)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":10213,"name":"Alon Marx","email":"alonma@il.ibm.com","username":"alonmarx"},"change_message_id":"33cfd0bd815e271fb82581b5faeb5c8515582678","unresolved":false,"context_lines":[{"line_number":6,"context_line":""},{"line_number":7,"context_line":"Migrate volume between backends in an async way"},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"Currently if we migrate a available volume from backend A to"},{"line_number":10,"context_line":"backend B who doesn\u0027t refer to the same array as backend A,"},{"line_number":11,"context_line":"we can\u0027t attach this volume to a server for real use untill"},{"line_number":12,"context_line":"the whole migration is totally completed."}],"source_content_type":"text/x-gerrit-commit-message","patch_set":21,"id":"9abb7d3a_72cd4d65","line":9,"updated":"2016-06-02 05:36:30.000000000","message":"a \u003d\u003e an","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"c064f99cb677f0843fa1363b5b9d7d724cece4da","unresolved":false,"context_lines":[{"line_number":6,"context_line":""},{"line_number":7,"context_line":"Migrate volume between backends in an async way"},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"Currently if we migrate a available volume from backend A to"},{"line_number":10,"context_line":"backend B who doesn\u0027t refer to the same array as backend A,"},{"line_number":11,"context_line":"we can\u0027t attach this volume to a server for real use untill"},{"line_number":12,"context_line":"the whole migration is totally completed."}],"source_content_type":"text/x-gerrit-commit-message","patch_set":21,"id":"9abb7d3a_4459428e","line":9,"in_reply_to":"9abb7d3a_72cd4d65","updated":"2016-06-06 07:46:12.000000000","message":"Done, thanks!","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":10213,"name":"Alon Marx","email":"alonma@il.ibm.com","username":"alonmarx"},"change_message_id":"33cfd0bd815e271fb82581b5faeb5c8515582678","unresolved":false,"context_lines":[{"line_number":8,"context_line":""},{"line_number":9,"context_line":"Currently if we migrate a available volume from backend A to"},{"line_number":10,"context_line":"backend B who doesn\u0027t refer to the same array as backend A,"},{"line_number":11,"context_line":"we can\u0027t attach this volume to a server for real use untill"},{"line_number":12,"context_line":"the whole migration is totally completed."},{"line_number":13,"context_line":"This spec is trying to find a way to migrate volume"},{"line_number":14,"context_line":"asynchronously to make the volume usable immediately after"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":21,"id":"9abb7d3a_32e625e8","line":11,"range":{"start_line":11,"start_character":53,"end_line":11,"end_character":59},"updated":"2016-06-02 05:36:30.000000000","message":"until","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"c064f99cb677f0843fa1363b5b9d7d724cece4da","unresolved":false,"context_lines":[{"line_number":8,"context_line":""},{"line_number":9,"context_line":"Currently if we migrate a available volume from backend A to"},{"line_number":10,"context_line":"backend B who doesn\u0027t refer to the same array as backend A,"},{"line_number":11,"context_line":"we can\u0027t attach this volume to a server for real use untill"},{"line_number":12,"context_line":"the whole migration is totally completed."},{"line_number":13,"context_line":"This spec is trying to find a way to migrate volume"},{"line_number":14,"context_line":"asynchronously to make the volume usable immediately after"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":21,"id":"9abb7d3a_245ebe76","line":11,"range":{"start_line":11,"start_character":53,"end_line":11,"end_character":59},"in_reply_to":"9abb7d3a_32e625e8","updated":"2016-06-06 07:46:12.000000000","message":"Done","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"}],"specs/newton/async-volume-migration.rst":[{"author":{"_account_id":6491,"name":"xing-yang","email":"xingyang105@gmail.com","username":"xing-yang"},"change_message_id":"d3457ac62f5d48d635177362bd7e695a4a8e03b2","unresolved":false,"context_lines":[{"line_number":61,"context_line":"remote(source) LUN and no data will be lost."},{"line_number":62,"context_line":""},{"line_number":63,"context_line":"To enable one array to take over other array\u0027s volume, we should allow one"},{"line_number":64,"context_line":"driver to call other drivers\u0027 interfaces directly or indirectly."},{"line_number":65,"context_line":""},{"line_number":66,"context_line":"To enable an available volume can be used(attach to a server) immediately"},{"line_number":67,"context_line":"after we issued migration from one backend to another backend, we should allow"}],"source_content_type":"text/x-rst","patch_set":9,"id":"1a122d0e_bb308b9f","line":64,"updated":"2016-05-05 13:46:18.000000000","message":"Are the local array and remote array managed by two different Cinder volume nodes or by the same Cinder volume node? Please specify.","commit_id":"b5fd9fd194c71200ef090849b36747087c2f5693"},{"author":{"_account_id":1207,"name":"Duncan Thomas","email":"duncan.thomas@gmail.com","username":"duncan-thomas"},"change_message_id":"e2ad547f3e9af71360f44fc4e437024df84ec232","unresolved":false,"context_lines":[{"line_number":51,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":52,"context_line":""},{"line_number":53,"context_line":"In brief, the async migration is to use some features of backend array and"},{"line_number":54,"context_line":"I\u0027ve known that these features are already supported by most vendors."},{"line_number":55,"context_line":""},{"line_number":56,"context_line":"These features are:"},{"line_number":57,"context_line":"1. One array can take over other array\u0027s LUN as a remote LUN if these two"}],"source_content_type":"text/x-rst","patch_set":10,"id":"dab17558_5221d2b3","line":54,"range":{"start_line":54,"start_character":0,"end_line":54,"end_character":69},"updated":"2016-05-07 22:51:13.000000000","message":"Can we get some idea of how many vendors can support this? The only one I know of is Huawei - it doesn\u0027t mean there aren\u0027t more, but I\u0027d like to know if there are more than one or two.","commit_id":"680cd5c345d4732e75a1be7030828d7df79931bb"},{"author":{"_account_id":7198,"name":"Jay Bryant","email":"jungleboyj@electronicjungle.net","username":"jsbryant"},"change_message_id":"1db2698f5d75f56cf6e640cba98c470907e4d4a2","unresolved":false,"context_lines":[{"line_number":73,"context_line":"To enable an available volume can be used(attach to a server) immediately"},{"line_number":74,"context_line":"after we issued migration from one backend to another backend, we should allow"},{"line_number":75,"context_line":"the migration task to be send to the target backend, not the source backend."},{"line_number":76,"context_line":"Driver should report the \u0027async_volume_migration\u0027 capability as Ture if it"},{"line_number":77,"context_line":"supports. If driver does not report this capability or reports it as False, we"},{"line_number":78,"context_line":"will take as it does not support. When scheduler received a migrate_volume"},{"line_number":79,"context_line":"task, it will firstly check whether the target backend support"}],"source_content_type":"text/x-rst","patch_set":11,"id":"dab17558_6a010438","line":76,"range":{"start_line":76,"start_character":64,"end_line":76,"end_character":68},"updated":"2016-05-10 23:49:01.000000000","message":"True","commit_id":"8ea0a467a78c53e0b9b36ee9072366c1369d30c9"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"04147948f7a9acf2f705a82f3422e702e01dee3c","unresolved":false,"context_lines":[{"line_number":73,"context_line":"To enable an available volume can be used(attach to a server) immediately"},{"line_number":74,"context_line":"after we issued migration from one backend to another backend, we should allow"},{"line_number":75,"context_line":"the migration task to be send to the target backend, not the source backend."},{"line_number":76,"context_line":"Driver should report the \u0027async_volume_migration\u0027 capability as Ture if it"},{"line_number":77,"context_line":"supports. If driver does not report this capability or reports it as False, we"},{"line_number":78,"context_line":"will take as it does not support. When scheduler received a migrate_volume"},{"line_number":79,"context_line":"task, it will firstly check whether the target backend support"}],"source_content_type":"text/x-rst","patch_set":11,"id":"dab17558_79b28095","line":76,"range":{"start_line":76,"start_character":64,"end_line":76,"end_character":68},"in_reply_to":"dab17558_6a010438","updated":"2016-05-11 03:55:25.000000000","message":"Done","commit_id":"8ea0a467a78c53e0b9b36ee9072366c1369d30c9"},{"author":{"_account_id":7198,"name":"Jay Bryant","email":"jungleboyj@electronicjungle.net","username":"jsbryant"},"change_message_id":"1db2698f5d75f56cf6e640cba98c470907e4d4a2","unresolved":false,"context_lines":[{"line_number":87,"context_line":"this. If a driver report \u0027async_volume_migration\u0027 as True, it should implement"},{"line_number":88,"context_line":"the \u0027migrate_volume_async\u0027 interface. We introduce a new interface"},{"line_number":89,"context_line":"\u0027migrate_volume_async\u0027 instead of use the existing \u0027migrate_volume\u0027 interface"},{"line_number":90,"context_line":"because there would be lots of different between them, a significant different"},{"line_number":91,"context_line":"is that the existing \u0027migrate_volume\u0027 interface always take them self as the"},{"line_number":92,"context_line":"source backend, but the new interface should take itself as the target"},{"line_number":93,"context_line":"backend."}],"source_content_type":"text/x-rst","patch_set":11,"id":"dab17558_0abe7864","line":90,"range":{"start_line":90,"start_character":69,"end_line":90,"end_character":78},"updated":"2016-05-10 23:49:01.000000000","message":"difference","commit_id":"8ea0a467a78c53e0b9b36ee9072366c1369d30c9"},{"author":{"_account_id":7198,"name":"Jay Bryant","email":"jungleboyj@electronicjungle.net","username":"jsbryant"},"change_message_id":"1db2698f5d75f56cf6e640cba98c470907e4d4a2","unresolved":false,"context_lines":[{"line_number":87,"context_line":"this. If a driver report \u0027async_volume_migration\u0027 as True, it should implement"},{"line_number":88,"context_line":"the \u0027migrate_volume_async\u0027 interface. We introduce a new interface"},{"line_number":89,"context_line":"\u0027migrate_volume_async\u0027 instead of use the existing \u0027migrate_volume\u0027 interface"},{"line_number":90,"context_line":"because there would be lots of different between them, a significant different"},{"line_number":91,"context_line":"is that the existing \u0027migrate_volume\u0027 interface always take them self as the"},{"line_number":92,"context_line":"source backend, but the new interface should take itself as the target"},{"line_number":93,"context_line":"backend."}],"source_content_type":"text/x-rst","patch_set":11,"id":"dab17558_aa8e4ccd","line":90,"range":{"start_line":90,"start_character":31,"end_line":90,"end_character":40},"updated":"2016-05-10 23:49:01.000000000","message":"differences","commit_id":"8ea0a467a78c53e0b9b36ee9072366c1369d30c9"},{"author":{"_account_id":7198,"name":"Jay Bryant","email":"jungleboyj@electronicjungle.net","username":"jsbryant"},"change_message_id":"1db2698f5d75f56cf6e640cba98c470907e4d4a2","unresolved":false,"context_lines":[{"line_number":102,"context_line":"2. Construct the connector with initiator or wwpns as if the target backend is"},{"line_number":103,"context_line":"a host, and then call source driver\u0027s initialize_connection() through"},{"line_number":104,"context_line":"volume.rpcapi.VolumeAPI()."