Hi, PR #38443[1] sets default value of parameter (of vstart_runner.LocalRemoteProcess._do_run()) "shell" to "True" because teuthology adds '&&'[2] to every command it executes. However, setting shell to True leads the ceph API CI job to fail without any error message or traceback in the log. Following is the stack of method calls through which the ceph API test crashes/quits - https://github.com/ceph/ceph/blob/master/qa/tasks/mgr/test_module_selftest.py#L238 https://github.com/ceph/ceph/blob/master/qa/tasks/ceph_test_case.py#L85 https://github.com/ceph/ceph/blob/master/qa/tasks/vstart_runner.py#L974 https://github.com/ceph/ceph/blob/master/qa/tasks/vstart_runner.py#L404 https://github.com/ceph/ceph/blob/master/qa/tasks/vstart_runner.py#L467 https://github.com/ceph/ceph/blob/master/qa/tasks/vstart_runner.py#L183 https://github.com/ceph/ceph/blob/master/qa/tasks/vstart_runner.py#L221 test_module_selftest.TestModuleSelftest.test_selftest_cluster_log crashes at the very first iteration of the loop in it[3]. I have been able to trace these exact details by creating a clone[4] of the original[1] PR and adding extra log messages to it. [5] is the ceph API CI job for this PR. An alternate way to avoid the job failure is by setting shell to False in the call to run_ceph_w()[6]. Commit here[7] got that done for the original PR. Thanks, - Rishabh [1] https://github.com/ceph/ceph/pull/38443 [2] https://github.com/ceph/teuthology/blob/master/teuthology/orchestra/run.py#L70 [3] https://github.com/ceph/ceph/blob/master/qa/tasks/mgr/test_module_selftest.py#L230 [4] https://github.com/ceph/ceph/pull/38471 [5] https://jenkins.ceph.com/job/ceph-api/10648/consoleFull#-85362248744e9240e-b50a-4693-bac0-8a991bac86ac [6] https://github.com/ceph/ceph/blob/master/qa/tasks/ceph_test_case.py#L109 [7] https://github.com/ceph/ceph/pull/38443/commits/c7c5ad878e06d0fd76a43bcfa890f435af110b4d _______________________________________________ Dev mailing list -- dev@xxxxxxx To unsubscribe send an email to dev-leave@xxxxxxx
