use of com.alibaba.otter.shared.etl.model.EventData in project otter by alibaba.
the class DbLoadMerger method mergeDelete.
private static void mergeDelete(EventData eventData, Map<RowKey, EventData> result) {
// 只保留pks,把columns去掉. 以后针对数据仓库可以开放delete columns记录
RowKey rowKey = new RowKey(eventData.getTableId(), eventData.getSchemaName(), eventData.getTableName(), eventData.getKeys());
if (!result.containsKey(rowKey)) {
result.put(rowKey, eventData);
} else {
EventData oldEventData = result.get(rowKey);
eventData.setSize(oldEventData.getSize() + eventData.getSize());
if (!CollectionUtils.isEmpty(oldEventData.getOldKeys())) {
// 存在主键变更
// insert/update -> delete记录组合时,delete的对应的pk为上一条记录的pk
eventData.setKeys(oldEventData.getOldKeys());
// 清除oldKeys
eventData.getOldKeys().clear();
// 删除老的对象
result.remove(rowKey);
result.put(new RowKey(eventData.getTableId(), eventData.getSchemaName(), eventData.getTableName(), eventData.getKeys()), // key发生变化,需要重新构造一个RowKey
eventData);
} else {
// 清除oldKeys
eventData.getOldKeys().clear();
result.put(rowKey, eventData);
}
}
}
use of com.alibaba.otter.shared.etl.model.EventData in project otter by alibaba.
the class DbLoadMerger method mergeInsert.
private static void mergeInsert(EventData eventData, Map<RowKey, EventData> result) {
// insert无主键变更的处理
RowKey rowKey = new RowKey(eventData.getTableId(), eventData.getSchemaName(), eventData.getTableName(), eventData.getKeys());
if (!result.containsKey(rowKey)) {
result.put(rowKey, eventData);
} else {
EventData oldEventData = result.get(rowKey);
eventData.setSize(oldEventData.getSize() + eventData.getSize());
// 如果上一条变更是delete的,就直接用insert替换
if (oldEventData.getEventType() == EventType.DELETE) {
result.put(rowKey, eventData);
} else if (oldEventData.getEventType() == EventType.UPDATE || oldEventData.getEventType() == EventType.INSERT) {
// insert之前出现了update逻辑上不可能,唯一的可能性主要是Freedom的介入,人为的插入了一条Insert记录
// 不过freedom一般不建议Insert操作,只建议执行update/delete操作. update默认会走merge
// sql,不存在即插入
logger.warn("update-insert/insert-insert happend. before[{}] , after[{}]", oldEventData, eventData);
// 如果上一条变更是update的,就用insert替换,并且把上一条存在而这一条不存在的字段值拷贝到这一条中
EventData mergeEventData = replaceColumnValue(eventData, oldEventData);
// 清空oldkeys,insert记录不需要
mergeEventData.getOldKeys().clear();
result.put(rowKey, mergeEventData);
}
}
}
use of com.alibaba.otter.shared.etl.model.EventData in project otter by alibaba.
the class DbLoadMerger method mergeUpdate.
private static void mergeUpdate(EventData eventData, Map<RowKey, EventData> result) {
RowKey rowKey = new RowKey(eventData.getTableId(), eventData.getSchemaName(), eventData.getTableName(), eventData.getKeys());
if (!CollectionUtils.isEmpty(eventData.getOldKeys())) {
// 存在主键变更
// 需要解决(1->2 , 2->3)级联主键变更的问题
RowKey oldKey = new RowKey(eventData.getTableId(), eventData.getSchemaName(), eventData.getTableName(), eventData.getOldKeys());
if (!result.containsKey(oldKey)) {
// 不需要级联
result.put(rowKey, eventData);
} else {
EventData oldEventData = result.get(oldKey);
eventData.setSize(oldEventData.getSize() + eventData.getSize());
// 如果上一条变更是insert的,就把这一条的eventType改成insert,并且把上一条存在而这一条不存在的字段值拷贝到这一条中
if (oldEventData.getEventType() == EventType.INSERT) {
eventData.setEventType(EventType.INSERT);
// 删除当前变更数据老主键的记录.
result.remove(oldKey);
EventData mergeEventData = replaceColumnValue(eventData, oldEventData);
// 清空oldkeys,insert记录不需要
mergeEventData.getOldKeys().clear();
result.put(rowKey, mergeEventData);
} else if (oldEventData.getEventType() == EventType.UPDATE) {
// 删除当前变更数据老主键的记录.
