Skip to content
Documentation
CLI
Check indexes

Check indexes

Overview

Checks a destination MongoDB database for missing indexes against a recommended index set, reports differences, and optionally creates missing indexes.

Prerequisites

  • A reachable MongoDB instance with appropriate credentials.
  • An index source file exported from a reference database (see below).

Kubernetes

Run the checker as a Kubernetes Job using the provided template. The Job image wraps the CLI and accepts the same flags via environment variables.

You can find the .yaml template for this job in this source file.

kubectl apply -f jobs/check-indexes.yaml

Usage

Run in dry-run mode to preview changes:

go run main.go -action check-indexes \
  -mongodb-uri "mongodb+srv://user:pass@cluster/?retryWrites=true&w=majority" \
  -mongodb-destination-database Kerberos \
  -collections "" \
  -index-version hub-08-12-2025 \
  -mode dry-run

Apply missing indexes in live mode:

go run main.go -action check-indexes \
  -mongodb-uri "mongodb+srv://user:pass@cluster/?retryWrites=true&w=majority" \
  -mongodb-destination-database Kerberos \
  -collections organisation,cache,counting \
  -index-version hub-08-12-2025 \
  -mode live

Arguments

  • -mongodb-uri: MongoDB connection URI with credentials (required).
  • -mongodb-destination-database: Destination database name to check (required).
  • -collections: Comma-separated list of collections to include. Defaults to all when omitted.
  • -mode: One of dry-run (no changes, report only) or live (create missing indexes).
  • -index-version: Version key used to resolve the index source file containing the recommended indexes.

Index Source File

You can find available index versions in the indexes directory on GitHub. The source file should be a plain text dump of collection names followed by their index definitions from the reference database. To generate it, use the MongoDB shell:

use("Kerberos");

db.getCollectionNames().forEach(function(collectionName) {
  print(collectionName);
  printjson(db.getCollection(collectionName).getIndexes());
});

Example excerpt:

organisation
[ { v: 2, key: { _id: 1 }, name: '_id_' } ]
cache
[
  { v: 2, key: { _id: 1 }, name: '_id_' },
  { v: 2, key: { key: 1, expiry: 1 }, name: 'key_1_expiry_1' },
  { v: 2, key: { key: 1 }, name: 'key_1' }
]
counting
[
  { v: 2, key: { _id: 1 }, name: '_id_' },
  { v: 2, key: { user_id: 1, timestamp: 1 }, name: 'user_id_1_timestamp_1' },
  { v: 2, key: { timestamp: 1 }, name: 'timestamp_1' }
]

How It Works

  • Parses the index source file into a map of collections → index definitions.
  • Fetches current indexes from the destination database.
  • Compares by index key and name to find missing indexes.
  • In dry-run, prints a report; in live, creates missing indexes where permitted.

Output

  • Missing indexes listed per collection.
  • Status codes per operation (created, existing, skipped on error).
  • Summary totals at the end in both modes.

Notes

  • Creating indexes may lock or impact performance; prefer running live during low traffic windows.
  • Ensure roles permit createIndex on target collections.
  • If -collections is omitted, all collections present in the source file are considered.