},{"line_number":105,"context_line":"3. Scan the source volume with is attached to the target backend."},{"line_number":106,"context_line":"4. If we successfully found the source volume on the target backend after"},{"line_number":107,"context_line":"scan, create a migration task on the target backend and start the migration"},{"line_number":108,"context_line":"task."}],"source_content_type":"text/x-rst","patch_set":11,"id":"dab17558_0a65d81c","line":105,"range":{"start_line":105,"start_character":26,"end_line":105,"end_character":30},"updated":"2016-05-10 23:49:01.000000000","message":"which","commit_id":"8ea0a467a78c53e0b9b36ee9072366c1369d30c9"},{"author":{"_account_id":7198,"name":"Jay Bryant","email":"jungleboyj@electronicjungle.net","username":"jsbryant"},"change_message_id":"1db2698f5d75f56cf6e640cba98c470907e4d4a2","unresolved":false,"context_lines":[{"line_number":110,"context_line":"refers to the target backend\u0027s remote LUN, whose real location is the source"},{"line_number":111,"context_line":"backend. This will make sure all the request to the source volume will be send"},{"line_number":112,"context_line":"to the target backend\u0027s remote LUN."},{"line_number":113,"context_line":"6. Return the modle_update after start the migration, no need to wait till the"},{"line_number":114,"context_line":"migration complete."},{"line_number":115,"context_line":""},{"line_number":116,"context_line":"After driver\u0027s migration returned to volume manager, volume manager update the"}],"source_content_type":"text/x-rst","patch_set":11,"id":"dab17558_6a7624d8","line":113,"range":{"start_line":113,"start_character":14,"end_line":113,"end_character":26},"updated":"2016-05-10 23:49:01.000000000","message":"model_update","commit_id":"8ea0a467a78c53e0b9b36ee9072366c1369d30c9"},{"author":{"_account_id":7198,"name":"Jay Bryant","email":"jungleboyj@electronicjungle.net","username":"jsbryant"},"change_message_id":"1db2698f5d75f56cf6e640cba98c470907e4d4a2","unresolved":false,"context_lines":[{"line_number":151,"context_line":"Other end user impact"},{"line_number":152,"context_line":"---------------------"},{"line_number":153,"context_line":""},{"line_number":154,"context_line":"Users should make sure the backends is connected to each other if they want to"},{"line_number":155,"context_line":"perform the async volume migration."},{"line_number":156,"context_line":""},{"line_number":157,"context_line":"Performance Impact"}],"source_content_type":"text/x-rst","patch_set":11,"id":"dab17558_2a47fc77","line":154,"range":{"start_line":154,"start_character":36,"end_line":154,"end_character":38},"updated":"2016-05-10 23:49:01.000000000","message":"are","commit_id":"8ea0a467a78c53e0b9b36ee9072366c1369d30c9"},{"author":{"_account_id":7198,"name":"Jay Bryant","email":"jungleboyj@electronicjungle.net","username":"jsbryant"},"change_message_id":"1db2698f5d75f56cf6e640cba98c470907e4d4a2","unresolved":false,"context_lines":[{"line_number":168,"context_line":"Developer impact"},{"line_number":169,"context_line":"----------------"},{"line_number":170,"context_line":""},{"line_number":171,"context_line":"None"},{"line_number":172,"context_line":""},{"line_number":173,"context_line":"Implementation"},{"line_number":174,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":11,"id":"dab17558_0a331815","line":171,"range":{"start_line":171,"start_character":0,"end_line":171,"end_character":4},"updated":"2016-05-10 23:49:01.000000000","message":"Is there really nothing that needs to be done by the driver developers to enable this?","commit_id":"8ea0a467a78c53e0b9b36ee9072366c1369d30c9"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"419488aced92a1eb1d29cf0aed6a91afaf109808","unresolved":false,"context_lines":[{"line_number":168,"context_line":"Developer impact"},{"line_number":169,"context_line":"----------------"},{"line_number":170,"context_line":""},{"line_number":171,"context_line":"None"},{"line_number":172,"context_line":""},{"line_number":173,"context_line":"Implementation"},{"line_number":174,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":11,"id":"dab17558_d9872bb3","line":171,"range":{"start_line":171,"start_character":0,"end_line":171,"end_character":4},"in_reply_to":"dab17558_0a331815","updated":"2016-05-11 04:02:02.000000000","message":"Driver developers should implement something if they want to support this,otherwise they needn\u0027t do anything. Have added this description in next patch, thanks!","commit_id":"8ea0a467a78c53e0b9b36ee9072366c1369d30c9"},{"author":{"_account_id":15506,"name":"zhangli","email":"zhangli09@huawei.com"},"change_message_id":"620693762f13bf213b05bc0ff0411dd26d1ec264","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":12,"id":"dab17558_a25b0c85","updated":"2016-05-17 06:33:36.000000000","message":"As I understand it, this spec includes 2 individual functions, but the description seems to be a little mixed up. \nFunction 1:\nUse the capability of backends instead of host assist (dd or file read/write) to perform volume migration between 2 different backends. \n\nThe mechanism of this function is relatively simple: \n1)\tAttach the source volume to the target backend device. Generally the target backend performs the role of a SCSI initiator. \n2)\tPerform volume migration inside the target bankend device.\n\nThe current volume manager doesn’t forbid drivers to do such kind of volume attaching stuff, so there is no need of change about this, but I think there could be one enhancement:\nSometimes we can only attach the newly created volume of the target backend to the source backend device, and then perform volume migration inside the source backend device. This depends on which backend has the capability of attaching an external volume to itself and performing volume migration; \nThis need a little change to volume manager:\nAs the current implementation, volume manager will invoke the migrate_volume routine of the source backend driver, if it fails, manager will invoke the _migrate_volume_generic routine. We can add a step that let the target backend driver to have a try (eg. migrate_volume_target) after the source backend driver return fail, this will help to increase the success probability of the backend based migration. \n\n\nFunction 2:\nMake the migration an online migration (or live migration called by some vendor), which means during the data coping period, the volume can still be used. Online migration is a very necessary feature in a real production environment, because customers usually can’t afford to shut down the business for a long time, especially when the volume size is big or the migration bandwidth is low.\n\nBut I think maybe it’s not necessary to add a new driver interface migrate_volume_async, we can use the current migrate_volume (and migrate_volume_target mentioned above) interface, just allow the drivers which support this function to return a new return value ‘migration_aync’ to tell the manager if an aync(online) migration is triggered.","commit_id":"8669bb83fdd263315a47beb02a15f7d7d2922cad"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"df934dd10adcbc3e7dfc5c6d9c891fc84050aca0","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":12,"id":"dab17558_c50d7590","in_reply_to":"dab17558_a25b0c85","updated":"2016-05-17 08:04:14.000000000","message":"I will update this spec to describe it more clearly, thanks!","commit_id":"8669bb83fdd263315a47beb02a15f7d7d2922cad"},{"author":{"_account_id":1736,"name":"Ivan Kolodyazhny","email":"e0ne@e0ne.info","username":"e0ne"},"change_message_id":"9ab1323173ae52d59b06e3c7ee378513aa82eeee","unresolved":false,"context_lines":[{"line_number":14,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":15,"context_line":""},{"line_number":16,"context_line":"If we want to migrate a available volume from one backend to another backend,"},{"line_number":17,"context_line":"the current way we can choose is host-copy, using \u0027dd\u0027 command which will take"},{"line_number":18,"context_line":"a mess of resources on the Cinder node, and with no guarantee for data"},{"line_number":19,"context_line":"integrity. Also, the \u0027dd\u0027 way make the volume not usable until the host-copy"},{"line_number":20,"context_line":"is totally completed, this will be a very long time even the volume is in a"}],"source_content_type":"text/x-rst","patch_set":12,"id":"dab17558_389aa86e","line":17,"range":{"start_line":17,"start_character":33,"end_line":17,"end_character":42},"updated":"2016-05-11 09:23:46.000000000","message":"We can use backend-specific migrations too if drivers implemented it. Please, check https://github.com/openstack/cinder-specs/blob/master/specs/liberty/generic-volume-migration.rst for more details","commit_id":"8669bb83fdd263315a47beb02a15f7d7d2922cad"},{"author":{"_account_id":15506,"name":"zhangli","email":"zhangli09@huawei.com"},"change_message_id":"620693762f13bf213b05bc0ff0411dd26d1ec264","unresolved":false,"context_lines":[{"line_number":14,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":15,"context_line":""},{"line_number":16,"context_line":"If we want to migrate a available volume from one backend to another backend,"},{"line_number":17,"context_line":"the current way we can choose is host-copy, using \u0027dd\u0027 command which will take"},{"line_number":18,"context_line":"a mess of resources on the Cinder node, and with no guarantee for data"},{"line_number":19,"context_line":"integrity. Also, the \u0027dd\u0027 way make the volume not usable until the host-copy"},{"line_number":20,"context_line":"is totally completed, this will be a very long time even the volume is in a"}],"source_content_type":"text/x-rst","patch_set":12,"id":"dab17558_022938e6","line":17,"range":{"start_line":17,"start_character":33,"end_line":17,"end_character":42},"in_reply_to":"dab17558_389aa86e","updated":"2016-05-17 06:33:36.000000000","message":"Hi Ivan,\nI checked the spec you mentioned, but the actually changed code is seems for _migrate_volume_generic routine? Could you give us an explaination how this affects driver\u0027s migrate_volume routine?  