)]}'
{"specs/keystone/newton/service-user-permissions.rst":[{"author":{"_account_id":5046,"name":"Lance Bragstad","email":"lbragstad@redhat.com","username":"ldbragst"},"change_message_id":"da4b6f323ce48b93a08b33a3355f3dd7d3def281","unresolved":false,"context_lines":[{"line_number":34,"context_line":"Because keystone tokens are bearer tokens the easiest way to accomplish this"},{"line_number":35,"context_line":"continuous user authentication was to have the service reuse the same token"},{"line_number":36,"context_line":"when communicating with others. This works most of the time, however there is"},{"line_number":37,"context_line":"an edge case that if the token expiry period is crossed between when the user"},{"line_number":38,"context_line":"submits a token and the service passes the token on then token authentication"},{"line_number":39,"context_line":"will fail. At scale this edge case becomes a serious problem and is a negative"},{"line_number":40,"context_line":"experience for users because the operation may have been running a significant"}],"source_content_type":"text/x-rst","patch_set":1,"id":"9abb7d3a_47988dc0","line":37,"updated":"2016-05-31 21:05:14.000000000","message":"nit: s/that/where/ ?","commit_id":"448ddfd94e86ea7eec988c69f0a1f615a6674d22"},{"author":{"_account_id":5046,"name":"Lance Bragstad","email":"lbragstad@redhat.com","username":"ldbragst"},"change_message_id":"da4b6f323ce48b93a08b33a3355f3dd7d3def281","unresolved":false,"context_lines":[{"line_number":51,"context_line":""},{"line_number":52,"context_line":"Auth_token middleware currently passes token data information down to the"},{"line_number":53,"context_line":"service it protects by inserting additional headers into the request. To ensure"},{"line_number":54,"context_line":"that there is no abuse currently it makes sure to unset any of the headers it"},{"line_number":55,"context_line":"may use before beginning validation so that extra data can never make it to the"},{"line_number":56,"context_line":"user."},{"line_number":57,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"9abb7d3a_67e2f130","line":54,"updated":"2016-05-31 21:05:14.000000000","message":"nit: \n\nTo ensure that there is no abuse, auth_token middleware unsets any of the headers it may have used before beginning validation so that extra data never makes it to the user.\n\nI\u0027m not sure if i read that sentence right","commit_id":"448ddfd94e86ea7eec988c69f0a1f615a6674d22"},{"author":{"_account_id":21013,"name":"Georgy Kibardin","email":"gkibardin@mirantis.com","username":"gkibardin"},"change_message_id":"75eec61d1df9245ca63fa3f356dc55f93f99d944","unresolved":false,"context_lines":[{"line_number":60,"context_line":"issued the request. When a request is received with an X-Service-Token header"},{"line_number":61,"context_line":"this token is also validate and it\u0027s information passed to the service which"},{"line_number":62,"context_line":"allows the service to enforce joint ownership and other permissions on the"},{"line_number":63,"context_line":"operation. For example in swift, an object was created for a user but created"},{"line_number":64,"context_line":"by the nova service. Policy should be able to enforce that the object can only"},{"line_number":65,"context_line":"be deleted via combination of the nova service user and the user token."},{"line_number":66,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"9abb7d3a_93d8a154","line":63,"updated":"2016-06-02 07:18:14.000000000","message":"I think that some kind of MAC would be more secure than just a token. Stolen token would allow an attacker to set any user id he wants, while MAC would only allow replay attacks (which may be mitigated using nonces or other techniques).\nThis, however, implies establishing trust relationship between services in advance. Which brings an idea of using existing means such as SSL certificates etc.","commit_id":"448ddfd94e86ea7eec988c69f0a1f615a6674d22"},{"author":{"_account_id":1916,"name":"Guang Yee","email":"gyee@suse.com","username":"guang-yee"},"change_message_id":"99e4b8532d095ff6b5fd1c0e647f67ba3e60f4bc","unresolved":false,"context_lines":[{"line_number":65,"context_line":"be deleted via combination of the nova service user and the user token."},{"line_number":66,"context_line":""},{"line_number":67,"context_line":"We can utilize the X-Service-Token concept to ensure that a request originated"},{"line_number":68,"context_line":"from another service and not directly from a user."