mongodb

Connect to MongoDB using a standard connection URI.

--quiet; --eval 'cmd'; --file script.js; --apiVersion 1/2

List databases; switch to one; list collections in the current db.

Server info: connections, opcounters, replication state; db stats: size & collections.

.pretty() to pretty-print; .verbose(); .json; --eval result

Query a collection; chain .pretty() .limit() .skip() .sort() to shape the result.

.projection({name:1, email:1, _id:0}); .sort({createdAt:-1}); .limit(20); .skip(20); .count(); .allowDiskUse()

Read a single document, or count docs matching a filter.

findOne with/without filter; estimatedDocumentCount is O(1)

Insert one or many documents; options: ordered:false writes all valid docs even if one fails.

Update one document. Use operators like $set $inc $push $pull $addToSet $rename $unset.

upsert:true; arrayFilters:[] for nested arrays; returnDocument:'after' (findOneAndUpdate)

Update many docs (write is bulk); findOneAndUpdate returns the doc pre/post update atomically.

Replace entire document except _id (not allowed in replacement).

Remove docs matching the filter. deleteMany({}) deletes everything (used with caution).

The standard comparison operators; $in matches a list of values.

Combine predicates; $not is regex/metadata only, regex needs explicit regex.

Field-level predicate: presence, BSON type, regex match, aggregation comparison ($expr), $mod.

Get unique values in a field (with optional filter).

Build a pipeline of stages: $match → $group → $project → $sort → $limit → $lookup → $unwind → $facet.

.explain('executionStats'); allowDiskUse:true; collation document

Join across collections ($lookup) and recursive tree walks ($graphLookup).

Useful aggregators: bucketing, sampling, top-N counting.

Create single/composite/geospatial/text/TTL indexes.

{unique:true, partialFilterExpression, sparse, expireAfterSeconds, weights, name}

List indexes; remove one / all (sometimes: just don't call this in prod!).

Inspect the winning plan & actual execution stats. Look for COLLSCAN, IXSCAN, FETCH, sort stage.

executionStats / queryPlanner / allPlansExecution; verbosity

Force the planner to use a particular index — debug bad plans, plan regressions.

Run MongoDB commands directly. escape for operations not in the shell helper.

Role management (run on admin DB).

Bootstrap a single-node replica set; show health.

rs.initiate({_id:'rs0',members:[{_id:0,host:'m1:27017'}]}); rs.status().myState

Add / remove members; force election to test failover.

rs.add('m4:27017'); rs.remove('m3:27017'); rs.stepDown(60)

Inspect the cluster topology / enable sharding on a database.

Logical backup / restore. Use --uri, --db, --archive to stream; --gzip for compressed.

--gzip; --archive=file.archive; --oplog for point-in-time; --nsInclude/Exclude

Export / import JSON or CSV. Stays as a separator engine for piping between Mongo and other systems.

--type=json|csv; --fields=a,b,c; --headerline; -q filter; --jsonArray (single top-level array)

Live monitoring: how much time each collection spent on read/write in the last second; statistics.

interval in seconds; --host

Run on admin DB to stop mongod cleanly — also useful after rm/poweroff to recover.

use admin; db.shutdownServer({force:true}? No; clean shutdown waits for oplog replication)