Thanks.","commit_id":"8669bb83fdd263315a47beb02a15f7d7d2922cad"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"df934dd10adcbc3e7dfc5c6d9c891fc84050aca0","unresolved":false,"context_lines":[{"line_number":14,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":15,"context_line":""},{"line_number":16,"context_line":"If we want to migrate a available volume from one backend to another backend,"},{"line_number":17,"context_line":"the current way we can choose is host-copy, using \u0027dd\u0027 command which will take"},{"line_number":18,"context_line":"a mess of resources on the Cinder node, and with no guarantee for data"},{"line_number":19,"context_line":"integrity. Also, the \u0027dd\u0027 way make the volume not usable until the host-copy"},{"line_number":20,"context_line":"is totally completed, this will be a very long time even the volume is in a"}],"source_content_type":"text/x-rst","patch_set":12,"id":"dab17558_384f4835","line":17,"range":{"start_line":17,"start_character":33,"end_line":17,"end_character":42},"in_reply_to":"dab17558_389aa86e","updated":"2016-05-17 08:04:14.000000000","message":"Maybe I did not described this clearly, please see my next patch, thanks Ivan!","commit_id":"8669bb83fdd263315a47beb02a15f7d7d2922cad"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"df934dd10adcbc3e7dfc5c6d9c891fc84050aca0","unresolved":false,"context_lines":[{"line_number":14,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":15,"context_line":""},{"line_number":16,"context_line":"If we want to migrate a available volume from one backend to another backend,"},{"line_number":17,"context_line":"the current way we can choose is host-copy, using \u0027dd\u0027 command which will take"},{"line_number":18,"context_line":"a mess of resources on the Cinder node, and with no guarantee for data"},{"line_number":19,"context_line":"integrity. Also, the \u0027dd\u0027 way make the volume not usable until the host-copy"},{"line_number":20,"context_line":"is totally completed, this will be a very long time even the volume is in a"}],"source_content_type":"text/x-rst","patch_set":12,"id":"dab17558_9860f4a6","line":17,"range":{"start_line":17,"start_character":33,"end_line":17,"end_character":42},"in_reply_to":"dab17558_389aa86e","updated":"2016-05-17 08:04:14.000000000","message":"Yes, we can use backend-specific migrations if drivers implemented it, but currently none driver has implemented it,","commit_id":"8669bb83fdd263315a47beb02a15f7d7d2922cad"},{"author":{"_account_id":1736,"name":"Ivan Kolodyazhny","email":"e0ne@e0ne.info","username":"e0ne"},"change_message_id":"9ab1323173ae52d59b06e3c7ee378513aa82eeee","unresolved":false,"context_lines":[{"line_number":40,"context_line":"migration case 1:"},{"line_number":41,"context_line":"Within the scope of this spec:"},{"line_number":42,"context_line":"  1.Available volume migration between two different backends. These backends"},{"line_number":43,"context_line":"    could from one single vendor or different vendors, this needs driver\u0027s"},{"line_number":44,"context_line":"    support."},{"line_number":45,"context_line":""},{"line_number":46,"context_line":"Out of the scope of the spec:"}],"source_content_type":"text/x-rst","patch_set":12,"id":"dab17558_38a8e876","line":43,"updated":"2016-05-11 09:23:46.000000000","message":"We have such option now, haven\u0027t we?","commit_id":"8669bb83fdd263315a47beb02a15f7d7d2922cad"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"df934dd10adcbc3e7dfc5c6d9c891fc84050aca0","unresolved":false,"context_lines":[{"line_number":40,"context_line":"migration case 1:"},{"line_number":41,"context_line":"Within the scope of this spec:"},{"line_number":42,"context_line":"  1.Available volume migration between two different backends. These backends"},{"line_number":43,"context_line":"    could from one single vendor or different vendors, this needs driver\u0027s"},{"line_number":44,"context_line":"    support."},{"line_number":45,"context_line":""},{"line_number":46,"context_line":"Out of the scope of the spec:"}],"source_content_type":"text/x-rst","patch_set":12,"id":"dab17558_65dac918","line":43,"in_reply_to":"dab17558_38a8e876","updated":"2016-05-17 08:04:14.000000000","message":"Yes we have such option now, but currently it doesn\u0027t support async migration, which this spec is talking about. I will update this for describe this more detailed in next patch.","commit_id":"8669bb83fdd263315a47beb02a15f7d7d2922cad"},{"author":{"_account_id":15961,"name":"lisali","email":"xiaoyan.li@intel.com","username":"lisali"},"change_message_id":"54d442b8e0f0841b011ee0d01d9e1f290501477f","unresolved":false,"context_lines":[{"line_number":74,"context_line":"There will be a new interface for driver which will be called"},{"line_number":75,"context_line":"\u0027migrate_volume_target\u0027 and we will put a \u0027pass\u0027 function in volume/driver.py"},{"line_number":76,"context_line":"so other drivers won\u0027t support this feature needn\u0027t modify anything."},{"line_number":77,"context_line":"We introduce a new interface \u0027migrate_volume_target\u0027 instead of use the existing"},{"line_number":78,"context_line":"\u0027migrate_volume\u0027 interface because there would be lots of differences between"},{"line_number":79,"context_line":"them, a significant difference is that the existing \u0027migrate_volume\u0027 interface"},{"line_number":80,"context_line":"always take them self as the source backend, but the new interface should take"}],"source_content_type":"text/x-rst","patch_set":15,"id":"bab6814e_0806f4cd","line":77,"updated":"2016-05-19 07:48:21.000000000","message":"To put a \u0027pass\u0027 in migrate_volume_target in volume/driver is not correct. You should return moved as False. This means by default this function doesn\u0027t migrate volume.","commit_id":"6551fd72bc20554af400b17a0a7434d16d95e827"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"a54b4ea097557bb0b4852956a9246b6279c9a5d4","unresolved":false,"context_lines":[{"line_number":74,"context_line":"There will be a new interface for driver which will be called"},{"line_number":75,"context_line":"\u0027migrate_volume_target\u0027 and we will put a \u0027pass\u0027 function in volume/driver.py"},{"line_number":76,"context_line":"so other drivers won\u0027t support this feature needn\u0027t modify anything."},{"line_number":77,"context_line":"We introduce a new interface \u0027migrate_volume_target\u0027 instead of use the existing"},{"line_number":78,"context_line":"\u0027migrate_volume\u0027 interface because there would be lots of differences between"},{"line_number":79,"context_line":"them, a significant difference is that the existing \u0027migrate_volume\u0027 interface"},{"line_number":80,"context_line":"always take them self as the source backend, but the new interface should take"}],"source_content_type":"text/x-rst","patch_set":15,"id":"bab6814e_186f492c","line":77,"in_reply_to":"bab6814e_0806f4cd","updated":"2016-05-19 14:15:35.000000000","message":"Yes, thanks for checking this!","commit_id":"6551fd72bc20554af400b17a0a7434d16d95e827"},{"author":{"_account_id":15961,"name":"lisali","email":"xiaoyan.li@intel.com","username":"lisali"},"change_message_id":"54d442b8e0f0841b011ee0d01d9e1f290501477f","unresolved":false,"context_lines":[{"line_number":100,"context_line":"  perform host-copy migration. If migrate_volume_target() returns moved as"},{"line_number":101,"context_line":"  True and async as False, change the \"migration_status\" to \"success\" and"},{"line_number":102,"context_line":"  update the volume in db. If migrate_volume_target() returns moved as True"},{"line_number":103,"context_line":"  and async as True, we will go to step 3."},{"line_number":104,"context_line":""},{"line_number":105,"context_line":"3.Update the volume.host to the target host and update the db with model_update."},{"line_number":106,"context_line":"  Now the volume is usable for server to perform read/write on it."}],"source_content_type":"text/x-rst","patch_set":15,"id":"bab6814e_681ce013","line":103,"updated":"2016-05-19 07:48:21.000000000","message":"I think no matter async is true or false, step 3 should be executed.\n\nOnly step 4 is executed when moved as true and async as true.","commit_id":"6551fd72bc20554af400b17a0a7434d16d95e827"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"a54b4ea097557bb0b4852956a9246b6279c9a5d4","unresolved":false,"context_lines":[{"line_number":100,"context_line":"  perform host-copy migration. If migrate_volume_target() returns moved as"},{"line_number":101,"context_line":"  True and async as False, change the \"migration_status\" to \"success\" and"},{"line_number":102,"context_line":"  update the volume in db. If migrate_volume_target() returns moved as True"},{"line_number":103,"context_line":"  and async as True, we will go to step 3."},{"line_number":104,"context_line":""},{"line_number":105,"context_line":"3.Update the volume.host to the target host and update the db with model_update."},{"line_number":106,"context_line":"  Now the volume is usable for server to perform read/write on it."}],"source_content_type":"text/x-rst","patch_set":15,"id":"bab6814e_f892459e","line":103,"in_reply_to":"bab6814e_681ce013","updated":"2016-05-19 14:15:35.000000000","message":"Yes LisaLi, I will correct this in next patch, thanks for the check!","commit_id":"6551fd72bc20554af400b17a0a7434d16d95e827"},{"author":{"_account_id":15961,"name":"lisali","email":"xiaoyan.li@intel.com","username":"lisali"},"change_message_id":"54d442b8e0f0841b011ee0d01d9e1f290501477f","unresolved":false,"context_lines":[{"line_number":148,"context_line":""},{"line_number":149,"context_line":"Basicly none"},{"line_number":150,"context_line":""},{"line_number":151,"context_line":"Other deployer impact"},{"line_number":152,"context_line":"---------------------"},{"line_number":153,"context_line":""},{"line_number":154,"context_line":"None"}],"source_content_type":"text/x-rst","patch_set":15,"id":"bab6814e_683b4039","line":151,"updated":"2016-05-19 07:48:21.000000000","message":"Anything add in this part?\n\nAs the target storage needs to access the source volume.","commit_id":"6551fd72bc20554af400b17a0a7434d16d95e827"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"a54b4ea097557bb0b4852956a9246b6279c9a5d4","unresolved":false,"context_lines":[{"line_number":148,"context_line":""},{"line_number":149,"context_line":"Basicly none"},{"line_number":150,"context_line":""},{"line_number":151,"context_line":"Other deployer impact"},{"line_number":152,"context_line":"---------------------"},{"line_number":153,"context_line":""},{"line_number":154,"context_line":"None"}],"source_content_type":"text/x-rst","patch_set":15,"id":"bab6814e_981339b8","line":151,"in_reply_to":"bab6814e_683b4039","updated":"2016-05-19 14:15:35.000000000","message":"Thanks, will add this in next patch.","commit_id":"6551fd72bc20554af400b17a0a7434d16d95e827"},{"author":{"_account_id":15506,"name":"zhangli","email":"zhangli09@huawei.com"},"change_message_id":"76094cb34a0501712483c0aebb678101cd04f88c","unresolved":false,"context_lines":[{"line_number":99,"context_line":"  True and async as False, change the \"migration_status\" to \"success\" and"},{"line_number":100,"context_line":"  update the volume in db, then go to step 3. If migrate_volume_target()"},{"line_number":101,"context_line":"  returns moved as True and async as True, leaving the \"migration_status\" as"},{"line_number":102,"context_line":"  \"migrating\" and just go to step 3."},{"line_number":103,"context_line":""},{"line_number":104,"context_line":"3.Update the volume.host to the target host and update the db with"},{"line_number":105,"context_line":"  model_update. Now the volume is usable for server to perform read/write"}],"source_content_type":"text/x-rst","patch_set":17,"id":"bab6814e_1151bbd8","line":102,"range":{"start_line":102,"start_character":3,"end_line":102,"end_character":12},"updated":"2016-05-20 07:30:19.000000000","message":"How can a user distinguishs between sync and async migration process? The migration_status is the same as \"migrating\".How can a user knows whether it is  safe to attach the volume?","commit_id":"51e8df51264174eb8d7bbb5b296e808fef91aed0"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"d9c0b87389f9145d53739d3ba8947b87fa3c272e","unresolved":false,"context_lines":[{"line_number":99,"context_line":"  True and async as False, change the \"migration_status\" to \"success\" and"},{"line_number":100,"context_line":"  update the volume in db, then go to step 3. If migrate_volume_target()"},{"line_number":101,"context_line":"  returns moved as True and async as True, leaving the \"migration_status\" as"},{"line_number":102,"context_line":"  \"migrating\" and just go to step 3."