// make sure that log isn't needlessly replayed after a clean close
@Test
public void testCleanShutdown() throws Exception {
DirectUpdateHandler2.commitOnClose = true;
final Semaphore logReplay = new Semaphore(0);
final Semaphore logReplayFinish = new Semaphore(0);
UpdateLog.testing_logReplayHook =
new Runnable() {
@Override
public void run() {
try {
assertTrue(logReplay.tryAcquire(timeout, TimeUnit.SECONDS));
} catch (Exception e) {
throw new RuntimeException(e);
}
}
};
UpdateLog.testing_logReplayFinishHook =
new Runnable() {
@Override
public void run() {
logReplayFinish.release();
}
};
SolrQueryRequest req = req();
UpdateHandler uhandler = req.getCore().getUpdateHandler();
UpdateLog ulog = uhandler.getUpdateLog();
try {
clearIndex();
assertU(commit());
assertU(adoc("id", "E1", "val_i", "1"));
assertU(adoc("id", "E2", "val_i", "1"));
// set to a high enough number so this test won't hang on a bug
logReplay.release(10);
h.close();
createCore();
// make sure the docs got committed
assertJQ(req("q", "*:*"), "/response/numFound==2");
// make sure no replay happened
assertEquals(10, logReplay.availablePermits());
} finally {
DirectUpdateHandler2.commitOnClose = true;
UpdateLog.testing_logReplayHook = null;
UpdateLog.testing_logReplayFinishHook = null;
req().close();
}
}
@Override
public void write(NodeInfo node, String uri) {
UpdateHandler updateHandler = core.getUpdateHandler();
// Create a version of the document for saving to the transaction log,
// or for cloud update via HTTP
SolrInputDocument solrDoc = new SolrInputDocument();
solrDoc.addField(uriFieldName, uri);
if (isCloud()) {
// TODO: write as binary, but we need to enable the binary update request writer for this
// TinyBinary tinybin = new TinyBinary(((TinyNodeImpl)node).getTree());
// solrDoc.addField(xmlFieldName, tinybin.getByteBuffer().array());
Serializer serializer = xqueryComponent.solrIndexConfig.checkoutSerializer();
try {
String xmlString = serializer.serializeNodeToString(new XdmNode(node));
solrDoc.addField(xmlFieldName, xmlString);
} catch (SaxonApiException e) {
throw new LuxException(e);
} finally {
xqueryComponent.solrIndexConfig.returnSerializer(serializer);
}
// TODO -- if we can determine this doc only gets added locally??
// solrDoc.addField(xmlFieldName, node);
} else if (updateHandler.getUpdateLog() != null) {
if (node instanceof TinyNodeImpl) {
TinyBinary tinybin = new TinyBinary(((TinyNodeImpl) node).getTree());
solrDoc.addField(xmlFieldName, tinybin.getByteBuffer());
} else {
String xml = node.toString();
solrDoc.addField(xmlFieldName, xml);
}
}
if (isCloud()) {
writeToCloud(solrDoc, uri);
} else {
writeLocal(solrDoc, node, uri);
}
}
public DistributedUpdateProcessor(
SolrQueryRequest req, SolrQueryResponse rsp, UpdateRequestProcessor next) {
super(next);
this.rsp = rsp;
this.next = next;
this.idField = req.getSchema().getUniqueKeyField();
// version init
this.updateHandler = req.getCore().getUpdateHandler();
this.ulog = updateHandler.getUpdateLog();
this.vinfo = ulog == null ? null : ulog.getVersionInfo();
versionsStored = this.vinfo != null && this.vinfo.getVersionField() != null;
returnVersions = req.getParams().getBool(UpdateParams.VERSIONS, false);
// TODO: better way to get the response, or pass back info to it?
