Scheduling reroute for delayed shards in delayed shards – How to solve this OpenSearch error

Overview

Data in an OpenSearch index can grow to massive proportions. In order to keep it manageable, it is split into a number of shards. Each OpenSearch shard is an Apache Lucene index, with each individual Lucene index containing a subset of the documents in the OpenSearch index. Splitting indices in this way keeps resource usage under control. An Apache Lucene index has a limit of 2,147,483,519 documents.

Examples

The number of shards is set when an index is created, and this number cannot be changed later without reindexing the data. When creating an index, you can set the number of shards and replicas as properties of the index using:

PUT /sensor
{
    "settings" : {
        "index" : {
            "number_of_shards" : 6,
            "number_of_replicas" : 2
        }
    }
}

The ideal number of shards should be determined based on the amount of data in an index. Generally, an optimal shard should hold 30-50GB of data. For example, if you expect to accumulate around 300GB of application logs in a day, having around 10 shards in that index would be reasonable.

During their lifetime, shards can go through a number of states, including:

• Initializing: An initial state before the shard can be used.
• Started: A state in which the shard is active and can receive requests.
• Relocating: A state that occurs when shards are in the process of being moved to a different node. This may be necessary under certain conditions, such as when the node they are on is running out of disk space.
• Unassigned: The state of a shard that has failed to be assigned. A reason is provided when this happens. For example, if the node hosting the shard is no longer in the cluster (NODE_LEFT) or due to restoring into a closed index (EXISTING_INDEX_RESTORED).

In order to view all shards, their states, and other metadata, use the following request:

GET _cat/shards

To view shards for a specific index, append the name of the index to the URL, for example:

sensor:
GET _cat/shards/sensor

This command produces output, such as in the following example. By default, the columns shown include the name of the index, the name (i.e. number) of the shard, whether it is a primary shard or a replica, its state, the number of documents, the size on disk, the IP address, and the node ID.

sensor 5 p STARTED    0  283b 127.0.0.1 ziap
sensor 5 r UNASSIGNED                   
sensor 2 p STARTED    1 3.7kb 127.0.0.1 ziap
sensor 2 r UNASSIGNED                   
sensor 3 p STARTED    3 7.2kb 127.0.0.1 ziap
sensor 3 r UNASSIGNED                   
sensor 1 p STARTED    1 3.7kb 127.0.0.1 ziap
sensor 1 r UNASSIGNED                   
sensor 4 p STARTED    2 3.8kb 127.0.0.1 ziap
sensor 4 r UNASSIGNED                   
sensor 0 p STARTED    0  283b 127.0.0.1 ziap
sensor 0 r UNASSIGNED

Notes and good things to know

• Having shards that are too large is simply inefficient. Moving huge indices across machines is both a time- and labor-intensive process. First, the Lucene merges would take longer to complete and would require greater resources. Moreover, moving the shards across the nodes for rebalancing would also take longer and recovery time would be extended. Thus by splitting the data and spreading it across a number of machines, it can be kept in manageable chunks and minimize risks.

• Having the right number of shards is important for performance. It is thus wise to plan in advance. When queries are run across different shards in parallel, they execute faster than an index composed of a single shard, but only if each shard is located on a different node and there are sufficient nodes in the cluster. At the same time, however, shards consume memory and disk space, both in terms of indexed data and cluster metadata. Having too many shards can slow down queries, indexing requests, and management operations, and so maintaining the right balance is critical.

How to reduce your OpenSearch costs by optimizing your shards

Watch the video below to learn how to save money on your deployment by optimizing your shards.

Overview

The shard is selected based on the following formula:

shard = hash(_routing) % number_of_primary_shards

Where the default value of _routing is _id.
It is important to know which shard the document is routed to because OpenSearch will need to determine where to find that document later on for document retrieval requests. 

Examples

In twitter index with 2 primary shards, the document with _id equal to “440” gets routed to the shard number:   

shard = hash( 440 ) % 2
PUT twitter/_doc/440
{
...
}

Notes and good things to know

• In order to improve search speed, you can create custom routing. For example, you can enable custom routing that will ensure that only a single shard will be queried (the shard that contains your data).
• To create custom routing in OpenSearch, you will need to configure and define that not all routing will be completed by default settings. ( v <= 5.0)

PUT my_index/customer/_mapping
{
   "order":{
      "_routing":{
         "required":true
      }
   }
}

•  This will ensure that every document in the “customer” type must specify a custom routing. 

Overview

Search or indexing requests will usually be load-balanced across the OpenSearch data nodes, and the node that receives the request will relay requests to other nodes as necessary and coordinate the response back to the user.

Notes and good things to know

The key elements to clustering are:

Cluster State – Refers to information about which indices are in the cluster, their data mappings and other information that must be shared between all the nodes to ensure that all operations across the cluster are coherent.

Master Node – Each cluster must elect a single master node responsible for coordinating the cluster and ensuring that each node contains an up-to-date copy of the cluster state.

Cluster Formation – OpenSearch requires a set of configurations to determine how the cluster is formed, which nodes can join the cluster, and how the nodes collectively elect a master node responsible for controlling the cluster state. These configurations are usually held in the opensearch.yml config file, environment variables on the node, or within the cluster state.

Node Roles – In small clusters it is common for all nodes to fill all roles; all nodes can store data, become master nodes or process ingestion pipelines. However as the cluster grows, it is common to allocate specific roles to specific nodes in order to simplify configuration and to make operation more efficient. In particular, it is common to define a limited number of dedicated master nodes.

Replication – Data may be replicated across a number of data nodes. This means that if one node goes down, data is not lost. It also means that a search request can be dealt with by more than one node.

Common problems

Many OpenSearch problems are caused by operations which place an excessive burden on the cluster because they require an excessive amount of information to be held and transmitted between the nodes as part of the cluster state. For example:

• Shards too small
• Too many fields (field explosion)

Problems may also be caused by inadequate configurations causing situations where the OpenSearch cluster is unable to safely elect a Master node. These type problems include: 

• Master node not discovered
• Split brain problem

Backups

Because OpenSearch is a clustered technology, it is not sufficient to have backups of each node’s data directory. This is because the backups will have been made at different times and so there may not be complete coherency between them. As such, the only way to backup an OpenSearch cluster is through the use of snapshots, which contain the full picture of an index at any one time.

Cluster resilience

When designing an OpenSearch cluster, it is important to think about cluster resilience. In particular – what happens when a single node goes down? And for larger clusters where several nodes may share common services such as a network or power supply – what happens if that network or power supply goes down? This is where it is useful to ensure that the master eligible nodes are spread across availability zones, and to use shard allocation awareness to ensure that shards are spread across different racks or availability zones in your data center.