},{"line_number":103,"context_line":""},{"line_number":104,"context_line":"3.Update the volume.host to the target host and update the db with"},{"line_number":105,"context_line":"  model_update. Now the volume is usable for server to perform read/write"}],"source_content_type":"text/x-rst","patch_set":17,"id":"bab6814e_b94e9138","line":102,"range":{"start_line":102,"start_character":3,"end_line":102,"end_character":12},"in_reply_to":"bab6814e_1151bbd8","updated":"2016-05-23 12:40:00.000000000","message":"Maybe we should change it to \"migrating_attable\" to let end users know whether it\u0027s safe to attach.","commit_id":"51e8df51264174eb8d7bbb5b296e808fef91aed0"},{"author":{"_account_id":8846,"name":"Hao Wang","display_name":"Hao Wang","email":"sxmatch1986@gmail.com","username":"sxmatch"},"change_message_id":"2899b04c9d892436c29a998ef91150f0daa3359f","unresolved":false,"context_lines":[{"line_number":38,"context_line":"1.Available volume migration between two different backends. These backends"},{"line_number":39,"context_line":"  could from one single vendor or different vendors."},{"line_number":40,"context_line":"2.Available volume migration between two pools of one single backend."},{"line_number":41,"context_line":"3.In-use(attached) volume migration using Cinder generic migration."},{"line_number":42,"context_line":""},{"line_number":43,"context_line":"This spec will only focus on case1, to make an available volume"},{"line_number":44,"context_line":"usable(can be attached to a server) immediately after we issue migration, no"}],"source_content_type":"text/x-rst","patch_set":19,"id":"9abb7d3a_da566cd1","line":41,"range":{"start_line":41,"start_character":0,"end_line":41,"end_character":67},"updated":"2016-05-30 09:32:04.000000000","message":"I think there is one more case: In-use volume migration using specific driver way.","commit_id":"709b8ad7e84ff9ea5ac49b85012e7c71947b48cb"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"d973a8f935de02625bc65973967969976ffcdca0","unresolved":false,"context_lines":[{"line_number":38,"context_line":"1.Available volume migration between two different backends. These backends"},{"line_number":39,"context_line":"  could from one single vendor or different vendors."},{"line_number":40,"context_line":"2.Available volume migration between two pools of one single backend."},{"line_number":41,"context_line":"3.In-use(attached) volume migration using Cinder generic migration."},{"line_number":42,"context_line":""},{"line_number":43,"context_line":"This spec will only focus on case1, to make an available volume"},{"line_number":44,"context_line":"usable(can be attached to a server) immediately after we issue migration, no"}],"source_content_type":"text/x-rst","patch_set":19,"id":"9abb7d3a_ca8af5be","line":41,"range":{"start_line":41,"start_character":0,"end_line":41,"end_character":67},"in_reply_to":"9abb7d3a_da566cd1","updated":"2016-05-31 11:35:50.000000000","message":"Thanks, I will add this in next patch.","commit_id":"709b8ad7e84ff9ea5ac49b85012e7c71947b48cb"},{"author":{"_account_id":8846,"name":"Hao Wang","display_name":"Hao Wang","email":"sxmatch1986@gmail.com","username":"sxmatch"},"change_message_id":"2899b04c9d892436c29a998ef91150f0daa3359f","unresolved":false,"context_lines":[{"line_number":40,"context_line":"2.Available volume migration between two pools of one single backend."},{"line_number":41,"context_line":"3.In-use(attached) volume migration using Cinder generic migration."},{"line_number":42,"context_line":""},{"line_number":43,"context_line":"This spec will only focus on case1, to make an available volume"},{"line_number":44,"context_line":"usable(can be attached to a server) immediately after we issue migration, no"},{"line_number":45,"context_line":"need to wait untill the migration complete."},{"line_number":46,"context_line":""}],"source_content_type":"text/x-rst","patch_set":19,"id":"9abb7d3a_1a0634ce","line":43,"range":{"start_line":43,"start_character":0,"end_line":43,"end_character":35},"updated":"2016-05-30 09:32:04.000000000","message":"So why we don\u0027t consider case2?  we also can attach volume in this case.","commit_id":"709b8ad7e84ff9ea5ac49b85012e7c71947b48cb"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"d973a8f935de02625bc65973967969976ffcdca0","unresolved":false,"context_lines":[{"line_number":40,"context_line":"2.Available volume migration between two pools of one single backend."},{"line_number":41,"context_line":"3.In-use(attached) volume migration using Cinder generic migration."},{"line_number":42,"context_line":""},{"line_number":43,"context_line":"This spec will only focus on case1, to make an available volume"},{"line_number":44,"context_line":"usable(can be attached to a server) immediately after we issue migration, no"},{"line_number":45,"context_line":"need to wait untill the migration complete."},{"line_number":46,"context_line":""}],"source_content_type":"text/x-rst","patch_set":19,"id":"9abb7d3a_8bb3c0e8","line":43,"range":{"start_line":43,"start_character":0,"end_line":43,"end_character":35},"in_reply_to":"9abb7d3a_1a0634ce","updated":"2016-05-31 11:35:50.000000000","message":"Case 2 has been well handled by most drivers, and it the routine goes into case 2, it won\u0027t goto case 1 again. So case 2 is out of the scope.","commit_id":"709b8ad7e84ff9ea5ac49b85012e7c71947b48cb"},{"author":{"_account_id":8846,"name":"Hao Wang","display_name":"Hao Wang","email":"sxmatch1986@gmail.com","username":"sxmatch"},"change_message_id":"2899b04c9d892436c29a998ef91150f0daa3359f","unresolved":false,"context_lines":[{"line_number":62,"context_line":"migration is completed, the local(target) LUN is exactly the same as the"},{"line_number":63,"context_line":"remote(source) LUN and no data will be lost."},{"line_number":64,"context_line":""},{"line_number":65,"context_line":"To enable one array to take over other array\u0027s volume, we should allow one"},{"line_number":66,"context_line":"driver to call other drivers\u0027 interfaces directly or indirectly. These two"},{"line_number":67,"context_line":"drivers are two independent backends, can be managed by one single volume"},{"line_number":68,"context_line":"node or two different volume nodes."},{"line_number":69,"context_line":""}],"source_content_type":"text/x-rst","patch_set":19,"id":"9abb7d3a_fa92504a","line":66,"range":{"start_line":65,"start_character":55,"end_line":66,"end_character":64},"updated":"2016-05-30 09:32:04.000000000","message":"So will this feature only be available between two independent backends, could it work in two pools of one backend? And the two backends could belong to two different vendor or same vendor, right?","commit_id":"709b8ad7e84ff9ea5ac49b85012e7c71947b48cb"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"d973a8f935de02625bc65973967969976ffcdca0","unresolved":false,"context_lines":[{"line_number":62,"context_line":"migration is completed, the local(target) LUN is exactly the same as the"},{"line_number":63,"context_line":"remote(source) LUN and no data will be lost."},{"line_number":64,"context_line":""},{"line_number":65,"context_line":"To enable one array to take over other array\u0027s volume, we should allow one"},{"line_number":66,"context_line":"driver to call other drivers\u0027 interfaces directly or indirectly. These two"},{"line_number":67,"context_line":"drivers are two independent backends, can be managed by one single volume"},{"line_number":68,"context_line":"node or two different volume nodes."},{"line_number":69,"context_line":""}],"source_content_type":"text/x-rst","patch_set":19,"id":"9abb7d3a_2ae101d1","line":66,"range":{"start_line":65,"start_character":55,"end_line":66,"end_character":64},"in_reply_to":"9abb7d3a_fa92504a","updated":"2016-05-31 11:35:50.000000000","message":"If they are two pools of one backend, the existing migrate_volume() will success and we won\u0027t go to migrate_volume_target() again. Yes, the two backends could belong to two different vendor or same vendor.","commit_id":"709b8ad7e84ff9ea5ac49b85012e7c71947b48cb"},{"author":{"_account_id":8846,"name":"Hao Wang","display_name":"Hao Wang","email":"sxmatch1986@gmail.com","username":"sxmatch"},"change_message_id":"2899b04c9d892436c29a998ef91150f0daa3359f","unresolved":false,"context_lines":[{"line_number":97,"context_line":"  perform the migration, we will go to _migrate_volume_generic() as usual to"},{"line_number":98,"context_line":"  perform host-copy migration. If migrate_volume_target() returns moved as"},{"line_number":99,"context_line":"  True and async as False, change the \"migration_status\" to \"success\" and"},{"line_number":100,"context_line":"  update the volume in db, then go to step 3. If migrate_volume_target()"},{"line_number":101,"context_line":"  returns moved as True and async as True, change the \"migration_status\" to"},{"line_number":102,"context_line":"  \"migrating_attachable\" and update the volume in db, then just go to step 3."},{"line_number":103,"context_line":"  \"migrating_attachable\" means this volume is migrating and is safe to attach"},{"line_number":104,"context_line":"  it to a server."},{"line_number":105,"context_line":""},{"line_number":106,"context_line":"3.Update the volume.host to the target host and update the db with"},{"line_number":107,"context_line":"  model_update. Now the volume is usable for server to perform read/write"}],"source_content_type":"text/x-rst","patch_set":19,"id":"9abb7d3a_fa66d027","line":104,"range":{"start_line":100,"start_character":45,"end_line":104,"end_character":17},"updated":"2016-05-30 09:32:04.000000000","message":"There is some confuse I feel, if async as True, and then user attach volume, so actually they attach the volume which is already existing on target host, but now driver is also copying data from remote LUN(source) to local LUN(target), so if user is writing data to this volume(local LUN), do it have some data corruption issue here?","commit_id":"709b8ad7e84ff9ea5ac49b85012e7c71947b48cb"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"d973a8f935de02625bc65973967969976ffcdca0","unresolved":false,"context_lines":[{"line_number":97,"context_line":"  perform the migration, we will go to _migrate_volume_generic() as usual to"},{"line_number":98,"context_line":"  perform host-copy migration. If migrate_volume_target() returns moved as"},{"line_number":99,"context_line":"  True and async as False, change the \"migration_status\" to \"success\" and"},{"line_number":100,"context_line":"  update the volume in db, then go to step 3. If migrate_volume_target()"},{"line_number":101,"context_line":"  returns moved as True and async as True, change the \"migration_status\" to"},{"line_number":102,"context_line":"  \"migrating_attachable\" and update the volume in db, then just go to step 3."},{"line_number":103,"context_line":"  \"migrating_attachable\" means this volume is migrating and is safe to attach"},{"line_number":104,"context_line":"  it to a server."},{"line_number":105,"context_line":""},{"line_number":106,"context_line":"3.Update the volume.host to the target host and update the db with"},{"line_number":107,"context_line":"  model_update. Now the volume is usable for server to perform read/write"}],"source_content_type":"text/x-rst","patch_set":19,"id":"9abb7d3a_7bec818f","line":104,"range":{"start_line":100,"start_character":45,"end_line":104,"end_character":17},"in_reply_to":"9abb7d3a_fa66d027","updated":"2016-05-31 11:35:50.000000000","message":"This is really import for this feature. If a driver developer implemented the migrate_volume_target() interface, it should make sure the corresponding backend storage is safe to write data from host when the data migration is undergoing, make sure there is no data corruption.","commit_id":"709b8ad7e84ff9ea5ac49b85012e7c71947b48cb"},{"author":{"_account_id":10213,"name":"Alon Marx","email":"alonma@il.ibm.com","username":"alonmarx"},"change_message_id":"33cfd0bd815e271fb82581b5faeb5c8515582678","unresolved":false,"context_lines":[{"line_number":19,"context_line":"is totally complete."