},{"line_number":69,"context_line":""},{"line_number":70,"context_line":"Alternatives"},{"line_number":71,"context_line":"------------"}],"source_content_type":"text/x-rst","patch_set":1,"id":"dab17558_f4900d45","line":68,"updated":"2016-05-17 16:24:32.000000000","message":"Are you proposing that if the X-Service-Token is successfully validated, regardless of its roles and scope, that we automatically trust the incoming X-* token headers? Can you please be more specific?","commit_id":"448ddfd94e86ea7eec988c69f0a1f615a6674d22"},{"author":{"_account_id":17860,"name":"Samuel de Medeiros Queiroz","email":"samueldmq@gmail.com","username":"samueldmq"},"change_message_id":"f88808239ad8e9c8d01c48b2dbcbc5702fcdbfa7","unresolved":false,"context_lines":[{"line_number":65,"context_line":"be deleted via combination of the nova service user and the user token."},{"line_number":66,"context_line":""},{"line_number":67,"context_line":"We can utilize the X-Service-Token concept to ensure that a request originated"},{"line_number":68,"context_line":"from another service and not directly from a user."},{"line_number":69,"context_line":""},{"line_number":70,"context_line":"Alternatives"},{"line_number":71,"context_line":"------------"}],"source_content_type":"text/x-rst","patch_set":1,"id":"3aaa91ec_36f08a90","line":68,"in_reply_to":"dab17558_46a04690","updated":"2016-06-22 11:42:56.000000000","message":"++ I like it","commit_id":"448ddfd94e86ea7eec988c69f0a1f615a6674d22"},{"author":{"_account_id":2903,"name":"Morgan Fainberg","email":"morgan.fainberg@gmail.com","username":"mdrnstm"},"change_message_id":"d112299307b4255bc234d639dd1806ed711a9837","unresolved":false,"context_lines":[{"line_number":65,"context_line":"be deleted via combination of the nova service user and the user token."},{"line_number":66,"context_line":""},{"line_number":67,"context_line":"We can utilize the X-Service-Token concept to ensure that a request originated"},{"line_number":68,"context_line":"from another service and not directly from a user."},{"line_number":69,"context_line":""},{"line_number":70,"context_line":"Alternatives"},{"line_number":71,"context_line":"------------"}],"source_content_type":"text/x-rst","patch_set":1,"id":"dab17558_46a04690","line":68,"in_reply_to":"dab17558_f4900d45","updated":"2016-05-17 18:45:21.000000000","message":"That is the general idea.","commit_id":"448ddfd94e86ea7eec988c69f0a1f615a6674d22"},{"author":{"_account_id":17860,"name":"Samuel de Medeiros Queiroz","email":"samueldmq@gmail.com","username":"samueldmq"},"change_message_id":"f88808239ad8e9c8d01c48b2dbcbc5702fcdbfa7","unresolved":false,"context_lines":[{"line_number":103,"context_line":"delegation story to the services from the start that doesn\u0027t involve bearer"},{"line_number":104,"context_line":"tokens."},{"line_number":105,"context_line":""},{"line_number":106,"context_line":"Don\u0027t check token expiry"},{"line_number":107,"context_line":"++++++++++++++++++++++++"},{"line_number":108,"context_line":""},{"line_number":109,"context_line":"A variation of this proposal is when passing the X-Auth-Token and"}],"source_content_type":"text/x-rst","patch_set":1,"id":"3aaa91ec_d63b06ea","line":106,"range":{"start_line":106,"start_character":0,"end_line":106,"end_character":24},"updated":"2016-06-22 11:42:56.000000000","message":"I was proposing this in the reservation spec (https://review.openstack.org/#/c/330329/)\n\nI understand the issues you list below, but what if tokens had an \u0027extended_expiry\u0027 attribute or something like that ? It\u0027s only added if asked by a service token AND is only valid when service token validates it in a service to service workflow.\n\nI think this is much simpler than implementing still another concept.","commit_id":"448ddfd94e86ea7eec988c69f0a1f615a6674d22"},{"author":{"_account_id":5046,"name":"Lance Bragstad","email":"lbragstad@redhat.com","username":"ldbragst"},"change_message_id":"da4b6f323ce48b93a08b33a3355f3dd7d3def281","unresolved":false,"context_lines":[{"line_number":126,"context_line":"user_id and project_id."},{"line_number":127,"context_line":""},{"line_number":128,"context_line":"This in my opinion is the most reasonable alternative as a service would be"},{"line_number":129,"context_line":"unable to pass invalid information to another service. It would requiring"},{"line_number":130,"context_line":"adding a new API to keystone that is capable of constructing this token data"},{"line_number":131,"context_line":"that is protected by policy to only be accessible to the service users. Ideally"},{"line_number":132,"context_line":"for performance we could roll this into one API that is capable of validating"}],"source_content_type":"text/x-rst","patch_set":1,"id":"9abb7d3a_3d2756cb","line":129,"updated":"2016-05-31 21:05:14.000000000","message":"nit: require","commit_id":"448ddfd94e86ea7eec988c69f0a1f615a6674d22"},{"author":{"_account_id":5046,"name":"Lance Bragstad","email":"lbragstad@redhat.com","username":"ldbragst"},"change_message_id":"da4b6f323ce48b93a08b33a3355f3dd7d3def281","unresolved":false,"context_lines":[{"line_number":148,"context_line":""},{"line_number":149,"context_line":"We would still need to provide an auth plugin for services that handles sending"},{"line_number":150,"context_line":"this core auth data. As proven by fernet tokens these core auth parameters that"},{"line_number":151,"context_line":"cannot be reconstructed keep growing."