leases on node %d, found %d",
expectedLeaseCount, node, count)
}
return nil
}

// Wait until the leaseholder for the test table ranges are on n3.
testutils.SucceedsSoon(t, func() error {
return checkLeaseCount(3, numRanges)
})

p.EnablePartition(true)

// TODO(kvoli,andrewbaptist): Currently this may return an error on first
// attempt with 'pq: partitioned from 127.0.0.1:XXXXX, n3'. Remove the retry
// loop below once this is resolved.
testutils.SucceedsSoon(t, func() error {
_, err = conn.Exec("SET TRACING = on; SELECT FROM t where i = 987654321; SET TRACING = off")
return err
})

// Expect the "proxy request complete" message to be in the trace and that it
// comes from the proxy node n2.
row := conn.QueryRowContext(ctx, "SELECT message, tag, location "+
"FROM [SHOW TRACE FOR SESSION] "+
"WHERE message LIKE '%proxy request complete%'"+
"AND location LIKE '%server/node%'",
"AND tag LIKE '%n2%'",
)
require.NotNil(t, row)
}