Skip to content

Frequently Asked Questions

Why do we need to follow “the Kubernetes way” when Kubernetes was never intended to run databases?

As it is well known, the Kubernetes approach is targeted at stateless applications but provides ways to store state (in Persistent Volumes, etc.) if the application needs it. Generally, a stateless mode of operation is supposed to provide better safety, sustainability, and scalability, it makes the already-deployed components interchangeable. You can find more about substantial benefits brought by Kubernetes to databases in this blog post.

The architecture of state-centric applications (like databases) should be composed in a right way to avoid crashes, data loss, or data inconsistencies during hardware failure. Percona Operator for MongoDB provides out-of-the-box functionality to automate provisioning and management of highly available MongoDB database clusters on Kubernetes.

How can I contact the developers?

The best place to discuss Percona Operator for MongoDB with developers and other community members is the community forum.

If you would like to report a bug, use the Percona Operator for MongoDB project in JIRA.

What is the difference between the Operator quickstart and advanced installation ways?

As you have noticed, the installation section of docs contains both quickstart and advanced installation guides.

The quickstart guide is simpler. It has fewer installation steps in favor of predefined default choices. Particularly, in advanced installation guides, you separately apply the Custom Resource Definition and Role-based Access Control configuration files with possible edits in them. At the same time, quickstart guides rely on the all-inclusive bundle configuration.

At another point, quickstart guides are related to specific platforms you are going to use (Minikube, Google Kubernetes Engine, etc.) and therefore include some additional steps needed for these platforms.

Generally, rely on the quickstart guide if you are a beginner user of the specific platform and/or you are new to the Percona Operator for MongoDB as a whole.

Which versions of MongoDB does the Operator support?

Percona Operator for MongoDB provides a ready-to-use installation of the MongoDB-based database cluster inside your Kubernetes installation. It works with Percona Server for MongoDB 4.2, and 4.4, and the exact version is determined by the Docker image in use.

Percona-certified Docker images used by the Operator are listed here. For example, Percona Server for MongoDB 4.4 is supported with the following recommended version: 4.4.24-23. More details on the exact Percona Server for MongoDB version can be found in the release notes (5.0, 4.4, and 4.2).

How can I add custom sidecar containers to my cluster?

The Operator allows you to deploy additional (so-called sidecar) containers to the Pod. You can use this feature to run debugging tools, some specific monitoring solutions, etc. Add such sidecar container to the deploy/cr.yaml configuration file, specifying its name and image, and possibly a command to run:

spec:
  replsets:
  - name: rs0
    ....
    sidecars:
    - image: busybox
      command: ["/bin/sh"]
      args: ["-c", "while true; do echo echo $(date -u) 'test' >> /dev/null; sleep 5; done"]
      name: rs-sidecar-1
    ....

You can add sidecars subsection to replsets, sharding.configsvrReplSet, and sharding.mongos sections.

Note

Custom sidecar containers can easily access other components of your cluster. Therefore they should be used carefully and by experienced users only.

Find more information on sidecar containers in the appropriate documentation page.

How to provoke the initial sync of a Pod?

There are certain situations where it might be necessary to delete all MongoDB instance data to force the resync. For example, there may be the following reasons:

  • rebuilding the node to defragment the database,
  • recreating the member failing to sync due to some bug.

In the case of a “regular” MongoDB, wiping the dbpath would trigger such resync. In the case of a MongoDB cluster controlled by the Operator, you will need to do the following steps:

  1. Find out the names of the Persistent Volume Claim and Pod you are going to delete (use kubectl get pvc command for PVC and kubectl get pod one for Pods).
  2. Delete the appropriate PVC and Pod. For example, wiping out the my-cluster-name-rs0-2 Pod should look as follows:

    $ kubectl delete pod/my-cluster-name-rs0-2 pvc/mongod-data-my-cluster-name-rs0-2
    

The Operator will automatically recreate the needed Pod and PVC after deletion.

Get expert help

If you need assistance, visit the community forum for comprehensive and free database knowledge, or contact our Percona Database Experts for professional support and services. Join K8S Squad to benefit from early access to features and “ask me anything” sessions with the Experts.


Last update: 2024-03-28