SOLR-13408: Cannot start/stop DaemonStream repeatedly, other API improvements

This commit is contained in:
erick 2019-04-17 16:02:06 -07:00
parent 3a6f2f7543
commit a9771a5849
5 changed files with 421 additions and 72 deletions

View File

@ -202,12 +202,14 @@ Bug Fixes
* SOLR-13336: add maxBooleanClauses (default to 1024) setting to solr.xml, reverting previous effective * SOLR-13336: add maxBooleanClauses (default to 1024) setting to solr.xml, reverting previous effective
value of Integer.MAX_VALUE-1, to restrict risk of pathalogical query expansion. (hossman) value of Integer.MAX_VALUE-1, to restrict risk of pathalogical query expansion. (hossman)
* SOLR-13386: OverseerTaskQueue#remove should not throw an exception when no node exists after an exists * SOLR-13386: OverseerTaskQueue#remove should not throw an exception when no node exists after an exists
check and the Overseer work loop should not allow free spinning the loop when it hits a KeeperException. check and the Overseer work loop should not allow free spinning the loop when it hits a KeeperException.
(Mark Miller, Fernandez-Lobbe, Mike Drob) (Mark Miller, Fernandez-Lobbe, Mike Drob)
* SOLR-12371: Editing authorization config via REST API now works in standalone mode (janhoy) * SOLR-12371: Editing authorization config via REST API now works in standalone mode (janhoy)
* SOLR-13408: Cannot start/stop DaemonStream repeatedly, other API improvements (Erick Erickson)
Improvements Improvements
---------------------- ----------------------

View File

@ -24,6 +24,7 @@ import java.util.Collections;
import java.util.HashMap; import java.util.HashMap;
import java.util.Iterator; import java.util.Iterator;
import java.util.List; import java.util.List;
import java.util.Locale;
import java.util.Map; import java.util.Map;
import java.util.concurrent.ConcurrentHashMap; import java.util.concurrent.ConcurrentHashMap;
import java.util.concurrent.ConcurrentMap; import java.util.concurrent.ConcurrentMap;
@ -192,35 +193,46 @@ public class StreamHandler extends RequestHandlerBase implements SolrCoreAware,
} }
private void handleAdmin(SolrQueryRequest req, SolrQueryResponse rsp, SolrParams params) { private void handleAdmin(SolrQueryRequest req, SolrQueryResponse rsp, SolrParams params) {
String action = params.get("action"); String action = params.get("action").toLowerCase(Locale.ROOT).trim();
if ("stop".equalsIgnoreCase(action)) {
String id = params.get(ID); if ("list".equals(action)) {
DaemonStream d = daemons.get(id); Collection<DaemonStream> vals = daemons.values();
if (d != null) { rsp.add("result-set", new DaemonCollectionStream(vals));
return;
}
String id = params.get(ID);
DaemonStream d = daemons.get(id);
if (d == null) {
rsp.add("result-set", new DaemonResponseStream("Deamon:" + id + " not found on " + coreName));
return;
}
switch (action) {
case "stop":
d.close(); d.close();
rsp.add("result-set", new DaemonResponseStream("Deamon:" + id + " stopped on " + coreName)); rsp.add("result-set", new DaemonResponseStream("Deamon:" + id + " stopped on " + coreName));
} else { break;
rsp.add("result-set", new DaemonResponseStream("Deamon:" + id + " not found on " + coreName));
} case "start":
} else { try {
if ("start".equalsIgnoreCase(action)) { d.open();
String id = params.get(ID); } catch (IOException e) {
DaemonStream d = daemons.get(id); rsp.add("result-set", new DaemonResponseStream("Daemon: " + id + " error: " + e.getMessage()));
d.open();
rsp.add("result-set", new DaemonResponseStream("Deamon:" + id + " started on " + coreName));
} else if ("list".equalsIgnoreCase(action)) {
Collection<DaemonStream> vals = daemons.values();
rsp.add("result-set", new DaemonCollectionStream(vals));
} else if ("kill".equalsIgnoreCase(action)) {
String id = params.get("id");
DaemonStream d = daemons.remove(id);
if (d != null) {
d.close();
rsp.add("result-set", new DaemonResponseStream("Deamon:" + id + " killed on " + coreName));
} else {
rsp.add("result-set", new DaemonResponseStream("Deamon:" + id + " not found on " + coreName));
} }
} rsp.add("result-set", new DaemonResponseStream("Deamon:" + id + " started on " + coreName));
break;
case "kill":
daemons.remove(id);
d.close(); // we already found it in the daemons list, so we don't need to verify we removed it.
rsp.add("result-set", new DaemonResponseStream("Deamon:" + id + " killed on " + coreName));
break;
default:
rsp.add("result-set", new DaemonResponseStream("Deamon:" + id + " action '"
+ action + "' not recognized on " + coreName));
break;
} }
} }