},{"line_number":20,"context_line":""},{"line_number":21,"context_line":"Both the dd way or driver-specific way will fail if we attach a volume when"},{"line_number":22,"context_line":"it\u0027s under migrating:"},{"line_number":23,"context_line":"1.If we migrate the volume through dd way, the source or target volume is not"},{"line_number":24,"context_line":"   usable to server."},{"line_number":25,"context_line":"2.If we migrate the volume through driver-specific way, the migration task is"}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_d217d9a0","line":22,"range":{"start_line":22,"start_character":0,"end_line":22,"end_character":20},"updated":"2016-06-02 05:36:30.000000000","message":"change to \u0027when it is migrating\u0027","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"c064f99cb677f0843fa1363b5b9d7d724cece4da","unresolved":false,"context_lines":[{"line_number":19,"context_line":"is totally complete."},{"line_number":20,"context_line":""},{"line_number":21,"context_line":"Both the dd way or driver-specific way will fail if we attach a volume when"},{"line_number":22,"context_line":"it\u0027s under migrating:"},{"line_number":23,"context_line":"1.If we migrate the volume through dd way, the source or target volume is not"},{"line_number":24,"context_line":"   usable to server."},{"line_number":25,"context_line":"2.If we migrate the volume through driver-specific way, the migration task is"}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_a402ae09","line":22,"range":{"start_line":22,"start_character":0,"end_line":22,"end_character":20},"in_reply_to":"9abb7d3a_d217d9a0","updated":"2016-06-06 07:46:12.000000000","message":"Done","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":14103,"name":"wangtianfa","email":"wang.tianfa@99cloud.net","username":"wangtianfa"},"change_message_id":"300adfff4515f39d3c79e9198b9f508ed2373b09","unresolved":false,"context_lines":[{"line_number":19,"context_line":"is totally complete."},{"line_number":20,"context_line":""},{"line_number":21,"context_line":"Both the dd way or driver-specific way will fail if we attach a volume when"},{"line_number":22,"context_line":"it\u0027s under migrating:"},{"line_number":23,"context_line":"1.If we migrate the volume through dd way, the source or target volume is not"},{"line_number":24,"context_line":"   usable to server."},{"line_number":25,"context_line":"2.If we migrate the volume through driver-specific way, the migration task is"}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_2787f7b5","line":22,"range":{"start_line":22,"start_character":0,"end_line":22,"end_character":20},"in_reply_to":"9abb7d3a_d217d9a0","updated":"2016-06-02 05:39:36.000000000","message":"I think so","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":10213,"name":"Alon Marx","email":"alonma@il.ibm.com","username":"alonmarx"},"change_message_id":"33cfd0bd815e271fb82581b5faeb5c8515582678","unresolved":false,"context_lines":[{"line_number":34,"context_line":"Use Cases"},{"line_number":35,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":36,"context_line":""},{"line_number":37,"context_line":"Currently cinder supports three cases for volume migration:"},{"line_number":38,"context_line":"1.Available volume migration between two different backends. These backends"},{"line_number":39,"context_line":"  could from one single vendor or different vendors."},{"line_number":40,"context_line":"2.Available volume migration between two pools of one single backend."}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_32d0e5c8","line":37,"range":{"start_line":37,"start_character":26,"end_line":37,"end_character":31},"updated":"2016-06-02 05:36:30.000000000","message":"four cases are listed below","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"c064f99cb677f0843fa1363b5b9d7d724cece4da","unresolved":false,"context_lines":[{"line_number":34,"context_line":"Use Cases"},{"line_number":35,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":36,"context_line":""},{"line_number":37,"context_line":"Currently cinder supports three cases for volume migration:"},{"line_number":38,"context_line":"1.Available volume migration between two different backends. These backends"},{"line_number":39,"context_line":"  could from one single vendor or different vendors."},{"line_number":40,"context_line":"2.Available volume migration between two pools of one single backend."}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_ca1c196d","line":37,"range":{"start_line":37,"start_character":26,"end_line":37,"end_character":31},"in_reply_to":"9abb7d3a_32d0e5c8","updated":"2016-06-06 07:46:12.000000000","message":"Done","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":10213,"name":"Alon Marx","email":"alonma@il.ibm.com","username":"alonmarx"},"change_message_id":"33cfd0bd815e271fb82581b5faeb5c8515582678","unresolved":false,"context_lines":[{"line_number":36,"context_line":""},{"line_number":37,"context_line":"Currently cinder supports three cases for volume migration:"},{"line_number":38,"context_line":"1.Available volume migration between two different backends. These backends"},{"line_number":39,"context_line":"  could from one single vendor or different vendors."},{"line_number":40,"context_line":"2.Available volume migration between two pools of one single backend."},{"line_number":41,"context_line":"3.In-use(attached) volume migration using driver specific way."},{"line_number":42,"context_line":"4.In-use(attached) volume migration using Cinder generic migration."}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_3259058f","line":39,"updated":"2016-06-02 05:36:30.000000000","message":"...could be from...","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"c064f99cb677f0843fa1363b5b9d7d724cece4da","unresolved":false,"context_lines":[{"line_number":36,"context_line":""},{"line_number":37,"context_line":"Currently cinder supports three cases for volume migration:"},{"line_number":38,"context_line":"1.Available volume migration between two different backends. These backends"},{"line_number":39,"context_line":"  could from one single vendor or different vendors."},{"line_number":40,"context_line":"2.Available volume migration between two pools of one single backend."},{"line_number":41,"context_line":"3.In-use(attached) volume migration using driver specific way."},{"line_number":42,"context_line":"4.In-use(attached) volume migration using Cinder generic migration."}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_ca61b9d5","line":39,"in_reply_to":"9abb7d3a_3259058f","updated":"2016-06-06 07:46:12.000000000","message":"Done","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":10213,"name":"Alon Marx","email":"alonma@il.ibm.com","username":"alonmarx"},"change_message_id":"33cfd0bd815e271fb82581b5faeb5c8515582678","unresolved":false,"context_lines":[{"line_number":44,"context_line":"This spec will only focus on case 1, to make an available volume"},{"line_number":45,"context_line":"usable(can be attached to a server) immediately after we issue migration, no"},{"line_number":46,"context_line":"need to wait untill the migration complete. Case 2 is already well handled by"},{"line_number":47,"context_line":"most drivers, and if we go into case 2, we won\u0027t go to case 1 again, so this"},{"line_number":48,"context_line":"spec will focus on case 1."},{"line_number":49,"context_line":""},{"line_number":50,"context_line":"Proposed change"},{"line_number":51,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_72b38d98","line":48,"range":{"start_line":47,"start_character":12,"end_line":48,"end_character":25},"updated":"2016-06-02 05:36:30.000000000","message":"The fact that most drivers implement this seems enough. No need in the rest of the sentence.","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"c064f99cb677f0843fa1363b5b9d7d724cece4da","unresolved":false,"context_lines":[{"line_number":44,"context_line":"This spec will only focus on case 1, to make an available volume"},{"line_number":45,"context_line":"usable(can be attached to a server) immediately after we issue migration, no"},{"line_number":46,"context_line":"need to wait untill the migration complete. Case 2 is already well handled by"},{"line_number":47,"context_line":"most drivers, and if we go into case 2, we won\u0027t go to case 1 again, so this"},{"line_number":48,"context_line":"spec will focus on case 1."},{"line_number":49,"context_line":""},{"line_number":50,"context_line":"Proposed change"},{"line_number":51,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":21,"id":"7aa08908_b22c9fac","line":48,"range":{"start_line":47,"start_character":12,"end_line":48,"end_character":25},"in_reply_to":"9abb7d3a_72b38d98","updated":"2016-06-06 07:46:12.000000000","message":"For the current implementation of most drivers, users need to wait till the whole migration is totally finished before using this volume, but now the volume is usable almost immediately. I think this gives users a big convenience and so it make sense. Thanks!","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":10213,"name":"Alon Marx","email":"alonma@il.ibm.com","username":"alonmarx"},"change_message_id":"33cfd0bd815e271fb82581b5faeb5c8515582678","unresolved":false,"context_lines":[{"line_number":72,"context_line":""},{"line_number":73,"context_line":"To enable an available volume usable(attach to a server) immediately"},{"line_number":74,"context_line":"after we issued migration from one backend to another backend, we should"},{"line_number":75,"context_line":"ALLOW the migration task to be send to the TARGET backend."},{"line_number":76,"context_line":""},{"line_number":77,"context_line":"There will be a new interface for driver which will be called"},{"line_number":78,"context_line":"\u0027migrate_volume_target\u0027. We introduce a new interface instead of use the"}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_f0949537","line":75,"range":{"start_line":75,"start_character":25,"end_line":75,"end_character":57},"updated":"2016-06-02 05:36:30.000000000","message":"For some storage types (e.g. XIV) the work should be done on the source backend. For the general case I suggest the manager should call both ends (e.g. migrate_volume_target and migrate_volume_source).\nAlso, there needs to be some reference to the remote side (at least in case of calling the source backend, it requires knowing who the target is).","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"c064f99cb677f0843fa1363b5b9d7d724cece4da","unresolved":false,"context_lines":[{"line_number":72,"context_line":""},{"line_number":73,"context_line":"To enable an available volume usable(attach to a server) immediately"},{"line_number":74,"context_line":"after we issued migration from one backend to another backend, we should"},{"line_number":75,"context_line":"ALLOW the migration task to be send to the TARGET backend."