)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":2889,"name":"Aeva Black","email":"aeva.online@gmail.com","username":"tenbrae"},"change_message_id":"4c52cb39f7f4ee7bb54d22cb1d53630479e467e7","unresolved":false,"context_lines":[{"line_number":4,"context_line":"Commit:     Josh Gachnang \u003cjosh@pcsforeducation.com\u003e"},{"line_number":5,"context_line":"CommitDate: 2015-04-10 15:23:02 -0700"},{"line_number":6,"context_line":""},{"line_number":7,"context_line":"Fix stuck locks for async clean steps"},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"RPC.call() was causing some nodes to become permanently locked using"},{"line_number":10,"context_line":"the agent driver during cleaning. The agent driver would issue a call() to"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"7aaa499b_d9682df8","line":7,"updated":"2015-04-13 16:40:32.000000000","message":"I think the commit header would be clearer if it said what is changed, eg:\n\nConvert internal RPC continue_node_cleaning to a \"cast\"","commit_id":"63df1d3fe019c9274784160b2558b76bd7d7930c"},{"author":{"_account_id":2889,"name":"Aeva Black","email":"aeva.online@gmail.com","username":"tenbrae"},"change_message_id":"4c52cb39f7f4ee7bb54d22cb1d53630479e467e7","unresolved":false,"context_lines":[{"line_number":6,"context_line":""},{"line_number":7,"context_line":"Fix stuck locks for async clean steps"},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"RPC.call() was causing some nodes to become permanently locked using"},{"line_number":10,"context_line":"the agent driver during cleaning. The agent driver would issue a call() to"},{"line_number":11,"context_line":"continue_node_cleaning() after either the first heartbeat"},{"line_number":12,"context_line":"(from prepare_cleaning) or a heartbeat after a clean step had completed."}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"7aaa499b_b96d6106","line":9,"updated":"2015-04-13 16:40:32.000000000","message":"\"call\" isn\u0027t really at fault here, it\u0027s the way the agent driver is invoking a method in the conductor that is awkward. Converting to a \"cast\" works around this by simply ignoring the errors. But the commit message shouldn\u0027t start out by saying that RPC.call() is somehow causing a problem.","commit_id":"63df1d3fe019c9274784160b2558b76bd7d7930c"},{"author":{"_account_id":2889,"name":"Aeva Black","email":"aeva.online@gmail.com","username":"tenbrae"},"change_message_id":"4c52cb39f7f4ee7bb54d22cb1d53630479e467e7","unresolved":false,"context_lines":[{"line_number":11,"context_line":"continue_node_cleaning() after either the first heartbeat"},{"line_number":12,"context_line":"(from prepare_cleaning) or a heartbeat after a clean step had completed."},{"line_number":13,"context_line":"The conductor would attempt to get a lock, but would not be able to."},{"line_number":14,"context_line":"The node would retain its locked state (so far as I could tell), even"},{"line_number":15,"context_line":"after the error. Other nodes would continue and complete cleaning"},{"line_number":16,"context_line":"just fine."},{"line_number":17,"context_line":""}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"7aaa499b_9972a5e4","line":14,"updated":"2015-04-13 16:40:32.000000000","message":"This may well be because the RPC client (inside the driver) was raising an exception that was getting ignored because the conductor thta called it is also ignoring this exception ... it\u0027s somewhat recursive ...","commit_id":"63df1d3fe019c9274784160b2558b76bd7d7930c"},{"author":{"_account_id":2889,"name":"Aeva Black","email":"aeva.online@gmail.com","username":"tenbrae"},"change_message_id":"4c52cb39f7f4ee7bb54d22cb1d53630479e467e7","unresolved":false,"context_lines":[{"line_number":18,"context_line":"Switching to cast() seems to fix the issue and seems more reasonable."},{"line_number":19,"context_line":""},{"line_number":20,"context_line":"Change-Id: I4dbb04ccb93199bba4e1a1614bc19b70a068a9ea"},{"line_number":21,"context_line":"Closes-Bug: 1442810"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"7aaa499b_7978f9c4","line":21,"updated":"2015-04-13 16:40:32.000000000","message":"I agree with lucas\u0027 assessment here: cast() is the right thing to do because you do not actually want any return from this RPC message. It\u0027s really a work-around to allow a driver to invoke the ConductorManager, and that should be documented as such.","commit_id":"63df1d3fe019c9274784160b2558b76bd7d7930c"},{"author":{"_account_id":6773,"name":"Lucas Alvares Gomes","email":"lucasagomes@gmail.com","username":"lucasagomes"},"change_message_id":"2a164c4b8ad438f15f1f3e305501049099b2693b","unresolved":false,"context_lines":[{"line_number":18,"context_line":"Switching to cast() seems to fix the issue and seems more reasonable."