View File

@ -0,0 +1,319 @@
/*
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership.
* The ASF licenses this file to You under the Apache License, Version 2.0
* (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.solr.handler.admin;
import java.io.IOException;
import java.util.ArrayList;
import java.util.Collections;
import java.util.List;
import java.util.concurrent.TimeUnit;
import org.apache.solr.SolrTestCaseJ4;
import org.apache.solr.client.solrj.SolrClient;
import org.apache.solr.client.solrj.SolrServerException;
import org.apache.solr.client.solrj.io.Tuple;
import org.apache.solr.client.solrj.io.stream.SolrStream;
import org.apache.solr.client.solrj.io.stream.TupleStream;
import org.apache.solr.client.solrj.request.CollectionAdminRequest;
import org.apache.solr.client.solrj.response.QueryResponse;
import org.apache.solr.cloud.MiniSolrCloudCluster;
import org.apache.solr.common.params.SolrParams;
import org.apache.solr.common.util.TimeSource;
import org.apache.solr.handler.TestSQLHandler;
import org.apache.solr.util.TimeOut;
import org.junit.After;
import org.junit.Before;
import org.junit.Test;
public class DaemonStreamApiTest extends SolrTestCaseJ4 {
private MiniSolrCloudCluster cluster;
private static final String SOURCE_COLL = "sourceColl";
private static final String TARGET_COLL = "targetColl";
private static final String CHECKPOINT_COLL = "checkpointColl";
private static final String DAEMON_ROOT = "daemon";
private static final String CONF_NAME = "conf";
private static final String DAEMON_OP = "DaemonOp";
// We want 2-5 daemons. Choose one of them to start/stop/kill to catch any off-by-one or other bookeeping errors.
final int numDaemons = random().nextInt(3) + 2;
String daemonOfInterest;
List<String> daemonNames = new ArrayList<>();
private String url;
@Override
@Before
public void setUp() throws Exception {
super.setUp();
cluster = new MiniSolrCloudCluster(1, createTempDir(), buildJettyConfig("/solr"));
url = cluster.getJettySolrRunners().get(0).getBaseUrl().toString() + "/" + CHECKPOINT_COLL;
cluster.uploadConfigSet(configset("cloud-minimal"), CONF_NAME);
// create a single shard, single replica collection. This is necessary until SOLR-13245 since the commands
// don't look in all replicas.
CollectionAdminRequest.createCollection(SOURCE_COLL, CONF_NAME, 1, 1)
.setMaxShardsPerNode(1)
.process(cluster.getSolrClient());
CollectionAdminRequest.createCollection(TARGET_COLL, CONF_NAME, 1, 1)
.setMaxShardsPerNode(1)
.process(cluster.getSolrClient());
CollectionAdminRequest.createCollection(CHECKPOINT_COLL, CONF_NAME, 1, 1)
.setMaxShardsPerNode(1)
.process(cluster.getSolrClient());
for (int idx = 0; idx < numDaemons; ++idx) {
String name = DAEMON_ROOT + idx;
daemonNames.add(name);
}
daemonOfInterest = daemonNames.get(random().nextInt(numDaemons));
}
@Override
@After
public void tearDown() throws Exception {
cluster.shutdown();
super.tearDown();
}
@Test
public void testAPIs() throws IOException, SolrServerException, InterruptedException {
checkCmdsNoDaemon(daemonOfInterest); // test no daemon defined
// Now create all our daemons.
for (String name : daemonNames) {
createDaemon(DAEMON_DEF.replace("DAEMON_NAME", name), name);
}
List<Tuple> tuples = getTuples(TestSQLHandler.mapParams("qt", "/stream", "action", "list"));
assertEquals("Should have all daemons listed", numDaemons, tuples.size());
for (int idx = 0; idx < numDaemons; ++idx) {
assertEquals("Daemon should be running ", tuples.get(idx).getString("id"), daemonNames.get(idx));
}
// Are all the daemons in a good state?
for (String daemon : daemonNames) {
checkAlive(daemon);
}
// We shouldn't be able to open a daemon twice without closing., leads to thread leeks.
Tuple tupleOfInterest = getTupleOfInterest(TestSQLHandler.mapParams("qt", "/stream", "action", "start", "id", daemonOfInterest)
