×
Community Blog How to Mitigate High Disk Usage in Elasticsearch

How to Mitigate High Disk Usage in Elasticsearch

This guide delves into rectifying disk usage that trumps the watermark level and ensures uninterrupted data operations, using features and offerings of Alibaba Cloud Elasticsearch.

Elasticsearch clusters are dynamic systems that require close monitoring, especially concerning disk usage. Exceeding the designated watermark levels can prompt errors and disrupt the normal operations of an Elasticsearch instance. In this article, we will navigate through resolving high disk usage demonstrated on Alibaba Cloud Elasticsearch, which provides robust features to efficiently handle such scenarios.

Understanding Disk Watermark Levels

The Elasticsearch cluster uses three watermark levels to safeguard against excessive disk usage:

Property Name Default Value Description
cluster.routing.allocation.disk.watermark.low 85% Low watermark
cluster.routing.allocation.disk.watermark.high 90% High watermark
cluster.routing.allocation.disk.watermark.flood_stage 95% Flood stage watermark

When disk usage reaches the flood stage watermark (95%), Elasticsearch prevents any further writes to the disk to avoid a situation where the disk is completely full. This invokes a read-only mode for indices on the node.

Detecting and Analyzing Cluster Issues

To begin resolving issues, it's vital first to analyze our cluster's state by checking shard allocation:

GET _cat/shards?v=true

If shards are still allocated to a node nearing full capacity, further investigation is needed using the allocation explain API:

GET _cluster/allocation/explain
{
  "index": "my-index",
  "shard": 0,
  "primary": false,
  "current_node": "my-node"
}

Immediate Response: Adjusting Watermark Levels

To resume writing immediately, you might consider temporarily increasing the watermark levels and removing write blocks:

PUT _cluster/settings
{
  "persistent": {
    "cluster.routing.allocation.disk.watermark.low": "90%",
    "cluster.routing.allocation.disk.watermark.high": "95%",
    "cluster.routing.allocation.disk.watermark.flood_stage": "97%"
  }
}

And to remove the read-only block:

PUT */_settings?expand_wildcards=all
{
  "index.blocks.read_only_allow_delete": null
}

It's crucial to note that while the immediate response can quickly restore functionality, it is an interim solution.

Long-Term Solutions: Adding Nodes and Disk Expansion

For a sustainable resolution, consider adding nodes to your cluster or expanding existing nodes' disks. For instance, if a data_hot node is overwhelmed, you might:

  • Add a new hot node
  • Expand the disk of an existing hot node

Deleting older, unnecessary indices can also free up disk space:

DELETE my-index

Once you've implemented a long-term solution, you can reset the disk watermark levels:

PUT _cluster/settings
{
  "persistent": {
    "cluster.routing.allocation.disk.watermark.low": null,
    "cluster.routing.allocation.disk.watermark.high": null,
    "cluster.routing.allocation.disk.watermark.flood_stage": null
  }
}

Conclusion

Avoiding disk space crunches requires proactive measures, including regular monitoring and alerting for disk usage, and thoughtful planning with total_shards_per_node settings to balance shard distribution among nodes.

We invite users to share their experiences and solutions regarding disk space issues in Elasticsearch.

Alibaba Cloud’s Elasticsearch offers a comprehensive environment to implement and manage your indexes effectively, handling high-volume data with ease. Beginning your data journey is straightforward with our tailored cloud solutions providing everything necessary to transform your data architecture.

Embark on Your 30-Day Free Trial with Alibaba Cloud Elasticsearch and elevate your data operations today.

0 1 0
Share on

Data Geek

86 posts | 4 followers

You may also like

Comments

Data Geek

86 posts | 4 followers

Related Products