Databases & Analytics - Handling «Disk Full» situations

Learn about how Cloud Databases and Analytics services behave before and when reaching such conditions, and what you can do about it.

No matter the engine technology, when no more physical disk space is available the service operation degrades significantly. At the very least your services cannot store any more data, but even logical read operations might start to get impacted, for example, querying might slow down or fail.

Requirements

• Access to the OVHcloud Control Panel
• A Cloud Database and Analytics service up and running

Instructions

Avoiding full disk conditions

Disk space usage metrics

As part of using your Cloud Databases and Analytics service efficiently, you should keep an eye on the service metrics. You can access those in the OVHcloud Control Panel or using the API. You can also make use of cross-service integrations to gather, observe, and alert based on service metrics.

Mail notifications

When your service storage begins to fill up and reaches a high mark, your service sends you an email to warn you of the situation. The specific threshold depends on the engine, it may range from 75 to 90 percent.

When the disk usage increases even more and reaches a critical level (depending on the engine, ranging from 90 to 95 percent), you will receive another email notification and the service will turn to a "disk full" mode, where it will start to refuse writes.

How to handle a disk full situation?

Different engines react in different ways, thus your services react differently when facing disk full conditions:

• Caching, Kafka MirrorMaker, Grafana®¹, and Kafka Connect do not store any user data on disk. Thus they will not fill up the underlying disk storage.
• Kafka, OpenSearch, and Cassandra turn to read-only.
• MySQL and PostgreSQL turn to read-only with a way to temporarily revert to read-write.
• MongoDB forbids writes but allows deletes.

Upgrading your service

It may be that your usage simply requires more storage. You can then increase the provisioned storage and/or upgrade to an offer with more storage. Once the upgrade finishes, the service will detect that more storage is available and thus revert to normal mode.

Reclaim disk space

It may be that you have reached the full disk situation because of a runaway application filling up your service, or that you are storing some old obsolete data. In these cases, stop whatever process is unduly filling up your storage, then remove unwanted data.

Kafka, OpenSearch

You can reclaim disk space by deleting a Kafka topic or an OpenSearch index.

MongoDB

MongoDB refuses any query that inserts data but allows queries deleting data. You can execute any MongoDB command that allows you to reclaim disk space.

PostgreSQL, MySQL

For these engines, call the respective API endpoint to temporarily allow write operations:

• PostgreSQL:
  POST /cloud/project/{serviceName}/database/postgresql/{clusterId}/enableWrites
• MySQL:
  POST /cloud/project/{serviceName}/database/mysql/{clusterId}/enableWrites

This will give you a 15-minute time window to write again to your database. At the end of this window, either you were able to execute queries that reduce disk usage (e.g. DROPs, DELETEs) allowing your service changes to the read-write state, or disk usage stays too high and your service will return to the read-only state.

Go further

For more information and tutorials, please see our other Managed Databases & Analytics or Platform as a Service guides. You can also explore the guides for other OVHcloud products and services.

[1]: The Grafana Labs Marks are trademarks of Grafana Labs, and are used with Grafana Labs’ permission. We are not affiliated with, endorsed or sponsored by Grafana Labs or its affiliates.