, DAEMON_OP);
assertTrue("Should not open twice without closing",
tupleOfInterest.getString(DAEMON_OP).contains("There is already an open daemon named"));
// Try stopping and check return.
tupleOfInterest = getTupleOfInterest(TestSQLHandler.mapParams("qt", "/stream", "action", "stop", "id", daemonOfInterest),
DAEMON_OP);
assertTrue("Should have been able to stop the daemon",
tupleOfInterest.getString(DAEMON_OP).contains(daemonOfInterest + " stopped"));
checkStopped();
// Are all the daemons alive? NOTE: a stopped daemon is still there, but in a TERMINATED state
for (String daemon : daemonNames) {
if (daemon.equals(daemonOfInterest) == false) {
checkAlive(daemon);
}
}
// Try starting and check return.
tupleOfInterest = getTupleOfInterest(TestSQLHandler.mapParams("qt", "/stream", "action", "start", "id", daemonOfInterest),
DAEMON_OP);
assertTrue("Should have been able to start the daemon",
tupleOfInterest.getString(DAEMON_OP).contains(daemonOfInterest + " started"));
// Are all the daemons alive?
for (String daemon : daemonNames) {
checkAlive(daemon);
}
// Try killing a daemon, it should be removed from lists.
tupleOfInterest = getTupleOfInterest(TestSQLHandler.mapParams("qt", "/stream", "action", "kill", "id", daemonOfInterest),
DAEMON_OP);
assertTrue("Daemon should have been killed",
tupleOfInterest.getString(DAEMON_OP).contains(daemonOfInterest + " killed"));
// Loop for a bit, waiting for the daemon to be removed from the list of possible entries.
checkDaemonKilled(daemonOfInterest);
// Should not be able to start a killed daemon
tupleOfInterest = getTupleOfInterest(TestSQLHandler.mapParams("qt", "/stream", "action", "start", "id", daemonOfInterest),
DAEMON_OP);
assertTrue("Daemon should not be found",
tupleOfInterest.getString(DAEMON_OP).contains(daemonOfInterest + " not found"));
// Should not be able to sop a killed daemon
tupleOfInterest = getTupleOfInterest(TestSQLHandler.mapParams("qt", "/stream", "action", "stop", "id", daemonOfInterest),
DAEMON_OP);
assertTrue("Daemon should not be found",
tupleOfInterest.getString(DAEMON_OP).contains(daemonOfInterest + " not found"));
// Should not be able to kill a killed daemon
tupleOfInterest = getTupleOfInterest(TestSQLHandler.mapParams("qt", "/stream", "action", "kill", "id", daemonOfInterest),
DAEMON_OP);
assertTrue("Daemon should not be found",
tupleOfInterest.getString(DAEMON_OP).contains(daemonOfInterest + " not found"));
// Let's bring the killed daemon back and see if it returns in our lists. Use the method that loops a bit to check
// in case there's a delay.
createDaemon(DAEMON_DEF.replace("DAEMON_NAME", daemonOfInterest), daemonOfInterest);
checkAlive(daemonOfInterest);
// Now kill them all so the threads disappear.
for (String daemon : daemonNames) {
getTuples(TestSQLHandler.mapParams("qt", "/stream", "action", "kill", "id", daemon));
checkDaemonKilled(daemon);
}
}
// There can be some delay while threads stabilize, so we need to loop;
private void checkAlive(String daemonName) throws InterruptedException, IOException {
TimeOut timeout = new TimeOut(10, TimeUnit.SECONDS, TimeSource.NANO_TIME);
while (timeout.hasTimedOut() == false) {
Tuple tuple = getTupleOfInterest(TestSQLHandler.mapParams("qt", "/stream", "action", "list"),
daemonName);
String state = tuple.getString("state");
if (state.equals("RUNNABLE") || state.equals("WAITING") || state.equals("TIMED_WAITING")) {
return;
}
TimeUnit.MILLISECONDS.sleep(100);
}
fail("State for daemon '" + daemonName + "' did not become RUNNABLE, WAITING or TIMED_WAITING in 10 seconds");
}
// There can be some delay while threads stabilize, so we need to loop. Evenutally, the statu of a stopped
// thread should be "TERMINATED"
private void checkStopped() throws InterruptedException, IOException {
TimeOut timeout = new TimeOut(10, TimeUnit.SECONDS, TimeSource.NANO_TIME);
