)]}'
{"tempest/stress/driver.py":[{"author":{"_account_id":8641,"name":"DennyZhang","email":"filebat.mark@gmail.com","username":"dennyzhang"},"change_message_id":"1b36cf985b931a08eaa7243f4e7c3266e5073e9c","unresolved":false,"context_lines":[{"line_number":196,"context_line":""},{"line_number":197,"context_line":"        if not logfiles:"},{"line_number":198,"context_line":"            continue"},{"line_number":199,"context_line":"        if _has_error_in_logs(logfiles, computes, stop_on_error):"},{"line_number":200,"context_line":"            had_errors \u003d True"},{"line_number":201,"context_line":"            break"},{"line_number":202,"context_line":""}],"source_content_type":"text/x-python","patch_set":1,"id":"AAAAS3%2F%2FGXY%3D","line":199,"updated":"2013-10-04 12:26:07.000000000","message":"I\u0027d suggest to change to: _has_error_in_logs(logfiles, computes, False).\n\nstop_on_error in _has_error_in_logs means whether we shall check all nodes for error logs, which is different from stop_on_error in stress_openstack.\n\nIt would be better, if we dump all the error messages in different nodes, since the performance penalty for this is not huge.","commit_id":"4ea019472dd3ec1868ab22bbeccc81d4ecfc3749"},{"author":{"_account_id":7872,"name":"Marc Koderer","email":"marc@koderer.com","username":"mkoderer"},"change_message_id":"efbf2cbd1c1defeaf4aca6aa584edd0adac6340c","unresolved":false,"context_lines":[{"line_number":196,"context_line":""},{"line_number":197,"context_line":"        if not logfiles:"},{"line_number":198,"context_line":"            continue"},{"line_number":199,"context_line":"        if _has_error_in_logs(logfiles, computes, stop_on_error):"},{"line_number":200,"context_line":"            had_errors \u003d True"},{"line_number":201,"context_line":"            break"},{"line_number":202,"context_line":""}],"source_content_type":"text/x-python","patch_set":1,"id":"AAAAS3%2F%2FDzc%3D","line":199,"in_reply_to":"AAAAS3%2F%2FET4%3D","updated":"2013-10-05 12:49:19.000000000","message":"Hi folks, the stop_on_error argument is used to stop the load of the system on the first occurrence of a failure. So I agree to Giulio in that point. If one of the nodes is not reachable in the list you need to wait the full ssh timeout period.\n\nIf we want to make it more flexible we could add a global parameter (like collect_all_logs) in tempest.conf.","commit_id":"4ea019472dd3ec1868ab22bbeccc81d4ecfc3749"},{"author":{"_account_id":8641,"name":"DennyZhang","email":"filebat.mark@gmail.com","username":"dennyzhang"},"change_message_id":"b61fa0f1baccacaee797de780d9797478c27c8a4","unresolved":false,"context_lines":[{"line_number":196,"context_line":""},{"line_number":197,"context_line":"        if not logfiles:"},{"line_number":198,"context_line":"            continue"},{"line_number":199,"context_line":"        if _has_error_in_logs(logfiles, computes, stop_on_error):"},{"line_number":200,"context_line":"            had_errors \u003d True"},{"line_number":201,"context_line":"            break"},{"line_number":202,"context_line":""}],"source_content_type":"text/x-python","patch_set":1,"id":"AAAAS3%2F%2FET4%3D","line":199,"in_reply_to":"AAAAS3%2F%2FEYo%3D","updated":"2013-10-04 23:37:31.000000000","message":"Giulio, current change actually change be behavior from _has_error_in_logs(logfiles, computes, False) to _has_error_in_logs(logfiles, computes, stop_on_error).\n\nThus I vote -1.\n\nIf the performance overhead is big enough, I would agree with you. Both way is ok for me.","commit_id":"4ea019472dd3ec1868ab22bbeccc81d4ecfc3749"},{"author":{"_account_id":6796,"name":"Giulio Fidente","email":"gfidente@redhat.com","username":"gfidente"},"change_message_id":"b6734368154e7511bdbaacafc91f453603ebb7fb","unresolved":false,"context_lines":[{"line_number":196,"context_line":""},{"line_number":197,"context_line":"        if not logfiles:"},{"line_number":198,"context_line":"            continue"},{"line_number":199,"context_line":"        if _has_error_in_logs(logfiles, computes, stop_on_error):"},{"line_number":200,"context_line":"            had_errors \u003d True"},{"line_number":201,"context_line":"            break"},{"line_number":202,"context_line":""}],"source_content_type":"text/x-python","patch_set":1,"id":"AAAAS3%2F%2FEYo%3D","line":199,"in_reply_to":"AAAAS3%2F%2FF2I%3D","updated":"2013-10-04 22:42:32.000000000","message":"Okay sorry for having worded badly my thinking. \"collecting the logs\" was not appropriate and to avoid any further misunderstanding, yes I agree we\u0027re only debating about logging the errors from all nodes or not.