SolrRequestInfo reqInfo = returnVersions ? SolrRequestInfo.getRequestInfo() : null;
this.req = req;
CoreDescriptor coreDesc = req.getCore().getCoreDescriptor();
this.zkEnabled = coreDesc.getCoreContainer().isZooKeeperAware();
zkController = req.getCore().getCoreDescriptor().getCoreContainer().getZkController();
if (zkEnabled) {
numNodes = zkController.getZkStateReader().getClusterState().getLiveNodes().size();
cmdDistrib =
new SolrCmdDistributor(
numNodes, coreDesc.getCoreContainer().getZkController().getCmdDistribExecutor());
}
// this.rsp = reqInfo != null ? reqInfo.getRsp() : null;
cloudDesc = coreDesc.getCloudDescriptor();
if (cloudDesc != null) {
collection = cloudDesc.getCollectionName();
}
}
@Test
public void testBufferingFlags() throws Exception {
DirectUpdateHandler2.commitOnClose = false;
final Semaphore logReplayFinish = new Semaphore(0);
UpdateLog.testing_logReplayFinishHook =
new Runnable() {
@Override
public void run() {
logReplayFinish.release();
}
};
SolrQueryRequest req = req();
UpdateHandler uhandler = req.getCore().getUpdateHandler();
UpdateLog ulog = uhandler.getUpdateLog();
try {
clearIndex();
assertU(commit());
assertEquals(UpdateLog.State.ACTIVE, ulog.getState());
ulog.bufferUpdates();
// simulate updates from a leader
updateJ(
jsonAdd(sdoc("id", "Q1", "_version_", "101")), params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
updateJ(
jsonAdd(sdoc("id", "Q2", "_version_", "102")), params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
updateJ(
jsonAdd(sdoc("id", "Q3", "_version_", "103")), params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
assertEquals(UpdateLog.State.BUFFERING, ulog.getState());
req.close();
h.close();
createCore();
req = req();
uhandler = req.getCore().getUpdateHandler();
ulog = uhandler.getUpdateLog();
logReplayFinish.acquire(); // wait for replay to finish
assertTrue(
(ulog.getStartingOperation() & UpdateLog.FLAG_GAP)
!= 0); // since we died while buffering, we should see this last
//
// Try again to ensure that the previous log replay didn't wipe out our flags
//
req.close();
h.close();
createCore();
req = req();
uhandler = req.getCore().getUpdateHandler();
ulog = uhandler.getUpdateLog();
assertTrue((ulog.getStartingOperation() & UpdateLog.FLAG_GAP) != 0);
// now do some normal non-buffered adds
updateJ(
jsonAdd(sdoc("id", "Q4", "_version_", "114")), params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
updateJ(
jsonAdd(sdoc("id", "Q5", "_version_", "115")), params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
updateJ(
jsonAdd(sdoc("id", "Q6", "_version_", "116")), params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
assertU(commit());
req.close();
h.close();
createCore();
req = req();
uhandler = req.getCore().getUpdateHandler();
ulog = uhandler.getUpdateLog();
assertTrue((ulog.getStartingOperation() & UpdateLog.FLAG_GAP) == 0);
ulog.bufferUpdates();
// simulate receiving no updates
ulog.applyBufferedUpdates();
updateJ(
jsonAdd(sdoc("id", "Q7", "_version_", "117")),
params(
DISTRIB_UPDATE_PARAM,
FROM_LEADER)); // do another add to make sure flags are back to normal
req.close();
h.close();
createCore();
req = req();
uhandler = req.getCore().getUpdateHandler();
ulog = uhandler.getUpdateLog();
assertTrue((ulog.getStartingOperation() & UpdateLog.FLAG_GAP) == 0); // check flags on Q7
logReplayFinish.acquire();
assertEquals(
UpdateLog.State.ACTIVE, ulog.getState()); // leave each test method in a good state
} finally {
DirectUpdateHandler2.commitOnClose = true;
UpdateLog.testing_logReplayHook = null;
UpdateLog.testing_logReplayFinishHook = null;
req().close();
}
}
@Test
@Ignore("HDFS-3107: no truncate support yet")
public void testDropBuffered() throws Exception {