while (timeout.hasTimedOut() == false) {
Tuple tuple = getTupleOfInterest(TestSQLHandler.mapParams("qt", "/stream", "action", "list"),
daemonOfInterest);
if (tuple.getString("state").equals("TERMINATED")) {
return;
}
TimeUnit.MILLISECONDS.sleep(100);
}
fail("State for daemon '" + daemonOfInterest + "' did not become TERMINATED in 10 seconds");
}
private void checkDaemonKilled(String daemon) throws IOException, InterruptedException {
TimeOut timeout = new TimeOut(10, TimeUnit.SECONDS, TimeSource.NANO_TIME);
while (timeout.hasTimedOut() == false) {
List<Tuple> tuples = getTuples(TestSQLHandler.mapParams("qt", "/stream", "action", "list"));
Boolean foundIt = false;
for (Tuple tuple : tuples) {
if (tuple.get("id").equals(daemon)) {
foundIt = true;
}
}
if (foundIt == false) return;
TimeUnit.MILLISECONDS.sleep(100);
}
fail("'" + daemonOfInterest + "' did not disappear in 10 seconds");
}
private void createDaemon(String daemonDef, String errMsg) throws IOException, SolrServerException {
SolrClient client = cluster.getSolrClient();
// create a daemon
QueryResponse resp = client.query(CHECKPOINT_COLL, TestSQLHandler.mapParams("expr", daemonDef, "qt", "/stream"));
assertEquals(errMsg, 0, resp.getStatus());
// This should close and replace the current daemon and NOT leak threads.
resp = client.query(CHECKPOINT_COLL, TestSQLHandler.mapParams("expr", daemonDef, "qt", "/stream"));
assertEquals(errMsg, 0, resp.getStatus());
}
private void checkCmdsNoDaemon(String daemonName) throws IOException {
List<Tuple> tuples = getTuples(TestSQLHandler.mapParams("qt", "/stream", "action", "list"));
assertEquals("List should be empty", 0, tuples.size());
Tuple tupleOfInterest = getTupleOfInterest(TestSQLHandler.mapParams("qt", "/stream", "action", "start", "id", daemonName),
"DaemonOp");
assertTrue("Start for daemon should not be found", tupleOfInterest.getString("DaemonOp").contains("not found on"));
tupleOfInterest = getTupleOfInterest(TestSQLHandler.mapParams("qt", "/stream", "action", "stop", "id", daemonName),
"DaemonOp");
assertTrue("Stop for daemon should not be found", tupleOfInterest.getString("DaemonOp").contains("not found on"));
tupleOfInterest = getTupleOfInterest(TestSQLHandler.mapParams("qt", "/stream", "action", "kill", "id", daemonName),
"DaemonOp");
assertTrue("Kill for daemon should not be found", tupleOfInterest.getString("DaemonOp").contains("not found on"));
}
// It's _really_ useful to have the tuples sorted....
private List<Tuple> getTuples(final SolrParams params) throws IOException {
return getTuples(params, null);
}
private List<Tuple> getTuples(final SolrParams params, String ofInterest) throws IOException {
//log.info("Tuples from params: {}", params);
TupleStream tupleStream = new SolrStream(url, params);
tupleStream.open();
List<Tuple> tuples = new ArrayList<>();
for (; ; ) {
Tuple t = tupleStream.read();
//log.info(" ... {}", t.fields);
if (t.EOF) {
break;
} else if (ofInterest == null || t.getString("id").equals(ofInterest) || t.getString(ofInterest).equals("null") == false) {
// a failed return is a bit different, the onlyh key is DaemonOp
tuples.add(t);
}
}
tupleStream.close();
Collections.sort(tuples, (o1, o2) -> (o1.getString("id").compareTo(o2.getString("id"))));
return tuples;
}
private Tuple getTupleOfInterest(final SolrParams params, String ofInterest) throws IOException {
List<Tuple> tuples = getTuples(params, ofInterest);
if (tuples.size() != 1) {
fail("Should have found a tuple for tuple of interest: " + ofInterest);
}
return tuples.get(0);
}
private static String DAEMON_DEF =
" daemon(id=\"DAEMON_NAME\"," +
" runInterval=\"1000\"," +
" terminate=\"false\"," +
" update(targetColl," +
" batchSize=100," +
" topic(checkpointColl," +
" sourceColl," +
" q=\"*:*\"," +
" fl=\"id\"," +
" id=\"topic1\"," +
" initialCheckpoint=0)" +
"))";
}