result.remove(oldKey);
// 如果上一条变更是update的,把上一条存在而这一条不存在的数据拷贝到这一条中
EventData mergeEventData = replaceColumnValue(eventData, oldEventData);
result.put(rowKey, mergeEventData);
} else {
throw new LoadException("delete(has old pks) + update impossible happed!");
}
}
} else {
if (!result.containsKey(rowKey)) {
// 没有主键变更
result.put(rowKey, eventData);
} else {
EventData oldEventData = result.get(rowKey);
// 如果上一条变更是insert的,就把这一条的eventType改成insert,并且把上一条存在而这一条不存在的字段值拷贝到这一条中
if (oldEventData.getEventType() == EventType.INSERT) {
eventData.setEventType(EventType.INSERT);
EventData mergeEventData = replaceColumnValue(eventData, oldEventData);
result.put(rowKey, mergeEventData);
} else if (oldEventData.getEventType() == EventType.UPDATE) {
// 可能存在
// 1->2
// ,
// 2update的问题
// 如果上一条变更是update的,把上一条存在而这一条不存在的数据拷贝到这一条中
EventData mergeEventData = replaceColumnValue(eventData, oldEventData);
result.put(rowKey, mergeEventData);
} else if (oldEventData.getEventType() == EventType.DELETE) {
//异常情况,出现 delete + update,那就直接更新为update
result.put(rowKey, eventData);
}
}
}
}
use of com.alibaba.otter.shared.etl.model.EventData in project otter by alibaba.
the class FreedomExtractor method extract.
public void extract(DbBatch dbBatch) throws ExtractException {
Assert.notNull(dbBatch);
// 读取配置
Pipeline pipeline = getPipeline(dbBatch.getRowBatch().getIdentity().getPipelineId());
boolean skipFreedom = pipeline.getParameters().getSkipFreedom();
String bufferSchema = pipeline.getParameters().getSystemSchema();
String bufferTable = pipeline.getParameters().getSystemBufferTable();
List<EventData> eventDatas = dbBatch.getRowBatch().getDatas();
// 使用set,提升remove时的查找速度
Set<EventData> removeDatas = new HashSet<EventData>();
for (EventData eventData : eventDatas) {
if (StringUtils.equalsIgnoreCase(bufferSchema, eventData.getSchemaName()) && StringUtils.equalsIgnoreCase(bufferTable, eventData.getTableName())) {
if (eventData.getEventType().isDdl()) {
continue;
}
if (skipFreedom) {
// 判断是否需要忽略
removeDatas.add(eventData);
continue;
}
// 只处理insert / update记录
if (eventData.getEventType().isInsert() || eventData.getEventType().isUpdate()) {
// 重新改写一下EventData的数据,根据系统表的定义
EventColumn tableIdColumn = getMatchColumn(eventData.getColumns(), TABLE_ID);
// 获取到对应tableId的media信息
try {
DataMedia dataMedia = null;
Long tableId = Long.valueOf(tableIdColumn.getColumnValue());
eventData.setTableId(tableId);
if (tableId <= 0) {
// 直接按照full_name进行查找
// 尝试直接根据schema+table name进行查找
EventColumn fullNameColumn = getMatchColumn(eventData.getColumns(), FULL_NAME);
if (fullNameColumn != null) {
String[] names = StringUtils.split(fullNameColumn.getColumnValue(), ".");
if (names.length >= 2) {
dataMedia = ConfigHelper.findSourceDataMedia(pipeline, names[0], names[1]);
eventData.setTableId(dataMedia.getId());
} else {
throw new ConfigException("no such DataMedia " + names);
}
}
} else {
// 如果指定了tableId,需要按照tableId进行严格查找,如果没找到,那说明不需要进行同步
dataMedia = ConfigHelper.findDataMedia(pipeline, Long.valueOf(tableIdColumn.getColumnValue()));
}
DbDialect dbDialect = dbDialectFactory.getDbDialect(pipeline.getId(), (DbMediaSource) dataMedia.getSource());
// 考虑offer[1-128]的配置模式
if (!dataMedia.getNameMode().getMode().isSingle() || !dataMedia.getNamespaceMode().getMode().isSingle()) {
boolean hasError = true;
EventColumn fullNameColumn = getMatchColumn(eventData.getColumns(), FULL_NAME);
if (fullNameColumn != null) {
String[] names = StringUtils.split(fullNameColumn.getColumnValue(), ".");
if (names.length >= 2) {
eventData.setSchemaName(names[0]);
eventData.setTableName(names[1]);
hasError = false;
}
}
if (hasError) {
// 出现异常,需要记录一下
logger.warn("dataMedia mode:{} , fullname:{} ", dataMedia.getMode(), fullNameColumn == null ? null : fullNameColumn.getColumnValue());
removeDatas.add(eventData);
// 跳过这条记录
continue;
}
} else {
eventData.setSchemaName(dataMedia.getNamespace());
eventData.setTableName(dataMedia.getName());
}
// 更新业务类型
EventColumn typeColumn = getMatchColumn(eventData.getColumns(), TYPE);
EventType eventType = EventType.valuesOf(typeColumn.getColumnValue());
eventData.setEventType(eventType);
if (eventType.isUpdate()) {
// 如果是update强制修改为insert,这样可以在目标端执行merge
// sql
eventData.setEventType(EventType.INSERT);