DirectUpdateHandler2.commitOnClose = false;
final Semaphore logReplay = new Semaphore(0);
final Semaphore logReplayFinish = new Semaphore(0);
UpdateLog.testing_logReplayHook =
new Runnable() {
@Override
public void run() {
try {
assertTrue(logReplay.tryAcquire(timeout, TimeUnit.SECONDS));
} catch (Exception e) {
throw new RuntimeException(e);
}
}
};
UpdateLog.testing_logReplayFinishHook =
new Runnable() {
@Override
public void run() {
logReplayFinish.release();
}
};
SolrQueryRequest req = req();
UpdateHandler uhandler = req.getCore().getUpdateHandler();
UpdateLog ulog = uhandler.getUpdateLog();
try {
clearIndex();
assertU(commit());
assertEquals(UpdateLog.State.ACTIVE, ulog.getState());
ulog.bufferUpdates();
assertEquals(UpdateLog.State.BUFFERING, ulog.getState());
Future<UpdateLog.RecoveryInfo> rinfoFuture = ulog.applyBufferedUpdates();
assertTrue(rinfoFuture == null);
assertEquals(UpdateLog.State.ACTIVE, ulog.getState());
ulog.bufferUpdates();
assertEquals(UpdateLog.State.BUFFERING, ulog.getState());
// simulate updates from a leader
updateJ(
jsonAdd(sdoc("id", "C1", "_version_", "101")), params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
updateJ(
jsonAdd(sdoc("id", "C2", "_version_", "102")), params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
updateJ(
jsonAdd(sdoc("id", "C3", "_version_", "103")), params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
assertTrue(ulog.dropBufferedUpdates());
ulog.bufferUpdates();
updateJ(
jsonAdd(sdoc("id", "C4", "_version_", "104")), params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
updateJ(
jsonAdd(sdoc("id", "C5", "_version_", "105")), params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
logReplay.release(1000);
rinfoFuture = ulog.applyBufferedUpdates();
UpdateLog.RecoveryInfo rinfo = rinfoFuture.get();
assertEquals(2, rinfo.adds);
assertJQ(req("qt", "/get", "getVersions", "2"), "=={'versions':[105,104]}");
// this time add some docs first before buffering starts (so tlog won't be at pos 0)
updateJ(
jsonAdd(sdoc("id", "C100", "_version_", "200")),
params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
updateJ(
jsonAdd(sdoc("id", "C101", "_version_", "201")),
params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
ulog.bufferUpdates();
updateJ(
jsonAdd(sdoc("id", "C103", "_version_", "203")),
params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
updateJ(
jsonAdd(sdoc("id", "C104", "_version_", "204")),
params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
assertTrue(ulog.dropBufferedUpdates());
ulog.bufferUpdates();
updateJ(
jsonAdd(sdoc("id", "C105", "_version_", "205")),
params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
updateJ(
jsonAdd(sdoc("id", "C106", "_version_", "206")),
params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
rinfoFuture = ulog.applyBufferedUpdates();
rinfo = rinfoFuture.get();
assertEquals(2, rinfo.adds);
assertJQ(
req("q", "*:*", "sort", "_version_ asc", "fl", "id,_version_"),
"/response/docs==["
+ "{'id':'C4','_version_':104}"
+ ",{'id':'C5','_version_':105}"
+ ",{'id':'C100','_version_':200}"
+ ",{'id':'C101','_version_':201}"
+ ",{'id':'C105','_version_':205}"
+ ",{'id':'C106','_version_':206}"
+ "]");
assertJQ(req("qt", "/get", "getVersions", "6"), "=={'versions':[206,205,201,200,105,104]}");
ulog.bufferUpdates();
assertEquals(UpdateLog.State.BUFFERING, ulog.getState());
updateJ(
jsonAdd(sdoc("id", "C301", "_version_", "998")),
params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
updateJ(
jsonAdd(sdoc("id", "C302", "_version_", "999")),
params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
assertTrue(ulog.dropBufferedUpdates());
// make sure we can overwrite with a lower version
// TODO: is this functionality needed?
