Java Code Examples for org.apache.solr.client.solrj.request.CollectionAdminRequest#ReindexCollection

@Test
public void testBasicReindexing() throws Exception {
  final String sourceCollection = "basicReindexing";

  createCollection(sourceCollection, "conf1", 2, 2);

  indexDocs(sourceCollection, NUM_DOCS,
      i -> new SolrInputDocument("id", String.valueOf(i), "string_s", String.valueOf(i)));

  final String targetCollection = "basicReindexingTarget";

  CollectionAdminRequest.ReindexCollection req = CollectionAdminRequest.reindexCollection(sourceCollection)
      .setTarget(targetCollection);
  CollectionAdminResponse rsp = req.process(solrClient);
  assertNotNull(rsp.toString(), rsp.getResponse().get(ReindexCollectionCmd.REINDEX_STATUS));
  @SuppressWarnings({"unchecked"})
  Map<String, Object> status = (Map<String, Object>)rsp.getResponse().get(ReindexCollectionCmd.REINDEX_STATUS);
  assertEquals(status.toString(), (long)NUM_DOCS, ((Number)status.get("inputDocs")).longValue());
  assertEquals(status.toString(), (long)NUM_DOCS, ((Number)status.get("processedDocs")).longValue());

  CloudUtil.waitForState(cloudManager, "did not finish copying in time", targetCollection, (liveNodes, coll) -> {
    ReindexCollectionCmd.State state = ReindexCollectionCmd.State.get(coll.getStr(ReindexCollectionCmd.REINDEXING_STATE));
    return ReindexCollectionCmd.State.FINISHED == state;
  });

  // verify the target docs exist
  QueryResponse queryResponse = solrClient.query(targetCollection, params(CommonParams.Q, "*:*"));
  assertEquals("copied num docs", NUM_DOCS, queryResponse.getResults().getNumFound());
}
 

@Test
public void testLossySchema() throws Exception {
  final String sourceCollection = "sourceLossyReindexing";
  final String targetCollection = "targetLossyReindexing";


  createCollection(sourceCollection, "conf2", 2, 2);

  indexDocs(sourceCollection, NUM_DOCS, i ->
    new SolrInputDocument(
        "id", String.valueOf(i),
        "string_s", String.valueOf(i),
        "sind", "this is a test " + i)); // "sind": indexed=true, stored=false, will be lost...

  CollectionAdminRequest.ReindexCollection req = CollectionAdminRequest.reindexCollection(sourceCollection)
      .setTarget(targetCollection)
      .setConfigName("conf3");
  req.process(solrClient);

  CloudUtil.waitForState(cloudManager, "did not finish copying in time", targetCollection, (liveNodes, coll) -> {
    ReindexCollectionCmd.State state = ReindexCollectionCmd.State.get(coll.getStr(ReindexCollectionCmd.REINDEXING_STATE));
    return ReindexCollectionCmd.State.FINISHED == state;
  });
  // verify the target docs exist
  QueryResponse rsp = solrClient.query(targetCollection, params(CommonParams.Q, "*:*"));
  assertEquals("copied num docs", NUM_DOCS, rsp.getResults().getNumFound());
  for (SolrDocument doc : rsp.getResults()) {
    String id = (String)doc.getFieldValue("id");
    assertEquals(id, doc.getFieldValue("string_s"));
    assertFalse(doc.containsKey("sind")); // lost in translation ...
  }
}
 

@Test
public void testFailure() throws Exception {
  final String sourceCollection = "failReindexing";
  final String targetCollection = "failReindexingTarget";
  final String aliasTarget = "failAlias";
  createCollection(sourceCollection, "conf1", 2, 2);
  createCollection(targetCollection, "conf1", 1, 1);
  CollectionAdminRequest.createAlias(aliasTarget, targetCollection).process(solrClient);
  indexDocs(sourceCollection, NUM_DOCS,
      i -> new SolrInputDocument(
          "id", String.valueOf(i),
          "string_s", String.valueOf(i)));