} else if (eventType.isDdl()) {
dbDialect.reloadTable(eventData.getSchemaName(), eventData.getTableName());
// 删除当前记录
removeDatas.add(eventData);
continue;
}
// 重新构建新的业务主键字段
EventColumn pkDataColumn = getMatchColumn(eventData.getColumns(), PK_DATA);
String pkData = pkDataColumn.getColumnValue();
String[] pks = StringUtils.split(pkData, PK_SPLIT);
Table table = dbDialect.findTable(eventData.getSchemaName(), eventData.getTableName());
List<EventColumn> newColumns = new ArrayList<EventColumn>();
Column[] primaryKeyColumns = table.getPrimaryKeyColumns();
if (primaryKeyColumns.length > pks.length) {
throw new ExtractException("data pk column size not match , data:" + eventData.toString());
}
// 构建字段
Column[] allColumns = table.getColumns();
int pkIndex = 0;
for (int i = 0; i < allColumns.length; i++) {
Column column = allColumns[i];
if (column.isPrimaryKey()) {
EventColumn newColumn = new EventColumn();
// 设置下标
newColumn.setIndex(i);
newColumn.setColumnName(column.getName());
newColumn.setColumnType(column.getTypeCode());
newColumn.setColumnValue(pks[pkIndex]);
newColumn.setKey(true);
newColumn.setNull(pks[pkIndex] == null);
newColumn.setUpdate(true);
// 添加到记录
newColumns.add(newColumn);
pkIndex++;
}
}
// 设置数据
eventData.setKeys(newColumns);
eventData.setOldKeys(new ArrayList<EventColumn>());
eventData.setColumns(new ArrayList<EventColumn>());
// 设置为行记录+反查
eventData.setSyncMode(SyncMode.ROW);
eventData.setSyncConsistency(SyncConsistency.MEDIA);
eventData.setRemedy(true);
// 默认为1kb,如果还是按照binlog大小计算的话,可能会采用rpc传输,导致内存不够用
eventData.setSize(1024);
} catch (ConfigException e) {
// 忽略掉,因为系统表会被共享,所以这条记录会被不是该同步通道给获取到
logger.info("find DataMedia error " + eventData.toString(), e);
removeDatas.add(eventData);
continue;
} catch (Throwable e) {
// 出现异常时忽略掉
logger.warn("process freedom data error " + eventData.toString(), e);
removeDatas.add(eventData);
continue;
}
} else {
// 删除该记录
removeDatas.add(eventData);
}
}
}
if (!CollectionUtils.isEmpty(removeDatas)) {
eventDatas.removeAll(removeDatas);
}
}
use of com.alibaba.otter.shared.etl.model.EventData in project otter by alibaba.
the class ProcessorExtractor method extract.
public void extract(DbBatch param) throws ExtractException {
ExecutorTemplate executorTemplate = null;
try {
RowBatch rowBatch = param.getRowBatch();
final Pipeline pipeline = getPipeline(rowBatch.getIdentity().getPipelineId());
List<EventData> eventDatas = rowBatch.getDatas();
// 使用set,提升remove时的查找速度
final Set<EventData> removeDatas = Collections.synchronizedSet(new HashSet<EventData>());
executorTemplate = executorTemplateGetter.get();
executorTemplate.start();
// 重新设置下poolSize
executorTemplate.adjustPoolSize(pipeline.getParameters().getExtractPoolSize());
for (final EventData eventData : eventDatas) {
List<DataMediaPair> dataMediaPairs = ConfigHelper.findDataMediaPairByMediaId(pipeline, eventData.getTableId());
if (dataMediaPairs == null) {
throw new ExtractException("ERROR ## the dataMediaId = " + eventData.getTableId() + " dataMediaPair is null,please check");
}
for (DataMediaPair dataMediaPair : dataMediaPairs) {
if (!dataMediaPair.isExistFilter()) {
continue;
}
final EventProcessor eventProcessor = extensionFactory.getExtension(EventProcessor.class, dataMediaPair.getFilterData());
if (eventProcessor instanceof DataSourceFetcherAware) {
((DataSourceFetcherAware) eventProcessor).setDataSourceFetcher(new DataSourceFetcher() {
@Override
public DataSource fetch(Long tableId) {
DataMedia dataMedia = ConfigHelper.findDataMedia(pipeline, tableId);
return dataSourceService.getDataSource(pipeline.getId(), dataMedia.getSource());
}
});
executorTemplate.submit(new Runnable() {
@Override
public void run() {
MDC.put(OtterConstants.splitPipelineLogFileKey, String.valueOf(pipeline.getId()));
boolean process = eventProcessor.process(eventData);
if (!process) {
// 添加到删除记录中
removeDatas.add(eventData);
}
}
});
} else {
boolean process = eventProcessor.process(eventData);
if (!process) {
// 添加到删除记录中
removeDatas.add(eventData);
break;
}
}
}
}
// 等待所有都处理完成
executorTemplate.waitForResult();
if (!CollectionUtils.isEmpty(removeDatas)) {
eventDatas.removeAll(removeDatas);
}
} finally {
if (executorTemplate != null) {
executorTemplateGetter.release(executorTemplate);
}
}
}
Aggregations