updateJ(
jsonAdd(sdoc("id", "C301", "_version_", "301")),
params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
updateJ(
jsonAdd(sdoc("id", "C302", "_version_", "302")),
params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
assertU(commit());
assertJQ(req("qt", "/get", "getVersions", "2"), "=={'versions':[302,301]}");
assertJQ(
req("q", "*:*", "sort", "_version_ desc", "fl", "id,_version_", "rows", "2"),
"/response/docs==["
+ "{'id':'C302','_version_':302}"
+ ",{'id':'C301','_version_':301}"
+ "]");
updateJ(
jsonAdd(sdoc("id", "C2", "_version_", "302")), params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
assertEquals(
UpdateLog.State.ACTIVE, ulog.getState()); // leave each test method in a good state
} finally {
DirectUpdateHandler2.commitOnClose = true;
UpdateLog.testing_logReplayHook = null;
UpdateLog.testing_logReplayFinishHook = null;
req().close();
}
}
@Test
public void testBuffering() throws Exception {
DirectUpdateHandler2.commitOnClose = false;
final Semaphore logReplay = new Semaphore(0);
final Semaphore logReplayFinish = new Semaphore(0);
UpdateLog.testing_logReplayHook =
new Runnable() {
@Override
public void run() {
try {
assertTrue(logReplay.tryAcquire(timeout, TimeUnit.SECONDS));
} catch (Exception e) {
throw new RuntimeException(e);
}
}
};
UpdateLog.testing_logReplayFinishHook =
new Runnable() {
@Override
public void run() {
logReplayFinish.release();
}
};
SolrQueryRequest req = req();
UpdateHandler uhandler = req.getCore().getUpdateHandler();
UpdateLog ulog = uhandler.getUpdateLog();
try {
clearIndex();
assertU(commit());
assertEquals(UpdateLog.State.ACTIVE, ulog.getState());
ulog.bufferUpdates();
assertEquals(UpdateLog.State.BUFFERING, ulog.getState());
Future<UpdateLog.RecoveryInfo> rinfoFuture = ulog.applyBufferedUpdates();
assertTrue(rinfoFuture == null);
assertEquals(UpdateLog.State.ACTIVE, ulog.getState());
ulog.bufferUpdates();
assertEquals(UpdateLog.State.BUFFERING, ulog.getState());
// simulate updates from a leader
updateJ(
jsonAdd(sdoc("id", "B1", "_version_", "1010")),
params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
updateJ(
jsonAdd(sdoc("id", "B11", "_version_", "1015")),
params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
updateJ(
jsonDelQ("id:B1 id:B11 id:B2 id:B3"),
params(DISTRIB_UPDATE_PARAM, FROM_LEADER, "_version_", "-1017"));
updateJ(
jsonAdd(sdoc("id", "B2", "_version_", "1020")),
params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
updateJ(
jsonAdd(sdoc("id", "B3", "_version_", "1030")),
params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
deleteAndGetVersion("B1", params(DISTRIB_UPDATE_PARAM, FROM_LEADER, "_version_", "-2010"));
assertJQ(
req("qt", "/get", "getVersions", "6"),
"=={'versions':[-2010,1030,1020,-1017,1015,1010]}");
assertU(commit());
assertJQ(
req("qt", "/get", "getVersions", "6"),
"=={'versions':[-2010,1030,1020,-1017,1015,1010]}");
// updates should be buffered, so we should not see any results yet.
assertJQ(req("q", "*:*"), "/response/numFound==0");
// real-time get should also not show anything (this could change in the future,
// but it's currently used for validating version numbers too, so it would
// be bad for updates to be visible if we're just buffering.