  CollectionAdminRequest.ReindexCollection req = CollectionAdminRequest.reindexCollection(sourceCollection)
      .setTarget(targetCollection);
  CollectionAdminResponse rsp = req.process(solrClient);
  assertNotNull(rsp.getResponse().get("error"));
  assertTrue(rsp.toString(), rsp.getResponse().get("error").toString().contains("already exists"));

  req = CollectionAdminRequest.reindexCollection(sourceCollection)
      .setTarget(aliasTarget);
  rsp = req.process(solrClient);
  assertNotNull(rsp.getResponse().get("error"));
  assertTrue(rsp.toString(), rsp.getResponse().get("error").toString().contains("already exists"));

  CollectionAdminRequest.deleteAlias(aliasTarget).process(solrClient);
  CollectionAdminRequest.deleteCollection(targetCollection).process(solrClient);

  req = CollectionAdminRequest.reindexCollection(sourceCollection)
      .setTarget(targetCollection);

  TestInjection.reindexFailure = "true:100";
  rsp = req.process(solrClient);
  assertNotNull(rsp.getResponse().get("error"));
  assertTrue(rsp.toString(), rsp.getResponse().get("error").toString().contains("waiting for daemon"));

  // verify that the target and checkpoint collections don't exist
  cloudManager.getClusterStateProvider().getClusterState().forEachCollection(coll -> {
    assertFalse(coll.getName() + " still exists", coll.getName().startsWith(ReindexCollectionCmd.TARGET_COL_PREFIX));
    assertFalse(coll.getName() + " still exists", coll.getName().startsWith(ReindexCollectionCmd.CHK_COL_PREFIX));
  });
  // verify that the source collection is read-write and has no reindexing flags
  CloudUtil.waitForState(cloudManager, "collection state is incorrect", sourceCollection,
      ((liveNodes, collectionState) ->
          !collectionState.isReadOnly() &&
          collectionState.getStr(ReindexCollectionCmd.REINDEXING_STATE) == null &&
          getState(sourceCollection) == null));
}
 

@Test
@SuppressWarnings({"unchecked"})
public void testAbort() throws Exception {
  final String sourceCollection = "abortReindexing";
  final String targetCollection = "abortReindexingTarget";
  createCollection(sourceCollection, "conf1", 2, 1);

  TestInjection.reindexLatch = new CountDownLatch(1);
  CollectionAdminRequest.ReindexCollection req = CollectionAdminRequest.reindexCollection(sourceCollection)
      .setTarget(targetCollection);
  String asyncId = req.processAsync(solrClient);
  // wait for the source collection to be put in readOnly mode
  CloudUtil.waitForState(cloudManager, "source collection didn't become readOnly",
      sourceCollection, (liveNodes, coll) -> coll.isReadOnly());

  req = CollectionAdminRequest.reindexCollection(sourceCollection);
  req.setCommand("abort");
  CollectionAdminResponse rsp = req.process(solrClient);
  Map<String, Object> status = (Map<String, Object>)rsp.getResponse().get(ReindexCollectionCmd.REINDEX_STATUS);
  assertNotNull(rsp.toString(), status);
  assertEquals(status.toString(), "aborting", status.get("state"));

  CloudUtil.waitForState(cloudManager, "incorrect collection state", sourceCollection,
      ((liveNodes, collectionState) ->
          collectionState.isReadOnly() &&
          getState(sourceCollection) == ReindexCollectionCmd.State.ABORTED));

  // verify status
  req.setCommand("status");
  rsp = req.process(solrClient);
  status = (Map<String, Object>)rsp.getResponse().get(ReindexCollectionCmd.REINDEX_STATUS);
  assertNotNull(rsp.toString(), status);
  assertEquals(status.toString(), "aborted", status.get("state"));
  // let the process continue
  TestInjection.reindexLatch.countDown();
  CloudUtil.waitForState(cloudManager, "source collection is in wrong state",
      sourceCollection, (liveNodes, docCollection) -> !docCollection.isReadOnly() && getState(sourceCollection) == null);
  // verify the response
  rsp = CollectionAdminRequest.requestStatus(asyncId).process(solrClient);
  status = (Map<String, Object>)rsp.getResponse().get(ReindexCollectionCmd.REINDEX_STATUS);
  assertNotNull(rsp.toString(), status);
  assertEquals(status.toString(), "aborted", status.get("state"));
}