)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":15343,"name":"Tim Burke","email":"tburke@nvidia.com","username":"tburke"},"change_message_id":"da00b8a7fd5d5e9980d27a0b10eba32d75119ad9","unresolved":false,"context_lines":[{"line_number":20,"context_line":"Drive-bys:"},{"line_number":21,"context_line":""},{"line_number":22,"context_line":" * Don\u0027t make concurrent_ec_extra_requests unless you have enabled"},{"line_number":23,"context_line":"   concurrent_gets."},{"line_number":24,"context_line":" * Improved mock_http_connect extra requests tracking formatting"},{"line_number":25,"context_line":" * FakeStatus __repr__\u0027s w/ status code in AssertionErrors"},{"line_number":26,"context_line":""}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"9f560f44_a036dcb2","line":23,"updated":"2020-09-25 15:41:34.000000000","message":"I guess the argument is that that\u0027s how all the other concurrent_* options work?\n\nAt what point can/should we default concurrent_gets to True? How long after that until we can get rid of the option entirely?","commit_id":"5f95e1becec55f422bc27cfbfcfef43854c76be7"},{"author":{"_account_id":1179,"name":"Clay Gerrard","email":"clay.gerrard@gmail.com","username":"clay-gerrard"},"change_message_id":"226fc6135204c6d79d604f85e2c3264d96952f6f","unresolved":false,"context_lines":[{"line_number":20,"context_line":"Drive-bys:"},{"line_number":21,"context_line":""},{"line_number":22,"context_line":" * Don\u0027t make concurrent_ec_extra_requests unless you have enabled"},{"line_number":23,"context_line":"   concurrent_gets."},{"line_number":24,"context_line":" * Improved mock_http_connect extra requests tracking formatting"},{"line_number":25,"context_line":" * FakeStatus __repr__\u0027s w/ status code in AssertionErrors"},{"line_number":26,"context_line":""}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"9f560f44_82cee4b4","line":23,"in_reply_to":"9f560f44_a036dcb2","updated":"2020-10-01 21:08:07.000000000","message":"yeah, we should change the default!  orthogonal to the bug i\u0027m trying to fix tho\n\nmaybe I shouldn\u0027t be fixing two bugs at once, but I thought it made the code slightly cleaner.","commit_id":"5f95e1becec55f422bc27cfbfcfef43854c76be7"}],"swift/proxy/controllers/obj.py":[{"author":{"_account_id":15343,"name":"Tim Burke","email":"tburke@nvidia.com","username":"tburke"},"change_message_id":"da00b8a7fd5d5e9980d27a0b10eba32d75119ad9","unresolved":false,"context_lines":[{"line_number":2910,"context_line":"        ec_request_count \u003d policy.ec_ndata"},{"line_number":2911,"context_line":"        if policy_options.concurrent_gets:"},{"line_number":2912,"context_line":"            ec_request_count +\u003d policy_options.concurrent_ec_extra_requests"},{"line_number":2913,"context_line":"        with ContextPool(policy.ec_n_unique_fragments) as pool:"},{"line_number":2914,"context_line":"            pile \u003d GreenAsyncPile(pool)"},{"line_number":2915,"context_line":"            buckets \u003d ECGetResponseCollection(policy)"},{"line_number":2916,"context_line":"            node_iter.set_node_provider(buckets.provide_alternate_node)"}],"source_content_type":"text/x-python","patch_set":1,"id":"9f560f44_405a601d","line":2913,"range":{"start_line":2913,"start_character":32,"end_line":2913,"end_character":53},"updated":"2020-09-25 15:41:34.000000000","message":"Could we run into a similar problem with multi-region EC?","commit_id":"5f95e1becec55f422bc27cfbfcfef43854c76be7"},{"author":{"_account_id":15343,"name":"Tim Burke","email":"tburke@nvidia.com","username":"tburke"},"change_message_id":"18dc46244dd2a7fbb80280aeda84bb7487616d51","unresolved":false,"context_lines":[{"line_number":2910,"context_line":"        ec_request_count \u003d policy.ec_ndata"},{"line_number":2911,"context_line":"        if policy_options.concurrent_gets:"},{"line_number":2912,"context_line":"            ec_request_count +\u003d policy_options.concurrent_ec_extra_requests"},{"line_number":2913,"context_line":"        with ContextPool(policy.ec_n_unique_fragments) as pool:"},{"line_number":2914,"context_line":"            pile \u003d GreenAsyncPile(pool)"},{"line_number":2915,"context_line":"            buckets \u003d ECGetResponseCollection(policy)"},{"line_number":2916,"context_line":"            node_iter.set_node_provider(buckets.provide_alternate_node)"}],"source_content_type":"text/x-python","patch_set":1,"id":"9f560f44_c842bb68","line":2913,"range":{"start_line":2913,"start_character":32,"end_line":2913,"end_character":53},"in_reply_to":"9f560f44_22def869","updated":"2020-10-02 04:25:51.000000000","message":"But like, if you\u0027ve got a proxy with write affinity misconfigured so it\u0027s starting out asking for everything over the WAN and of course it takes a while... do we really not want to spin up more requests? I\u0027m trying to think of why we *don\u0027t* want this to just be replica_count... I know the extra connections aren\u0027t *free*, but I feel like if we\u0027re digging that deep and it\u0027s actually causing a problem, maybe we ought to be increasing the concurrency timeout instead...","commit_id":"5f95e1becec55f422bc27cfbfcfef43854c76be7"},{"author":{"_account_id":1179,"name":"Clay Gerrard","email":"clay.gerrard@gmail.com","username":"clay-gerrard"},"change_message_id":"226fc6135204c6d79d604f85e2c3264d96952f6f","unresolved":false,"context_lines":[{"line_number":2910,"context_line":"        ec_request_count \u003d policy.ec_ndata"},{"line_number":2911,"context_line":"        if policy_options.concurrent_gets:"},{"line_number":2912,"context_line":"            ec_request_count +\u003d policy_options.concurrent_ec_extra_requests"},{"line_number":2913,"context_line":"        with ContextPool(policy.ec_n_unique_fragments) as pool:"},{"line_number":2914,"context_line":"            pile \u003d GreenAsyncPile(pool)"},{"line_number":2915,"context_line":"            buckets \u003d ECGetResponseCollection(policy)"},{"line_number":2916,"context_line":"            node_iter.set_node_provider(buckets.provide_alternate_node)"}],"source_content_type":"text/x-python","patch_set":1,"id":"9f560f44_22def869","line":2913,"range":{"start_line":2913,"start_character":32,"end_line":2913,"end_character":53},"in_reply_to":"9f560f44_405a601d","updated":"2020-10-01 21:08:07.000000000","message":"well, but you never need more than ec_n_unique_fragments pending\n\nagain it all works out as soon as responses start coming back, the contextpool is only limiting in-flight spawns to the pile","commit_id":"5f95e1becec55f422bc27cfbfcfef43854c76be7"}]}