},{"line_number":152,"context_line":""},{"line_number":153,"context_line":"Security Impact"},{"line_number":154,"context_line":"---------------"}],"source_content_type":"text/x-rst","patch_set":1,"id":"9abb7d3a_18a38001","line":151,"updated":"2016-05-31 21:05:14.000000000","message":"Is this referencing the additional information required for federated tokens?","commit_id":"448ddfd94e86ea7eec988c69f0a1f615a6674d22"},{"author":{"_account_id":1916,"name":"Guang Yee","email":"gyee@suse.com","username":"guang-yee"},"change_message_id":"99e4b8532d095ff6b5fd1c0e647f67ba3e60f4bc","unresolved":false,"context_lines":[{"line_number":161,"context_line":"all of the user information that is passed. This means that someone possessing"},{"line_number":162,"context_line":"a service token is capable of issuing any request on behalf of any user"},{"line_number":163,"context_line":"regardless of this being a valid keystone relation. It would also be capable of"},{"line_number":164,"context_line":"for example adding roles to a user\u0027s information before passing it on."},{"line_number":165,"context_line":""},{"line_number":166,"context_line":"This escalates the privileges of a service to nearly that of an admin as it can"},{"line_number":167,"context_line":"emulate any request."}],"source_content_type":"text/x-rst","patch_set":1,"id":"dab17558_b46a057f","line":164,"updated":"2016-05-17 16:24:32.000000000","message":"Basically, compromising a service token is like compromising Keystone itself as service can impersonate any users at will right? But then again, if services are sharing the same memcached for token validation results, a compromised service can change the token data there anyway.","commit_id":"448ddfd94e86ea7eec988c69f0a1f615a6674d22"},{"author":{"_account_id":5046,"name":"Lance Bragstad","email":"lbragstad@redhat.com","username":"ldbragst"},"change_message_id":"da4b6f323ce48b93a08b33a3355f3dd7d3def281","unresolved":false,"context_lines":[{"line_number":164,"context_line":"for example adding roles to a user\u0027s information before passing it on."},{"line_number":165,"context_line":""},{"line_number":166,"context_line":"This escalates the privileges of a service to nearly that of an admin as it can"},{"line_number":167,"context_line":"emulate any request."},{"line_number":168,"context_line":""},{"line_number":169,"context_line":"Notifications Impact"},{"line_number":170,"context_line":"--------------------"}],"source_content_type":"text/x-rst","patch_set":1,"id":"9abb7d3a_584d083d","line":167,"updated":"2016-05-31 21:05:14.000000000","message":"From the nova-keystone session in Austin, it sounds like the nova folks took means to make sure the nova service didn\u0027t do any operations that required the admin role.","commit_id":"448ddfd94e86ea7eec988c69f0a1f615a6674d22"},{"author":{"_account_id":1916,"name":"Guang Yee","email":"gyee@suse.com","username":"guang-yee"},"change_message_id":"99e4b8532d095ff6b5fd1c0e647f67ba3e60f4bc","unresolved":false,"context_lines":[{"line_number":169,"context_line":"Notifications Impact"},{"line_number":170,"context_line":"--------------------"},{"line_number":171,"context_line":""},{"line_number":172,"context_line":"None."},{"line_number":173,"context_line":""},{"line_number":174,"context_line":"Other End User Impact"},{"line_number":175,"context_line":"---------------------"}],"source_content_type":"text/x-rst","patch_set":1,"id":"dab17558_74d37d04","line":172,"range":{"start_line":172,"start_character":4,"end_line":172,"end_character":5},"updated":"2016-05-17 16:24:32.000000000","message":"I would think this constitutes a different event, that server token, along with its headers are trusted.","commit_id":"448ddfd94e86ea7eec988c69f0a1f615a6674d22"},{"author":{"_account_id":5046,"name":"Lance Bragstad","email":"lbragstad@redhat.com","username":"ldbragst"},"change_message_id":"da4b6f323ce48b93a08b33a3355f3dd7d3def281","unresolved":false,"context_lines":[{"line_number":230,"context_line":""},{"line_number":231,"context_line":"* The X-Service-Token is validated by auth_token middleware but it is not"},{"line_number":232,"context_line":"  currently asserted that the service token actually came from another service."},{"line_number":233,"context_line":"  If a user pass the X-Service-Token header this would also be validated. If we"},{"line_number":234,"context_line":"  intend to bypass the user token validation process this is unacceptable and"},{"line_number":235,"context_line":"  we must verify that the token passed as the X-Service-Token is allowed to"},{"line_number":236,"context_line":"  invoke this bypass."}],"source_content_type":"text/x-rst","patch_set":1,"id":"9abb7d3a_180d40e8","line":233,"updated":"2016-05-31 21:05:14.000000000","message":"passes*","commit_id":"448ddfd94e86ea7eec988c69f0a1f615a6674d22"}]}
