Indexing Overview

WSO2 DAS has a distributed indexing engine that is built on top of Apache Lucene. Data is indexed and saved in the Data Access Layer of DAS in a store referred to as the Analytics File System. This section explains how indexing is carried out in WSO2 DAS.

A main requirement of DAS is to support arbitrary key value pairs. Most of the RDBMS have a fixed schema and do not support storing arbitrary key values. In DAS, records are stored as blobs. This way, the data fields in a record can be converted to a blob and stored together with the record ID and the timestamp of the record. Storing data as blobs has one drawback. That is database level indexing cannot be used for blobs to search by fields. To overcome this issue, records are sent through the indexer before the records are persisted in DAS so that the indexer can keep a searchable copy of record data along with the record ID. This indexer does not store the field values. It only stores the record ID and keeps an index of data fields. When a search query is sent to DAS, the indexer identifies the matching records, gets their record IDs, and sends them to the persistence layer/recordstore. In recordstore level, the record blobs are deserialized and returned as the record objects from the AnalyticsDataService implementation.

Indexing Architecture

DAS indexer is implemented using Apache Lucene which is a full text search library. Users can index records and search for records later via Lucene queries. Events received by DAS are converted to a list of  records and inserted into FileSystem based queues. These queues are created in <DAS_HOME>/repository/data/index_staging_queues directory. With a background thread, these queues are consumed and records are indexed. The indexed data is stored in the <DAS_HOME>/repository/data/index_data directory. The DAS index consists of smaller indexes known as shards. A shard can be accessed only by one index writer at a given time. (Index writer is the lucene class visible to outside world that is used to write lucene documents to a file system based index).Therefore having multiple shards can increase the write throughput (however, the write throughput can be limited by Disk IO operations). By default, DAS is configured to have six shards and one replica.

Indexing related configurations are done in the <DAS_HOME>/repository/conf/analytics/analytics-config.xml file. Additionally, information relating to shards is maintained in the <DAS_HOME>/repository/data/local-shard-allocation-config.conf file. This file stores the shard number along with its state (that can be INIT or NORMAL). The INIT is the initial state. Usually this state cannot be seen from outside. This is because the INIT state changes to the NORMAL state once the server starts. If the indexing node is running, the state of shards should be NORMAL and not INIT. The NORMAL state denotes that the indexing node has started indexing. Therefore, whenever the data is ready to be indexed, the indexer node indexes the incoming data.

By default, a single DAS server is configured to have six shards, and one replica for each shard (i.e., a total of twelve shards). Therefore, in a minimum HA setup, each DAS server contains three shards and replicas of the other three shards. Even if one server is shut down, the second server has replicas of the three shards that were there in the node that was shut down. This allows the high availability to be maintained.

The number of replicas and the number of shards are defined in the analytics-config.xml. By changing these properties we can change the number of shards and replicas, but once they are defined and the server is started, we cannot change them. Events are not needed to be published to both nodes. Replication happens through hazelcast messages by sending records to other nodes. 

The following example further explains how shards are managed in a clustered deployment.

If there are three DAS nodes in the cluster, all twelve shards (i.e., the shards and the replicas) are split among the three nodes. Each node will have four shrds which consists of 2 shards and 2 replicas.

Re-indexing existing data

To re-index the whole dataset, follow the steps below:

1. Shut down all the DAS nodes in your deployment.

2. Truncate the following tables.

   Table names are encoded in characters that cannot be read by humans for all database types other than HBase and Cassandra. Therefore, if the database type is not HBase or Cassandra, you need to use the encoded versions of these table names depending on the database type when performing this step.

   
   

   • ANX_X1000__INDEX_STAGING_DATA_*
   • IDX_X1000__INDEX_STAGING_DATA_*

3. Drop all the tables with names that start with ANX and IDX such as the examples given below.

   Table names are encoded in characters that cannot be read by humans for all database types other than HBase and Cassandra. Therefore, this step is relevant only if the database type used is HBase or Cassandra because the table names given below can only exist for ese two database types

   
   

   • ANX_X1000__INDEX_STAGING_DATA_*
   • IDX_X1000__INDEX_STAGING_DATA_*
4. For each node, remove all the index data stored in the <DAS_HOME>/repository/data directory.
5. Restart the DAS nodes.

Debugging Indexing in a cluster

If there seems to be an issue with indexing, do the following in the given order. 

1. In the <DAS_HOME>/repository/conf/analytics/analytics-config.xml file check whether the shards are properly allocated.