\n\nThis said, I think stopping as soon as am error is found in the logs is perfectly fine.\n\nLogging the error is helpful, maybe I\u0027ll have to check the logs myself anyway, but logging is good.\n\nLogging from all nodes, meh, it introduces some performance penalty especially when nodes aren\u0027t just a few and yet I\u0027m not how many chances there are it\u0027ll find errors (actually different errors) on different nodes in the same loop.\n\nStill, I can live happy with stop_on_error set to False by default and eventually let the loop run but we should at least provide for a way to control that as this is changing the initial behaviour. The initial code was just stopping as soon as an error was found.\n\nMy submission is just a fix to a problem introduced earlier, where we did not provide anything to control such a behaviour.\n\nSo I don\u0027t see why you\u0027re -1","commit_id":"4ea019472dd3ec1868ab22bbeccc81d4ecfc3749"},{"author":{"_account_id":8641,"name":"DennyZhang","email":"filebat.mark@gmail.com","username":"dennyzhang"},"change_message_id":"b8dd0569ad524c921f403565c598c09a778e0cee","unresolved":false,"context_lines":[{"line_number":196,"context_line":""},{"line_number":197,"context_line":"        if not logfiles:"},{"line_number":198,"context_line":"            continue"},{"line_number":199,"context_line":"        if _has_error_in_logs(logfiles, computes, stop_on_error):"},{"line_number":200,"context_line":"            had_errors \u003d True"},{"line_number":201,"context_line":"            break"},{"line_number":202,"context_line":""}],"source_content_type":"text/x-python","patch_set":1,"id":"AAAAS3%2F%2FF2I%3D","line":199,"in_reply_to":"AAAAS3%2F%2FGDA%3D","updated":"2013-10-04 15:19:52.000000000","message":"Guilio, thanks for clarify the two requirements of when to stop on test errors and when to not collect logs from all nodes in _has_error_in_logs function.\n\n_has_error_in_logs, simply check whether some nodes has error log, and report the status of the node. And nothing more.\n\nThus IMO, we don\u0027t need to distinguish the two requirements.\n\nMy questions is whether you agree we should log errors in all nodes for stress_openstack function, or let it depends on the stop_on_error parameter of stress_openstack.","commit_id":"4ea019472dd3ec1868ab22bbeccc81d4ecfc3749"},{"author":{"_account_id":6796,"name":"Giulio Fidente","email":"gfidente@redhat.com","username":"gfidente"},"change_message_id":"72c90e4b5929f92f245f9aeaa41e38a4562698e7","unresolved":false,"context_lines":[{"line_number":196,"context_line":""},{"line_number":197,"context_line":"        if not logfiles:"},{"line_number":198,"context_line":"            continue"},{"line_number":199,"context_line":"        if _has_error_in_logs(logfiles, computes, stop_on_error):"},{"line_number":200,"context_line":"            had_errors \u003d True"},{"line_number":201,"context_line":"            break"},{"line_number":202,"context_line":""}],"source_content_type":"text/x-python","patch_set":1,"id":"AAAAS3%2F%2FGDA%3D","line":199,"in_reply_to":"AAAAS3%2F%2FGXY%3D","updated":"2013-10-04 14:25:12.000000000","message":"I agree with you on the meaning of \"stop_on_error\" but I think reusing it for the purpose of not collecting the logs is acceptable.\n\nAlso if you did not think so in the first place, then why would you accept that as parameter in _has_error_in_logs ?\n\nThe only other solution is see is an additional parameter intended to distinguish when to stop on test errors and when to not collect logs from all nodes; looks overkilling to me.","commit_id":"4ea019472dd3ec1868ab22bbeccc81d4ecfc3749"}]}