Intermittent CI test failure

[esabol]

This test in t/client suddenly has started to fail frequently but not always:

tests/libgearman-1.0/client_test.cc:821: in hostname_resolution() pid(37552) Assertion 'GEARMAN_COULD_NOT_CONNECT' != 'gearman_client_echo(&client, test_literal_param("foo"))' client.gearman_client_st.gearman_client_add_servers(GEARMAN_GETADDRINFO) [ failed ]

That corresponds to this code:

gearmand/tests/libgearman-1.0/client_test.cc

Lines 821 to 822 in c24d899

	ASSERT_EQ(GEARMAN_COULD_NOT_CONNECT,
	gearman_client_echo(&client, test_literal_param("foo")));

In what scenario would that ASSERT_EQ() fail?

[esabol]

Actually, more often than not the CI workflow will just hang for a long time – 30+ minutes is not uncommon – before it finally fails that test.

[esabol]

A couple lines earlier before the test that intermittently fails is this skipped test:

gearmand/tests/libgearman-1.0/client_test.cc

Line 813 in c24d899

test_skip(GEARMAN_SUCCESS, gearman_client_add_servers(&client, "exist.gearman.info:12345"));

Does a skipped test still execute the code? I think so? test_skip is a #define which does a do { ... } while (0);

So that means the failing test is trying to send foo to exist.gearman.info:12345?

[esabol]

It took 12 attempts, but I finally got all of the CI workflow jobs to pass.

I'm tempted to just disable all of these hostname_resolution() tests in CI.

[esabol]

I suspect this issue may have been fixed by changing the *.gearman.info hostnames in the test suite to *.gearman.org, but I'll leave this open until I have more data.

EDIT: Yes, I'm 99% sure that was it. The CI tests are much more robust after making this change.

[esabol]

There's a new intermittent CI test failure:

t/round_robin apparently failed intermittently (only once so far) with a segmentation fault on Alpine 3.21. If anyone wants to help track that down, your assistance would be appreciated.

FAIL: t/round_robin
===================

round_robin.round_robin.add					0:001356165 [ ok ]
round_robin.round_robin.worker					0:001082215 [ ok ]
round_robin.--job-retries=1.GEARMAN_FATAL					0:011416687 [ ok ]
round_robin.--job-retries=1.GEARMAN_SUCCESS					0:001280151 [ ok ]
round_robin.--job-retries=1.limit					[ skipped ]
round_robin.--job-retries=2.GEARMAN_FATAL					0:021848231 [ ok ]
round_robin.--job-retries=2.GEARMAN_SUCCESS					0:011624626 [ ok ]
round_robin.--job-retries=2.limit					0:001276899 [ ok ]
Segmentation fault (core dumped)
FAIL t/round_robin (exit status: 139)