View File

@ -195,7 +195,12 @@ public class DaemonStream extends TupleStream implements Expressible {
return id; return id;
} }
public void open() { public void open() throws IOException {
if (this.streamRunner != null && this.closed == false) {
log.error("There is already a running daemon named '{}', no action taken", id);
throw new IOException("There is already an open daemon named '" + id + "', no action taken.");
}
this.closed = false;
this.streamRunner = new StreamRunner(runInterval, id); this.streamRunner = new StreamRunner(runInterval, id);
this.streamRunner.start(); this.streamRunner.start();
} }

View File

@ -1739,58 +1739,69 @@ public void testParallelRankStream() throws Exception {
daemonStream.setStreamContext(context); daemonStream.setStreamContext(context);
daemonStream.open(); daemonStream.open();
CheckDaemonStream(context, daemonStream);
// Wait for the checkpoint // We should get an error if we try to open an already-open stream.
JettySolrRunner jetty = cluster.getJettySolrRunners().get(0); final IOException ex = expectThrows(IOException.class, () -> {
daemonStream.open();
});
assertEquals("Should have an intelligible exception message", ex.getMessage(), "There is already an open daemon named 'daemon1', no action taken.");
daemonStream.close();
// We should be able to close then re-open the stream, then close it again, see SOLR-13408
SolrParams sParams1 = mapParams("qt", "/get", "ids", "50000000", "fl", "id"); daemonStream.open();
int count = 0; CheckDaemonStream(context, daemonStream);
while (count == 0) {
SolrStream solrStream = new SolrStream(jetty.getBaseUrl().toString() + "/" + COLLECTIONORALIAS, sParams1);
solrStream.setStreamContext(context);
List<Tuple> tuples = getTuples(solrStream);
count = tuples.size();
if (count > 0) {
Tuple t = tuples.get(0);
assertTrue(t.getLong("id") == 50000000);
} else {
System.out.println("###### Waiting for checkpoint #######:" + count);
}
}
new UpdateRequest()
.add(id, "0", "a_s", "hello0", "a_i", "0", "a_f", "1")
.add(id, "2", "a_s", "hello0", "a_i", "2", "a_f", "2")
.add(id, "3", "a_s", "hello0", "a_i", "3", "a_f", "3")
.add(id, "4", "a_s", "hello0", "a_i", "4", "a_f", "4")
.add(id, "1", "a_s", "hello0", "a_i", "1", "a_f", "5")
.commit(cluster.getSolrClient(), COLLECTIONORALIAS);
for (int i = 0; i < 5; i++) {
daemonStream.read();
}
new UpdateRequest()
.add(id, "5", "a_s", "hello0", "a_i", "4", "a_f", "4")
.add(id, "6", "a_s", "hello0", "a_i", "4", "a_f", "4")
.commit(cluster.getSolrClient(), COLLECTIONORALIAS);
for (int i = 0; i < 2; i++) {
daemonStream.read();
}
daemonStream.shutdown();
Tuple tuple = daemonStream.read();
assertTrue(tuple.EOF);
daemonStream.close(); daemonStream.close();
} finally { } finally {
cache.close(); cache.close();
} }
}
private void CheckDaemonStream(StreamContext context, DaemonStream daemonStream) throws IOException, SolrServerException {
// Wait for the checkpoint
JettySolrRunner jetty = cluster.getJettySolrRunners().get(0);
SolrParams sParams1 = mapParams("qt", "/get", "ids", "50000000", "fl", "id");
int count = 0;
while (count == 0) {
SolrStream solrStream = new SolrStream(jetty.getBaseUrl().toString() + "/" + COLLECTIONORALIAS, sParams1);
solrStream.setStreamContext(context);
List<Tuple> tuples = getTuples(solrStream);
count = tuples.size();
if (count > 0) {
Tuple t = tuples.get(0);
assertTrue(t.getLong("id") == 50000000);
} else {
System.out.println("###### Waiting for checkpoint #######:" + count);
}
}
new UpdateRequest()
.add(id, "0", "a_s", "hello0", "a_i", "0", "a_f", "1")
.add(id, "2", "a_s", "hello0", "a_i", "2", "a_f", "2")
.add(id, "3", "a_s", "hello0", "a_i", "3", "a_f", "3")
.add(id, "4", "a_s", "hello0", "a_i", "4", "a_f", "4")
.add(id, "1", "a_s", "hello0", "a_i", "1", "a_f", "5")
.commit(cluster.getSolrClient(), COLLECTIONORALIAS);
for (int i = 0; i < 5; i++) {
daemonStream.read();
}
new UpdateRequest()
.add(id, "5", "a_s", "hello0", "a_i", "4", "a_f", "4")
.add(id, "6", "a_s", "hello0", "a_i", "4", "a_f", "4")
.commit(cluster.getSolrClient(), COLLECTIONORALIAS);
for (int i = 0; i < 2; i++) {
daemonStream.read();
}
daemonStream.shutdown();
Tuple tuple = daemonStream.read();
assertTrue(tuple.EOF);
} }
@Test @Test