assertJQ(req("qt", "/get", "id", "B3"), "=={'doc':null}");
rinfoFuture = ulog.applyBufferedUpdates();
assertTrue(rinfoFuture != null);
assertEquals(UpdateLog.State.APPLYING_BUFFERED, ulog.getState());
logReplay.release(1000);
UpdateLog.RecoveryInfo rinfo = rinfoFuture.get();
assertEquals(UpdateLog.State.ACTIVE, ulog.getState());
assertJQ(
req("qt", "/get", "getVersions", "6"),
"=={'versions':[-2010,1030,1020,-1017,1015,1010]}");
assertJQ(req("q", "*:*"), "/response/numFound==2");
// move back to recovering
ulog.bufferUpdates();
assertEquals(UpdateLog.State.BUFFERING, ulog.getState());
Long ver = getVer(req("qt", "/get", "id", "B3"));
assertEquals(1030L, ver.longValue());
// add a reordered doc that shouldn't overwrite one in the index
updateJ(
jsonAdd(sdoc("id", "B3", "_version_", "3")), params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
// reorder two buffered updates
updateJ(
jsonAdd(sdoc("id", "B4", "_version_", "1040")),
params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
deleteAndGetVersion(
"B4",
params(
DISTRIB_UPDATE_PARAM,
FROM_LEADER,
"_version_",
"-940")); // this update should not take affect
updateJ(
jsonAdd(sdoc("id", "B6", "_version_", "1060")),
params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
updateJ(
jsonAdd(sdoc("id", "B5", "_version_", "1050")),
params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
updateJ(
jsonAdd(sdoc("id", "B8", "_version_", "1080")),
params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
// test that delete by query is at least buffered along with everything else so it will delete
// the
// currently buffered id:8 (even if it doesn't currently support versioning)
updateJ(
"{\"delete\": { \"query\":\"id:B2 OR id:B8\" }}",
params(DISTRIB_UPDATE_PARAM, FROM_LEADER, "_version_", "-3000"));
assertJQ(
req("qt", "/get", "getVersions", "13"),
"=={'versions':[-3000,1080,1050,1060,-940,1040,3,-2010,1030,1020,-1017,1015,1010]}" // the
// "3"
// appears because versions aren't checked while buffering
);
logReplay.drainPermits();
rinfoFuture = ulog.applyBufferedUpdates();
assertTrue(rinfoFuture != null);
assertEquals(UpdateLog.State.APPLYING_BUFFERED, ulog.getState());
// apply a single update
logReplay.release(1);
// now add another update
updateJ(
jsonAdd(sdoc("id", "B7", "_version_", "1070")),
params(DISTRIB_UPDATE_PARAM, FROM_LEADER));
// a reordered update that should be dropped
deleteAndGetVersion("B5", params(DISTRIB_UPDATE_PARAM, FROM_LEADER, "_version_", "-950"));
deleteAndGetVersion("B6", params(DISTRIB_UPDATE_PARAM, FROM_LEADER, "_version_", "-2060"));
logReplay.release(1000);
UpdateLog.RecoveryInfo recInfo = rinfoFuture.get();
assertJQ(
req("q", "*:*", "sort", "id asc", "fl", "id,_version_"),
"/response/docs==["
+ "{'id':'B3','_version_':1030}"
+ ",{'id':'B4','_version_':1040}"
+ ",{'id':'B5','_version_':1050}"
+ ",{'id':'B7','_version_':1070}"
+ "]");
assertEquals(1, recInfo.deleteByQuery);
assertEquals(
UpdateLog.State.ACTIVE, ulog.getState()); // leave each test method in a good state
} finally {
DirectUpdateHandler2.commitOnClose = true;
UpdateLog.testing_logReplayHook = null;
UpdateLog.testing_logReplayFinishHook = null;
req().close();
}
}