},{"line_number":19,"context_line":""},{"line_number":20,"context_line":"Change-Id: I4dbb04ccb93199bba4e1a1614bc19b70a068a9ea"},{"line_number":21,"context_line":"Closes-Bug: 1442810"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"7aaa499b_481457f4","line":21,"updated":"2015-04-13 09:09:56.000000000","message":"The only please I see continue_node_cleaning() being called is here[1], and it\u0027s calling task.release_sources() right before invoking continue_node_cleaning(). So for some reason the node remains locked or we have a race to lock the node.\n\nI\u0027m not against having cast(), I think it also makes sense since we don\u0027t expect any return values from continue_node_cleaning() but I don\u0027t think this solves the problem. It workaround it. Now the call will not wait for the return of the function, but the lock problem remains.\n\n[1] https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/agent_base_vendor.py#L136-L137","commit_id":"63df1d3fe019c9274784160b2558b76bd7d7930c"},{"author":{"_account_id":2889,"name":"Aeva Black","email":"aeva.online@gmail.com","username":"tenbrae"},"change_message_id":"b5435cc04afcc2df84c86f884c6f2616851582fa","unresolved":false,"context_lines":[{"line_number":18,"context_line":"Switching to cast() seems to fix the issue and seems more reasonable."},{"line_number":19,"context_line":""},{"line_number":20,"context_line":"Change-Id: I4dbb04ccb93199bba4e1a1614bc19b70a068a9ea"},{"line_number":21,"context_line":"Closes-Bug: 1442810"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":1,"id":"7aaa499b_e42b3cea","line":21,"in_reply_to":"7aaa499b_7978f9c4","updated":"2015-04-13 16:55:58.000000000","message":"sorry - I meant I agree that cast makes sense.\n\nThe bug is \"cleaning may leave nodes locked and require manual intervention to unlock\". If this change stops that behavior, I think \u0027closes-bug\u0027 is fine. \n\nFWIW, this code path appears needlessly complicated and would be really good candidate to clean up in Liberty (no pun intended).","commit_id":"63df1d3fe019c9274784160b2558b76bd7d7930c"}],"ironic/conductor/rpcapi.py":[{"author":{"_account_id":2889,"name":"Aeva Black","email":"aeva.online@gmail.com","username":"tenbrae"},"change_message_id":"4c52cb39f7f4ee7bb54d22cb1d53630479e467e7","unresolved":false,"context_lines":[{"line_number":68,"context_line":"    |    1.23 - Added do_provisioning_action"},{"line_number":69,"context_line":"    |    1.24 - Added inspect_hardware method"},{"line_number":70,"context_line":"    |    1.25 - Added destroy_port"},{"line_number":71,"context_line":"    |    1.26 - Added continue_node_clean"},{"line_number":72,"context_line":""},{"line_number":73,"context_line":"    \"\"\""},{"line_number":74,"context_line":""}],"source_content_type":"text/x-python","patch_set":1,"id":"7aaa499b_59753dc9","line":71,"updated":"2015-04-13 16:40:32.000000000","message":"new line here:\n  1.27 convert continue_node_clean to cast","commit_id":"63df1d3fe019c9274784160b2558b76bd7d7930c"},{"author":{"_account_id":2889,"name":"Aeva Black","email":"aeva.online@gmail.com","username":"tenbrae"},"change_message_id":"4c52cb39f7f4ee7bb54d22cb1d53630479e467e7","unresolved":false,"context_lines":[{"line_number":73,"context_line":"    \"\"\""},{"line_number":74,"context_line":""},{"line_number":75,"context_line":"    # NOTE(rloo): This must be in sync with manager.ConductorManager\u0027s."},{"line_number":76,"context_line":"    RPC_API_VERSION \u003d \u00271.26\u0027"},{"line_number":77,"context_line":""},{"line_number":78,"context_line":"    def __init__(self, topic\u003dNone):"},{"line_number":79,"context_line":"        super(ConductorAPI, self).__init__()"}],"source_content_type":"text/x-python","patch_set":1,"id":"7aaa499b_39827195","line":76,"updated":"2015-04-13 16:40:32.000000000","message":"this needs to be bumped to 1.27","commit_id":"63df1d3fe019c9274784160b2558b76bd7d7930c"}]}