   <!-- The number of index data replicas the system should keep, for H/A, this should be at least 1, e.g. the value 0 means
           there aren't any copies of the data -->
      <indexReplicationFactor>1</indexReplicationFactor>
      <!-- The number of index shards, should be equal or higher to the number of indexing nodes that is going to be working,
           ideal count being 'number of indexing nodes * [CPU cores used for indexing per node]' -->
      <shardCount>6</shardCount>

2. Check the logs that are printed when the cluster is initilized. The log must be similar to the example shown below.

   TID: [-1] [] [2017-05-12 13:17:27,538]  INFO {org.wso2.carbon.analytics.dataservice.core.indexing.IndexNodeCoordinator} -  Indexing Initialized: CLUSTERED {0={c13d3a23-b15a-4b9c-ac8f-a30df2811c98=Member [10.3.24.70]:4000 this, bc751d36-d345-4b8f-b133-b77793f04805=Member [10.3.24.67]:4000}, 1={c13d3a23-b15a-4b9c-ac8f-a30df2811c98=Member [10.3.24.70]:4000 this, bc751d36-d345-4b8f-b133-b77793f04805=Member [10.3.24.67]:4000}, 2={c13d3a23-b15a-4b9c-ac8f-a30df2811c98=Member [10.3.24.70]:4000 this, bc751d36-d345-4b8f-b133-b77793f04805=Member [10.3.24.67]:4000}, 3={c13d3a23-b15a-4b9c-ac8f-a30df2811c98=Member [10.3.24.70]:4000 this, bc751d36-d345-4b8f-b133-b77793f04805=Member [10.3.24.67]:4000}, 4={c13d3a23-b15a-4b9c-ac8f-a30df2811c98=Member [10.3.24.70]:4000 this, bc751d36-d345-4b8f-b133-b77793f04805=Member [10.3.24.67]:4000}, 5={c13d3a23-b15a-4b9c-ac8f-a30df2811c98=Member [10.3.24.70]:4000 this, bc751d36-d345-4b8f-b133-b77793f04805=Member [10.3.24.67]:4000}} | Current Node Indexing: Yes {org.wso2.carbon.analytics.dataservice.core.indexing.IndexNodeCoordinator}

   Here, {0={c13d3a23-b15a-4b9c-ac8f-a30df2811c98=Member [10.36.241.70]:4000 this, bc751d36-d345-4b8f-b133-b77793f04805=Member [10.36.241.67]:4000} means that the shard 0 is allocated to two nodes, and their IDs are c13d3a23-b15a-4b9c-ac8f-a30df2811c98 and bc751d36-d345-4b8f-b133-b77793f04805. The IPs of the two nodes are also mentioned in the log line. For a correctly configured two-node DAS cluster, this log line must contain all six shards (from 0 to 5) and the node IDs to which they are allocated.

   This log line does not contain both the node IDs when you initially set up the cluster and start the first server because the other node has not joined the cluster. The log line is printed with the complete mapping of shards to node IDs only after the 2nd node joins the cluster.

   The two node IDs mentioned in the log line must match the IDs mentioned in the my-node-id.dat of both DAS nodes. If there are more than two unique IDs in a 2 node cluster, the shard allocation may have been affected and it mayneed to be corrected.
   
   One reason to have more than one node ID is to allow the cluster to be reconfigured with new DAS packs without clearing the data sources configured in the analytics-datasources.xml file. When a new DAS pack that points to the same analytics data sources is started, a new ID is generated and the new DAS pack is considered the third node to join the cluster. When you re-configure the cluster with new DAS packs, make sure the older IDs are first backed up and restored in the new packs. For more information, see Backing up and Restoring Analytics Data. If the two node cluster already has more than two unique IDs ( a result of starting a new DAS pack with a new ID), remove the unnecessary/additional IDs by following the steps given below.

   1. Keep backup of the current my-node-id.dat of both nodes.

   2. Separate the IDs that do not match with the my-node-id.dat of the two nodes.
   3. Shut down both the nodes.

   4. Replace the my-node-id.dat of either DAS node with one of the non-matching/unnecessary/additional node IDs.
      

      Make sure that you keep a backup of the my-node-id.dat before doing this.

       

   5. Start the DAS node mentioned in substep d with the -DdisableIndexing=true property. This property removes the node ID that you enter in the my-node-id.dat from the indexing cluster.
   6. Repeat substeps c, d, and e for all the additional node IDs.
   7. Restore the two node IDs you backed up in substep d.

   8. Clear the information in the in the <DAS_HOME>/repository/data directory.

   9. Start both the nodes, and cheeck the content of the the <DAS_HOME>/repository/data/local-shard-allocation-config.conf file. It should be changed as follows.

      0,NORMAL
      1,NORMAL
      2,NORMAL
      3,NORMAL
      4,NORMAL
      5,NORMAL

      The nodes in this example are configured with six shards and one replicas for each shard. This step must change based on the number of shards and replicas configured.

Indexing in Spark Environment

Spark runs in a different JVM. Therefore, it cannot write directly to indexing queues. When You instruct Spark to index its summarized data, it sends the summarized data to be indexed to a storage area called the staging area in the primary record store. This data is fetched and inserted to the indexing que via the staging index worker threads. From this point onwards, the indexing process is the same as described in the previous sections.