Daniel Hilst
1a531128de
Appendix A decisions 9, 21, 22 and 24. Each one is a difference an
operator never chose: the backend is picked by distribution version, so
whichever side is wrong is wrong on half the clusters.
9 -- Kea never loaded libdhcp_bootp.so, so a BOOTP-only client that ISC
answered got nothing. The hook now sits alongside host_cmds in one
hooks-libraries array, and when the hooks package is absent the
operator is told which clients that leaves unanswered rather than
being left to find out from a node that never boots.
21 -- The ISC host statements sent "send host-name", which is a dhcpd
*client* keyword: it never reached option 12 and the node was
handed no hostname at all. Kea has always sent one.
22 -- The PXE short lease existed on ISC in name only. dhcpd applies
min-lease-time after max-lease-time, so the cluster default won and
a pool address taken by a PXE ROM was held for half a day. The
class now names all three bounds, and Kea is given the same 600
seconds through xcat-pxe-lease.
24 -- ISC writes "authoritative;" into every generated subnet; Kea
defaults to the opposite. A node that moved rack asked to keep an
address from the network it had left and was answered with silence,
so it waited out a lease that would never be renewed instead of
being told to start over.
Also makes $xCAT_plugin::dhcp::callback a package variable. It was a file
lexical, so the tests' local() set something nothing read, and warnings
were only ever captured because an earlier process_request had left its
collector behind.
xCAT-test/integration
Integration tests. These run against an installed management node -- they need a
real xCAT installation, and depending on the test a populated /install, a service
binary they can execute, or a live daemon.
They are driven by xcattest through the testcase in
../autotest/testcase/integration/, which proves the copy installed by the
xcat-test package:
prove -I/opt/xcat/lib/perl -I/opt/xcat/lib/perl/xCAT \
-r /opt/xcat/share/xcat/tools/autotest/integration
Run the case by hand on an MN with:
xcattest -f <cluster.conf> -t integration_tests
The case carries the ci_test label, so it also runs on every pull request: the
xcat_test GitHub Actions workflow installs and configures xCAT on the runner, which
makes that runner a (single node) management node, and then runs every ci_test case
against it.
Note the -I flags: unlike the unit tests these run from the installed location, so
they pick up xCAT modules from /opt/xcat/lib/perl rather than from a source tree.
The case checks rc==0 and output=~Files=5,. The second assertion is there because
prove exits 0 both when tests pass and when they all skip, so rc==0 alone would let
the case report green having run nothing. Matching Files=5, proves prove actually
found all five files, which catches a packaging regression or a test renamed away from
.t without updating the count. Add to that number when you add a test. A missing
directory is already caught by rc==0 -- prove -r on a path that does not exist exits
2.
Note also that github_action_xcat_test.pl invokes each case through sudo, so in CI
these tests run as root while the unit tests run unprivileged. That is the right
way round -- integration tests legitimately need to write to places like /etc/kea,
whereas running the unit tests as root would let permission-related assertions pass
for the wrong reason.
What belongs here
A test belongs in integration/ when it needs something the checkout cannot provide:
| Test | Requires |
|---|---|
copycds_packages_integrity.t |
/install populated by a real copycds |
dhcp_isc_config_validation.t |
a dhcpd binary that can read the generated config, as root |
dhcp_kea_config_validation.t |
a kea-dhcp4 binary that can read the generated config |
dhcp_kea_control_agent_smoke.t |
live kea-dhcp4 and kea-ctrl-agent, root, and the Kea host-commands hook |
html_form_runtime_dependency.t |
the installed HTML::Form module |
Environment guards
Tests with optional runtime dependencies guard with plan skip_all -- an MN with no
Kea installed should skip the Kea tests, not go red. The HTML::Form test is different:
it verifies a required installed dependency and fails when that module is absent. A
skip in this directory is therefore expected and normal.
Which tests actually run consequently varies by node. On a GitHub runner, for example,
/install is empty so copycds_packages_integrity.t skips, while
dhcp_kea_config_validation.t does run because the case executes as root and can
therefore validate from /etc/kea.
dhcp_kea_control_agent_smoke.t is opt-in on top of that:
plan skip_all => 'set XCAT_KEA_LIVE_SMOKE=1 to run live Kea daemon smoke test'
unless $ENV{XCAT_KEA_LIVE_SMOKE};
It starts real Kea daemons, so it stays off unless asked for. Do not enable it on a node whose DHCP service is in use.
What does not belong here
Anything that only needs the checkout. Those go in ../unit and
run on every pull request, which is much faster feedback than waiting for a cluster
test.