},{"line_number":76,"context_line":""},{"line_number":77,"context_line":"There will be a new interface for driver which will be called"},{"line_number":78,"context_line":"\u0027migrate_volume_target\u0027. We introduce a new interface instead of use the"}],"source_content_type":"text/x-rst","patch_set":21,"id":"7aa08908_4d44bc73","line":75,"range":{"start_line":75,"start_character":25,"end_line":75,"end_character":57},"in_reply_to":"9abb7d3a_f0949537","updated":"2016-06-06 07:46:12.000000000","message":"For these storage types that should do the work on the source backend, the existing migration code is already OK to work, because the existing migration code will send the migration task to source backend by default and call source driver\u0027s migrate_volume(). You can just implement your code in the existing migrate_volume().","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":15506,"name":"zhangli","email":"zhangli09@huawei.com"},"change_message_id":"ef34bb45d1dd7aa17630f3c3321b02790abf1785","unresolved":false,"context_lines":[{"line_number":77,"context_line":"There will be a new interface for driver which will be called"},{"line_number":78,"context_line":"\u0027migrate_volume_target\u0027. We introduce a new interface instead of use the"},{"line_number":79,"context_line":"existing \u0027migrate_volume\u0027 interface because there would be lots of"},{"line_number":80,"context_line":"differences between them, a significant difference is that the existing"},{"line_number":81,"context_line":"\u0027migrate_volume\u0027 interface always take them self as the source backend, but"},{"line_number":82,"context_line":"the new interface should take itself as the target backend."},{"line_number":83,"context_line":""},{"line_number":84,"context_line":"Some change will be made in the volume/manager.py for migrate_volume routine:"}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_410bf47d","line":81,"range":{"start_line":80,"start_character":59,"end_line":81,"end_character":26},"updated":"2016-06-02 12:32:55.000000000","message":"the drivers executing the \u0027migrate_volume\u0027 interface","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"c064f99cb677f0843fa1363b5b9d7d724cece4da","unresolved":false,"context_lines":[{"line_number":77,"context_line":"There will be a new interface for driver which will be called"},{"line_number":78,"context_line":"\u0027migrate_volume_target\u0027. We introduce a new interface instead of use the"},{"line_number":79,"context_line":"existing \u0027migrate_volume\u0027 interface because there would be lots of"},{"line_number":80,"context_line":"differences between them, a significant difference is that the existing"},{"line_number":81,"context_line":"\u0027migrate_volume\u0027 interface always take them self as the source backend, but"},{"line_number":82,"context_line":"the new interface should take itself as the target backend."},{"line_number":83,"context_line":""},{"line_number":84,"context_line":"Some change will be made in the volume/manager.py for migrate_volume routine:"}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_6af52dca","line":81,"range":{"start_line":80,"start_character":59,"end_line":81,"end_character":26},"in_reply_to":"9abb7d3a_410bf47d","updated":"2016-06-06 07:46:12.000000000","message":"Done","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":17045,"name":"xiaoqin","email":"xiaoqin.li@hotmail.com","username":"xiaoqin"},"change_message_id":"732129b9c03fd94e5d1367de112791db4bd72b5d","unresolved":false,"context_lines":[{"line_number":79,"context_line":"existing \u0027migrate_volume\u0027 interface because there would be lots of"},{"line_number":80,"context_line":"differences between them, a significant difference is that the existing"},{"line_number":81,"context_line":"\u0027migrate_volume\u0027 interface always take them self as the source backend, but"},{"line_number":82,"context_line":"the new interface should take itself as the target backend."},{"line_number":83,"context_line":""},{"line_number":84,"context_line":"Some change will be made in the volume/manager.py for migrate_volume routine:"},{"line_number":85,"context_line":""}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_991a317b","line":82,"range":{"start_line":82,"start_character":0,"end_line":82,"end_character":59},"updated":"2016-06-03 11:57:01.000000000","message":"So the migration task is issued to the migration destination backend by scheduler? And volume attach is also issued to the migration destination backend by scheduler after issue cinder migrate/nova volume-attach command?","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"c064f99cb677f0843fa1363b5b9d7d724cece4da","unresolved":false,"context_lines":[{"line_number":79,"context_line":"existing \u0027migrate_volume\u0027 interface because there would be lots of"},{"line_number":80,"context_line":"differences between them, a significant difference is that the existing"},{"line_number":81,"context_line":"\u0027migrate_volume\u0027 interface always take them self as the source backend, but"},{"line_number":82,"context_line":"the new interface should take itself as the target backend."},{"line_number":83,"context_line":""},{"line_number":84,"context_line":"Some change will be made in the volume/manager.py for migrate_volume routine:"},{"line_number":85,"context_line":""}],"source_content_type":"text/x-rst","patch_set":21,"id":"7aa08908_cdfc2c41","line":82,"range":{"start_line":82,"start_character":0,"end_line":82,"end_character":59},"in_reply_to":"9abb7d3a_991a317b","updated":"2016-06-06 07:46:12.000000000","message":"We will firstly call source backend\u0027s migrati_volume(), and only if the source backend\u0027s migrati_volume() returned \"removed\" as false, then we call migration destination backend\u0027s migration_volume_target() to give the destination backend a chance to perform migration.\n\nIf the migration_status of the volume if \"migrating_attachable\", the host of the volume is already changed to the destination backend so latter operations to the volume will all be issued to the migration destination backend by scheduler.","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":17045,"name":"xiaoqin","email":"xiaoqin.li@hotmail.com","username":"xiaoqin"},"change_message_id":"732129b9c03fd94e5d1367de112791db4bd72b5d","unresolved":false,"context_lines":[{"line_number":83,"context_line":""},{"line_number":84,"context_line":"Some change will be made in the volume/manager.py for migrate_volume routine:"},{"line_number":85,"context_line":""},{"line_number":86,"context_line":"1.If not force_host_copy and new_type_id is None, firstly call source backend"},{"line_number":87,"context_line":"  driver\u0027s migrate_volume(). If source backend driver\u0027s migrate_volume() return"},{"line_number":88,"context_line":"  True which means it has migrated successfully, change the \"migration_status\""},{"line_number":89,"context_line":"  to \"success\" and update the volume, then go to step 3. If source backend"},{"line_number":90,"context_line":"  driver\u0027s migrate_volume() returned False, go to step 2."},{"line_number":91,"context_line":""},{"line_number":92,"context_line":"2.Call target backend driver\u0027s migrate_volume_target() through rpcapi, give a"},{"line_number":93,"context_line":"  chance for target backend to perform the migration. Give target backend a"}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_22f30ea1","line":90,"range":{"start_line":86,"start_character":0,"end_line":90,"end_character":57},"updated":"2016-06-03 11:57:01.000000000","message":"What is the behavior once the volume is in CG? For example, the source volume is in the CG　of source backend, would it be removed from CG after the migration or the migration will be refused? Also how to handle it one the volume has snapshot?","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"c064f99cb677f0843fa1363b5b9d7d724cece4da","unresolved":false,"context_lines":[{"line_number":83,"context_line":""},{"line_number":84,"context_line":"Some change will be made in the volume/manager.py for migrate_volume routine:"},{"line_number":85,"context_line":""},{"line_number":86,"context_line":"1.If not force_host_copy and new_type_id is None, firstly call source backend"},{"line_number":87,"context_line":"  driver\u0027s migrate_volume(). If source backend driver\u0027s migrate_volume() return"},{"line_number":88,"context_line":"  True which means it has migrated successfully, change the \"migration_status\""},{"line_number":89,"context_line":"  to \"success\" and update the volume, then go to step 3. If source backend"},{"line_number":90,"context_line":"  driver\u0027s migrate_volume() returned False, go to step 2."},{"line_number":91,"context_line":""},{"line_number":92,"context_line":"2.Call target backend driver\u0027s migrate_volume_target() through rpcapi, give a"},{"line_number":93,"context_line":"  chance for target backend to perform the migration. Give target backend a"}],"source_content_type":"text/x-rst","patch_set":21,"id":"7aa08908_6da09891","line":90,"range":{"start_line":86,"start_character":0,"end_line":90,"end_character":57},"in_reply_to":"9abb7d3a_22f30ea1","updated":"2016-06-06 07:46:12.000000000","message":"In the existing migration code we don\u0027t check if the volume is in CG or not, and  I\u0027m not sure this is a bug or not. So I won\u0027t take care of the CG as the existing migration doesn\u0027t take care of the CG. Same for the volume has snapshot.","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":10213,"name":"Alon Marx","email":"alonma@il.ibm.com","username":"alonmarx"},"change_message_id":"33cfd0bd815e271fb82581b5faeb5c8515582678","unresolved":false,"context_lines":[{"line_number":105,"context_line":"  \"migrating_attachable\" and update the volume in db, then just go to step 3."},{"line_number":106,"context_line":"  \"migrating_attachable\" means this volume is migrating and is safe to attach"},{"line_number":107,"context_line":"  it to a server. Note that driver developer should make sure the volume on"},{"line_number":108,"context_line":"  the orresponding backend is safe to read/write and not data corruption"},{"line_number":109,"context_line":"  will occur while the migration is undergoing before implemente the"},{"line_number":110,"context_line":"  migrate_volume_target() interface."},{"line_number":111,"context_line":""}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_359e0f07","line":108,"range":{"start_line":108,"start_character":6,"end_line":108,"end_character":18},"updated":"2016-06-02 05:36:30.000000000","message":"corresponding","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":10213,"name":"Alon Marx","email":"alonma@il.ibm.com","username":"alonmarx"},"change_message_id":"33cfd0bd815e271fb82581b5faeb5c8515582678","unresolved":false,"context_lines":[{"line_number":105,"context_line":"  \"migrating_attachable\" and update the volume in db, then just go to step 3."},{"line_number":106,"context_line":"  \"migrating_attachable\" means this volume is migrating and is safe to attach"},{"line_number":107,"context_line":"  it to a server. Note that driver developer should make sure the volume on"},{"line_number":108,"context_line":"  the orresponding backend is safe to read/write and not data corruption"},{"line_number":109,"context_line":"  will occur while the migration is undergoing before implemente the"},{"line_number":110,"context_line":"  migrate_volume_target() interface."},{"line_number":111,"context_line":""}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_75053744","line":108,"range":{"start_line":108,"start_character":53,"end_line":108,"end_character":56},"updated":"2016-06-02 05:36:30.000000000","message":"not\u003d\u003e no","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"c064f99cb677f0843fa1363b5b9d7d724cece4da","unresolved":false,"context_lines":[{"line_number":105,"context_line":"  \"migrating_attachable\" and update the volume in db, then just go to step 3."},{"line_number":106,"context_line":"  \"migrating_attachable\" means this volume is migrating and is safe to attach"},{"line_number":107,"context_line":"  it to a server. Note that driver developer should make sure the volume on"},{"line_number":108,"context_line":"  the orresponding backend is safe to read/write and not data corruption"},{"line_number":109,"context_line":"  will occur while the migration is undergoing before implemente the"},{"line_number":110,"context_line":"  migrate_volume_target() interface."