Check infrastructure health with CIT
Build a CIT test that verifies a resource handler’s connection, so you find a broken cloud before your users do
4 minute read
Continuous Infrastructure Testing (CIT) re-runs orders that previously succeeded and verifies they still succeed, so any change in the chain shows up as a failed test rather than a failed order for a user. Tests can run on a schedule or on demand, and Cloud Management Platform (CMP) reports on them and can email you when one fails.
CIT exercises the whole provisioning path through the CMP web interface, so a passing test proves every system involved in provisioning and decommissioning is working together. Depending on your environment, that can include networking, storage, virtualization hosts, public cloud providers, DNS, asset tracking, monitoring, backup, hostname reservation, and change management systems.
Order-based provision and decommission tests verify that CMP can:
Action tests cover anything not tied to provisioning, from checking the status of an external system to confirming that your own scripts still run.
A CIT test is built from either an order or an action.
Go to Admin > Continuous Infrastructure Testing.
Select Add, then select New Action Test.
Choose an existing action, or create one.
Go to Admin > Continuous Infrastructure Testing.
Select Execute Tests, then select Execute All Tests to run everything, Execute tests with labels to run a subset, Execute enabled, failing tests to run failing tests, or the play button on a single test.
Scheduled runs are driven by the Run all CIT Tests recurring job. Edit its schedule at Admin > Recurring Jobs; see Schedule format.
On a newly installed instance, the Run all CIT Tests job is disabled, so no tests run on schedule until you enable it. See Built-in recurring jobs.
Jobs and orders created by scheduled test runs have no owner by default. To give them one, select Edit CIT Settings on the Continuous Infrastructure Testing page and choose a default owner. Pick an account whose name makes clear no human started the run, such as cit_owner, and check that it has the permissions your tests need.
Tests run by hand are always owned by the user who started them, regardless of whether a default owner is set.
Set Max Retries on the test to have it re-run automatically until it passes or the limit is reached. A test that fails first and then passes is reported as a Warning rather than a Failure.
CMP waits 10 minutes before a retry. Change that under Edit CIT Settings.
CIT runs tests in serial unless you group them. Label a test serialgroup:<group name> and every test sharing that exact label runs in serial with the others in its group, while separate groups run in parallel. Unlabelled tests form an implicit group of their own, running in serial with one another alongside the labeled groups.
Give a test only one serialgroup: label. A test can belong to a single serial group.
The Progress section of the job details page shows the detailed output for each test, including what failed and links to the associated jobs.
The order behind a test cannot be edited, nor can the test itself. Work around it by rebuilding the test from a corrected order:
CMP emails Continuous Infrastructure Test results on failure.
Build a CIT test that verifies a resource handler’s connection, so you find a broken cloud before your users do