Enmasse backup

[pb82]

Creates the enmasse backup cronjob.

Sets up a CronJob to backup the kubernetes resources.

Verification:

To test that the files in the PV are backed up you need a brokered AMQ:

1. Go to the service catalog and select AMQ Online (brokered). Pick the enmasse namespace and a name and provision the brokered quee.
2. In your enmasse namespace you should see an additional pod for the brokered queue.
3. Open the terminal of this pod and go to /var/run/artemis
4. Create a file in the, like echo "test" > /var/run/artemis/test.txt

Then you can verify the backup using:

1. Make sure that the s3-credentials secret exists in the default namespace
2. Update the master URL in managed.template and set core_install=false if you've already setup integreatly
3. Set backup_schedule in manifest.yaml to '*/1 * * * *' to run each minute
4. Run: ansible-playbook -i inventories/managed.template playbooks/managed/install.yml
5. Wait a minute

In the enmasse namespace
1.1 Ensure the postgres-backup Pod succeeds
1.1 Ensure the PV backup succeeds and the archive contains the file generated for the brokered pod.
1.1 Check S3 and ensure the archives have been pushed up.
1.1 Delete the CronJobs in all namespaces

[david-martin]

@pb82 I'd prefer to avoid this pattern during install.
If the install failed for some reason, and had to be re-run, it should be idempotent i.e. not remove or recreate anything, but create anything that is missing.

This has caused issues for us in the past e.g. #73 & #402

[pb82]

Interesting, the idea was to make sure the job gets updated when the properties change but for that we can use apply. Will update this.

[matskiv]

👀

[matskiv]

Since we are creating all CronJobs in default namespace, two names above should be prepended with 'enmasse-', just to avoid any ambiguity.

[matskiv]

👍 Verified