Skip to content

Latest commit

 

History

History
170 lines (104 loc) · 3.39 KB

TEST-PLAN.md

File metadata and controls

170 lines (104 loc) · 3.39 KB

MongoDB release test plan

Unit tests

Templates to tests

Included:

  • mongod.conf
  • mongod_bootstrap.conf
  • pre-start
  • create_admin_users.js
  • initiate_rs.js

Excluded:

  • mongoadmin.sh

Rendering technologies:

  1. Test templates rendering with render.rb (with bosh-template Gem v1.x).
  2. Test templates with bosh-template 2.0.0 Gem

Run ShellCheck on Bash scripts.

Configurations

  1. Mandatory

    • replication or standalone
    • test various engines: wiredTiger, mmapv1.
  2. Optional

    • test various engines: RocksDB
    • with SSL or not
    • config server
    • shard server

Acceptance tests

Write BOSH errand job.

Common tests

These test will be run for both Stand Alone and Replica Set setups.

These tests use the mongo CLI.

  1. Read/Write test
  • Before each:

    • connect as admin
    • create database
    • create user with admin role on the database
  • Test case:

    • connect as user
    • create collection
    • write some data in collection
    • read the data back
    • update the data
    • delete data
    • verify data is absent
  • After each:

    • drop database
    • drop user
  1. Dump/Restore
  • Test case:
    • write some data
    • mongo dump
    • drop database
    • mongo restore
    • read the data back
  1. Engine

    • Create stand alone node with specific engine
    • Verify the mongo server is using that specific engine
  2. Replication modes

    • Verify mongodb.replication.enable: true provides a replica set
    • Verify that mongodb.replication.enable: true with only one instance keeps the ReplicaSet active
    • Verify mongodb.replication.enable: false and role: sa creates a standalone node
    • Verify that having both mongodb.replication.enable: true and role: sa raises an error when deploying
  3. Replication

    • Insert some data on master node
    • Toggle slaves to slaveOK
    • verify data is present on slaves
  4. Soft Failover

    • Build a 3-nodes Replica Set
    • Turn the BOSH resurrector off
    • Find the primary node (using rs.status().isMaster() on each node)
    • Stop the primary node gracefully
    • Verify another node takes the master role over
  5. Cluster re-join after graceful stop

    • Same begining as above

    • Start the node back

    • verify it joins the cluster as secondary node

    • verify the data is here

  6. Dirty Failover

    • Build a 3-nodes Replica Set
    • Turn the BOSH resurrector on
    • Find the primary node (using rs.status().isMaster() on each node)
    • Kill the primary node (access the IaaS API)
    • Verify another node takes the master role over
  7. Cluster recovery

    • Same begining as above

    • Wait for the resurrector to recreate the killed node

    • verify

  8. Scale-in (general case)

    • Build a 3-nodes Replica Set
    • Scale in to 1 node
    • Verify that deleted nodes have properly left the cluster
  9. Scale-in (involing a master node)

    • Build a 3-nodes Replica Set
    • Have a non-bootstrap node take over the master role
    • Scale in to 1 node
    • Verify that the master role has been transfered to the node that is left
    • Verify that deleted nodes have properly left the cluster
  10. Scale-out

    • Build a 1-node Replica Set
    • Scale out to 3 nodes
    • Verify that new nodes join the cluster properly
  11. SSL

    • Verify SSL is properly implemented