},{"line_number":111,"context_line":""}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_2a168555","line":108,"range":{"start_line":108,"start_character":6,"end_line":108,"end_character":18},"in_reply_to":"9abb7d3a_359e0f07","updated":"2016-06-06 07:46:12.000000000","message":"Done","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"c064f99cb677f0843fa1363b5b9d7d724cece4da","unresolved":false,"context_lines":[{"line_number":105,"context_line":"  \"migrating_attachable\" and update the volume in db, then just go to step 3."},{"line_number":106,"context_line":"  \"migrating_attachable\" means this volume is migrating and is safe to attach"},{"line_number":107,"context_line":"  it to a server. Note that driver developer should make sure the volume on"},{"line_number":108,"context_line":"  the orresponding backend is safe to read/write and not data corruption"},{"line_number":109,"context_line":"  will occur while the migration is undergoing before implemente the"},{"line_number":110,"context_line":"  migrate_volume_target() interface."},{"line_number":111,"context_line":""}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_8a0b312b","line":108,"range":{"start_line":108,"start_character":53,"end_line":108,"end_character":56},"in_reply_to":"9abb7d3a_75053744","updated":"2016-06-06 07:46:12.000000000","message":"Done","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":10213,"name":"Alon Marx","email":"alonma@il.ibm.com","username":"alonmarx"},"change_message_id":"33cfd0bd815e271fb82581b5faeb5c8515582678","unresolved":false,"context_lines":[{"line_number":106,"context_line":"  \"migrating_attachable\" means this volume is migrating and is safe to attach"},{"line_number":107,"context_line":"  it to a server. Note that driver developer should make sure the volume on"},{"line_number":108,"context_line":"  the orresponding backend is safe to read/write and not data corruption"},{"line_number":109,"context_line":"  will occur while the migration is undergoing before implemente the"},{"line_number":110,"context_line":"  migrate_volume_target() interface."},{"line_number":111,"context_line":""},{"line_number":112,"context_line":"3.Update the volume.host to the target host and update the db with"}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_75b37798","line":109,"range":{"start_line":109,"start_character":54,"end_line":109,"end_character":64},"updated":"2016-06-02 05:36:30.000000000","message":"implementing?","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"c064f99cb677f0843fa1363b5b9d7d724cece4da","unresolved":false,"context_lines":[{"line_number":106,"context_line":"  \"migrating_attachable\" means this volume is migrating and is safe to attach"},{"line_number":107,"context_line":"  it to a server. Note that driver developer should make sure the volume on"},{"line_number":108,"context_line":"  the orresponding backend is safe to read/write and not data corruption"},{"line_number":109,"context_line":"  will occur while the migration is undergoing before implemente the"},{"line_number":110,"context_line":"  migrate_volume_target() interface."},{"line_number":111,"context_line":""},{"line_number":112,"context_line":"3.Update the volume.host to the target host and update the db with"}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_55360ae8","line":109,"range":{"start_line":109,"start_character":54,"end_line":109,"end_character":64},"in_reply_to":"9abb7d3a_75b37798","updated":"2016-06-06 07:46:12.000000000","message":"Done","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":1207,"name":"Duncan Thomas","email":"duncan.thomas@gmail.com","username":"duncan-thomas"},"change_message_id":"94a067584cba00711c53dfdbdfe07210416a5775","unresolved":false,"context_lines":[{"line_number":147,"context_line":"Notifications impact"},{"line_number":148,"context_line":"--------------------"},{"line_number":149,"context_line":""},{"line_number":150,"context_line":"None"},{"line_number":151,"context_line":""},{"line_number":152,"context_line":"Other end user impact"},{"line_number":153,"context_line":"---------------------"}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_d319a823","line":150,"range":{"start_line":150,"start_character":0,"end_line":150,"end_character":4},"updated":"2016-06-01 17:16:12.000000000","message":"We probably want notifications that a driver assisted read-through migration has started and when it finishes","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"c064f99cb677f0843fa1363b5b9d7d724cece4da","unresolved":false,"context_lines":[{"line_number":147,"context_line":"Notifications impact"},{"line_number":148,"context_line":"--------------------"},{"line_number":149,"context_line":""},{"line_number":150,"context_line":"None"},{"line_number":151,"context_line":""},{"line_number":152,"context_line":"Other end user impact"},{"line_number":153,"context_line":"---------------------"}],"source_content_type":"text/x-rst","patch_set":21,"id":"7aa08908_d255ebc6","line":150,"range":{"start_line":150,"start_character":0,"end_line":150,"end_character":4},"in_reply_to":"9abb7d3a_d319a823","updated":"2016-06-06 07:46:12.000000000","message":"Currently there is no direct notifications. But users can know the status of the migration by \"cinder show\" command from the CLI.","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":10213,"name":"Alon Marx","email":"alonma@il.ibm.com","username":"alonmarx"},"change_message_id":"33cfd0bd815e271fb82581b5faeb5c8515582678","unresolved":false,"context_lines":[{"line_number":151,"context_line":""},{"line_number":152,"context_line":"Other end user impact"},{"line_number":153,"context_line":"---------------------"},{"line_number":154,"context_line":""},{"line_number":155,"context_line":"None"},{"line_number":156,"context_line":""},{"line_number":157,"context_line":"Performance Impact"}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_15594b35","line":154,"updated":"2016-06-02 05:36:30.000000000","message":"What operations are permitted during the migration? And what is not permitted? E.g. what happens when retype is called during migration..","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"c064f99cb677f0843fa1363b5b9d7d724cece4da","unresolved":false,"context_lines":[{"line_number":151,"context_line":""},{"line_number":152,"context_line":"Other end user impact"},{"line_number":153,"context_line":"---------------------"},{"line_number":154,"context_line":""},{"line_number":155,"context_line":"None"},{"line_number":156,"context_line":""},{"line_number":157,"context_line":"Performance Impact"}],"source_content_type":"text/x-rst","patch_set":21,"id":"7aa08908_32b54fcd","line":154,"in_reply_to":"9abb7d3a_15594b35","updated":"2016-06-06 07:46:12.000000000","message":"For the permitted and not permitted operations, I think it is same as the existing migration. And I have addressed this in next patch, thanks!","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":1207,"name":"Duncan Thomas","email":"duncan.thomas@gmail.com","username":"duncan-thomas"},"change_message_id":"94a067584cba00711c53dfdbdfe07210416a5775","unresolved":false,"context_lines":[{"line_number":157,"context_line":"Performance Impact"},{"line_number":158,"context_line":"------------------"},{"line_number":159,"context_line":""},{"line_number":160,"context_line":"Basicly none"},{"line_number":161,"context_line":""},{"line_number":162,"context_line":"Other deployer impact"},{"line_number":163,"context_line":"---------------------"}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_d3f088cb","line":160,"range":{"start_line":160,"start_character":0,"end_line":160,"end_character":12},"updated":"2016-06-01 17:16:12.000000000","message":"Is volume performance not affected by the read-through process?","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":4523,"name":"Eric Harney","email":"eharney@redhat.com","username":"eharney"},"change_message_id":"2460ae64583bb9cfba11517d9d5e3495aa580ade","unresolved":false,"context_lines":[{"line_number":157,"context_line":"Performance Impact"},{"line_number":158,"context_line":"------------------"},{"line_number":159,"context_line":""},{"line_number":160,"context_line":"Basicly none"},{"line_number":161,"context_line":""},{"line_number":162,"context_line":"Other deployer impact"},{"line_number":163,"context_line":"---------------------"}],"source_content_type":"text/x-rst","patch_set":21,"id":"9abb7d3a_018e3fcf","line":160,"updated":"2016-06-01 16:32:35.000000000","message":"Isn\u0027t the whole motivation for this feature to be able to perform migrations in a more efficient way?\n\nSo, it should have a positive performance impact when used in the right configuration?","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"c064f99cb677f0843fa1363b5b9d7d724cece4da","unresolved":false,"context_lines":[{"line_number":157,"context_line":"Performance Impact"},{"line_number":158,"context_line":"------------------"},{"line_number":159,"context_line":""},{"line_number":160,"context_line":"Basicly none"},{"line_number":161,"context_line":""},{"line_number":162,"context_line":"Other deployer impact"},{"line_number":163,"context_line":"---------------------"}],"source_content_type":"text/x-rst","patch_set":21,"id":"7aa08908_43aad3d5","line":160,"in_reply_to":"9abb7d3a_018e3fcf","updated":"2016-06-06 07:46:12.000000000","message":"Yes, if there is no read-through process the performance should be enhanced.","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"c064f99cb677f0843fa1363b5b9d7d724cece4da","unresolved":false,"context_lines":[{"line_number":157,"context_line":"Performance Impact"},{"line_number":158,"context_line":"------------------"},{"line_number":159,"context_line":""},{"line_number":160,"context_line":"Basicly none"},{"line_number":161,"context_line":""},{"line_number":162,"context_line":"Other deployer impact"},{"line_number":163,"context_line":"---------------------"}],"source_content_type":"text/x-rst","patch_set":21,"id":"7aa08908_a3dd7772","line":160,"range":{"start_line":160,"start_character":0,"end_line":160,"end_character":12},"in_reply_to":"9abb7d3a_d3f088cb","updated":"2016-06-06 07:46:12.000000000","message":"Thanks Duncan, I will address in next patch.","commit_id":"a02980a02bdc5e6301e922fd33a43ed64390ec1f"},{"author":{"_account_id":8846,"name":"Hao Wang","display_name":"Hao Wang","email":"sxmatch1986@gmail.com","username":"sxmatch"},"change_message_id":"d571268f2702aa6e9311c01dc135744bba1b5021","unresolved":false,"context_lines":[{"line_number":20,"context_line":""},{"line_number":21,"context_line":"Both the dd way or driver-specific way will fail if we attach a volume when"},{"line_number":22,"context_line":"it is migrating:"},{"line_number":23,"context_line":"1.If we migrate the volume through dd way, the source or target volume is not"},{"line_number":24,"context_line":"   usable to server."},{"line_number":25,"context_line":"2.If we migrate the volume through driver-specific way, the migration task is"},{"line_number":26,"context_line":"  issued to the source backend by scheduler and the volume is owned by the"}],"source_content_type":"text/x-rst","patch_set":22,"id":"7aa08908_b8ad9d33","line":23,"range":{"start_line":23,"start_character":0,"end_line":23,"end_character":2},"updated":"2016-06-06 08:47:23.000000000","message":"For spec formatting, I don\u0027t suggest to use this list, you should use list like this: http://docs.openstack.org/contributor-guide/rst-conv/lists.html.","commit_id":"eaf4dba2aaf655e9ca25c6bcb250cfbb7363bd60"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"1446e1bf144017187be63acd25a07fa2e08617b5","unresolved":false,"context_lines":[{"line_number":20,"context_line":""},{"line_number":21,"context_line":"Both the dd way or driver-specific way will fail if we attach a volume when"},{"line_number":22,"context_line":"it is migrating:"},{"line_number":23,"context_line":"1.If we migrate the volume through dd way, the source or target volume is not"},{"line_number":24,"context_line":"   usable to server."},{"line_number":25,"context_line":"2.If we migrate the volume through driver-specific way, the migration task is"},{"line_number":26,"context_line":"  issued to the source backend by scheduler and the volume is owned by the"}],"source_content_type":"text/x-rst","patch_set":22,"id":"7aa08908_86031e73","line":23,"range":{"start_line":23,"start_character":0,"end_line":23,"end_character":2},"in_reply_to":"7aa08908_b8ad9d33","updated":"2016-06-06 09:33:14.000000000","message":"Thanks, I will fix this latter.","commit_id":"eaf4dba2aaf655e9ca25c6bcb250cfbb7363bd60"},{"author":{"_account_id":8846,"name":"Hao Wang","display_name":"Hao Wang","email":"sxmatch1986@gmail.com","username":"sxmatch"},"change_message_id":"d571268f2702aa6e9311c01dc135744bba1b5021","unresolved":false,"context_lines":[{"line_number":34,"context_line":"Use Cases"},{"line_number":35,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":36,"context_line":""},{"line_number":37,"context_line":"Currently cinder supports four cases for volume migration:"},{"line_number":38,"context_line":"1.Available volume migration between two different backends. These backends"},{"line_number":39,"context_line":"  could be from one single vendor or different vendors."},{"line_number":40,"context_line":"2.Available volume migration between two pools of one single backend."}],"source_content_type":"text/x-rst","patch_set":22,"id":"7aa08908_b8747dcf","line":37,"updated":"2016-06-06 08:47:23.000000000","message":"need a blank line here.","commit_id":"eaf4dba2aaf655e9ca25c6bcb250cfbb7363bd60"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"1446e1bf144017187be63acd25a07fa2e08617b5","unresolved":false,"context_lines":[{"line_number":34,"context_line":"Use Cases"},{"line_number":35,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":36,"context_line":""},{"line_number":37,"context_line":"Currently cinder supports four cases for volume migration:"},{"line_number":38,"context_line":"1.Available volume migration between two different backends. These backends"},{"line_number":39,"context_line":"  could be from one single vendor or different vendors."},{"line_number":40,"context_line":"2.Available volume migration between two pools of one single backend."}],"source_content_type":"text/x-rst","patch_set":22,"id":"7aa08908_e6f8227f","line":37,"in_reply_to":"7aa08908_b8747dcf","updated":"2016-06-06 09:33:14.000000000","message":"Done","commit_id":"eaf4dba2aaf655e9ca25c6bcb250cfbb7363bd60"},{"author":{"_account_id":8846,"name":"Hao Wang","display_name":"Hao Wang","email":"sxmatch1986@gmail.com","username":"sxmatch"},"change_message_id":"d571268f2702aa6e9311c01dc135744bba1b5021","unresolved":false,"context_lines":[{"line_number":57,"context_line":"migration or host-copy way. We won\u0027t affect anything to the existing routine,"},{"line_number":58,"context_line":"just add a new way for developers or users to choose."},{"line_number":59,"context_line":""},{"line_number":60,"context_line":"These features are:"},{"line_number":61,"context_line":"1. One array can take over other array\u0027s LUN as a remote LUN if these two"},{"line_number":62,"context_line":"arrays are connected with FC or iSCSI fabric."},{"line_number":63,"context_line":"2. We can migrate a remote LUN to a local LUN and meanwhile the remote(source)"}],"source_content_type":"text/x-rst","patch_set":22,"id":"7aa08908_384fed93","line":60,"updated":"2016-06-06 08:47:23.000000000","message":"need a blank line here.","commit_id":"eaf4dba2aaf655e9ca25c6bcb250cfbb7363bd60"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"1446e1bf144017187be63acd25a07fa2e08617b5","unresolved":false,"context_lines":[{"line_number":57,"context_line":"migration or host-copy way. We won\u0027t affect anything to the existing routine,"},{"line_number":58,"context_line":"just add a new way for developers or users to choose."},{"line_number":59,"context_line":""},{"line_number":60,"context_line":"These features are:"},{"line_number":61,"context_line":"1. One array can take over other array\u0027s LUN as a remote LUN if these two"},{"line_number":62,"context_line":"arrays are connected with FC or iSCSI fabric."},{"line_number":63,"context_line":"2. We can migrate a remote LUN to a local LUN and meanwhile the remote(source)"}],"source_content_type":"text/x-rst","patch_set":22,"id":"7aa08908_06f0ce8f","line":60,"in_reply_to":"7aa08908_384fed93","updated":"2016-06-06 09:33:14.000000000","message":"Done","commit_id":"eaf4dba2aaf655e9ca25c6bcb250cfbb7363bd60"},{"author":{"_account_id":7198,"name":"Jay Bryant","email":"jungleboyj@electronicjungle.net","username":"jsbryant"},"change_message_id":"2ca1063637bd1f356cbf71d23233b28f9e4e6497","unresolved":false,"context_lines":[{"line_number":15,"context_line":""},{"line_number":16,"context_line":"Currently if we migrate a volume from one backend to another backend,"},{"line_number":17,"context_line":"especially when these two backends are handling two different vendors of"},{"line_number":18,"context_line":"arrays, we can\u0027t attach this migrating volume to server untill the migration"},{"line_number":19,"context_line":"is totally complete."},{"line_number":20,"context_line":""},{"line_number":21,"context_line":"Both the dd way or driver-specific way will fail if we attach a volume when"}],"source_content_type":"text/x-rst","patch_set":23,"id":"7aa08908_27d10f24","line":18,"range":{"start_line":18,"start_character":56,"end_line":18,"end_character":62},"updated":"2016-06-07 23:46:29.000000000","message":"untill -\u003e until","commit_id":"22ab8201d2af78a9210f190f0478f47a72415849"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"de238786725180b436ced176c30264762af30b5e","unresolved":false,"context_lines":[{"line_number":15,"context_line":""},{"line_number":16,"context_line":"Currently if we migrate a volume from one backend to another backend,"},{"line_number":17,"context_line":"especially when these two backends are handling two different vendors of"},{"line_number":18,"context_line":"arrays, we can\u0027t attach this migrating volume to server untill the migration"},{"line_number":19,"context_line":"is totally complete."},{"line_number":20,"context_line":""},{"line_number":21,"context_line":"Both the dd way or driver-specific way will fail if we attach a volume when"}],"source_content_type":"text/x-rst","patch_set":23,"id":"7aa08908_9d1ad67b","line":18,"range":{"start_line":18,"start_character":56,"end_line":18,"end_character":62},"in_reply_to":"7aa08908_27d10f24","updated":"2016-06-08 01:22:15.000000000","message":"I will fix this if I need commit another patch, thanks!","commit_id":"22ab8201d2af78a9210f190f0478f47a72415849"},{"author":{"_account_id":7198,"name":"Jay Bryant","email":"jungleboyj@electronicjungle.net","username":"jsbryant"},"change_message_id":"2ca1063637bd1f356cbf71d23233b28f9e4e6497","unresolved":false,"context_lines":[{"line_number":164,"context_line":""},{"line_number":165,"context_line":"After issued \"cinder migrate\" command, end users can check the"},{"line_number":166,"context_line":"\"migration_status\" by \"cinder show\" command. If \"migration_status\" is"},{"line_number":167,"context_line":"\"migrating\", this volume is probably not safe to attach, and if"},{"line_number":168,"context_line":"\"migration_status\" is \"migrating_attachable\" or \"success\", we can attach"},{"line_number":169,"context_line":"it safely. But one thing we should let end user know is that if"},{"line_number":170,"context_line":"\"migration_status\" is \"migrating_attachable\", the volume is safe to"}],"source_content_type":"text/x-rst","patch_set":23,"id":"7aa08908_a7dfbf0b","line":167,"range":{"start_line":167,"start_character":0,"end_line":167,"end_character":11},"updated":"2016-06-07 23:46:29.000000000","message":"We wouldn\u0027t allow attachment in this state.  Right?","commit_id":"22ab8201d2af78a9210f190f0478f47a72415849"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"de238786725180b436ced176c30264762af30b5e","unresolved":false,"context_lines":[{"line_number":164,"context_line":""},{"line_number":165,"context_line":"After issued \"cinder migrate\" command, end users can check the"},{"line_number":166,"context_line":"\"migration_status\" by \"cinder show\" command. If \"migration_status\" is"},{"line_number":167,"context_line":"\"migrating\", this volume is probably not safe to attach, and if"},{"line_number":168,"context_line":"\"migration_status\" is \"migrating_attachable\" or \"success\", we can attach"},{"line_number":169,"context_line":"it safely. But one thing we should let end user know is that if"},{"line_number":170,"context_line":"\"migration_status\" is \"migrating_attachable\", the volume is safe to"}],"source_content_type":"text/x-rst","patch_set":23,"id":"7aa08908_5d57ce25","line":167,"range":{"start_line":167,"start_character":0,"end_line":167,"end_character":11},"in_reply_to":"7aa08908_a7dfbf0b","updated":"2016-06-08 01:22:15.000000000","message":"For the driver specific migration  between two pools of one single backend, some vendors have already make it safe to attach to a server when it\u0027s \"migrating\". But for the case of host-copy, we certainly won\u0027t allow attachment when it\u0027s \"migrating\" because the volume is really unusable. Since end users can\u0027t distinguish whether the \"migrating\" is executing by driver specific way or host-copy way, so we can say we wouldn\u0027t allow attachment in this state. Thanks!","commit_id":"22ab8201d2af78a9210f190f0478f47a72415849"},{"author":{"_account_id":10628,"name":"Peter Wang","email":"peter.wang13@emc.com","username":"peter_wang"},"change_message_id":"3b4e3866079ecf2bbcdc2be92a2fdf6a5501fea3","unresolved":false,"context_lines":[{"line_number":63,"context_line":"* One array can take over other array\u0027s LUN as a remote LUN if these two"},{"line_number":64,"context_line":"arrays are connected with FC or iSCSI fabric."},{"line_number":65,"context_line":"* We can migrate a remote LUN to a local LUN and meanwhile the remote(source)"},{"line_number":66,"context_line":"LUN is writable and readable with the migration task is undergoing, after the"},{"line_number":67,"context_line":"migration is completed, the local(target) LUN is exactly the same as the"},{"line_number":68,"context_line":"remote(source) LUN and no data will be lost."},{"line_number":69,"context_line":""},{"line_number":70,"context_line":"To enable one array to take over other array\u0027s volume, we should allow one"},{"line_number":71,"context_line":"driver to call other drivers\u0027 interfaces directly or indirectly. These two"}],"source_content_type":"text/x-rst","patch_set":24,"id":"1aa78d24_6b09a231","line":68,"range":{"start_line":66,"start_character":67,"end_line":68,"end_character":44},"updated":"2016-07-04 07:36:04.000000000","message":"Suppose we migrate from lvm backend to vendor backend,\nand vendor driver implemented \"migrate_volume_target\", how can the vendor migrate the data from lvm to vendor storage and make sure no data lost?","commit_id":"5758603751eef6562d6c2bad2fb3befe428d0c5a"},{"author":{"_account_id":6491,"name":"xing-yang","email":"xingyang105@gmail.com","username":"xing-yang"},"change_message_id":"5703676ff55712320f49887192406ae6a8b6ee09","unresolved":false,"context_lines":[{"line_number":63,"context_line":"* One array can take over other array\u0027s LUN as a remote LUN if these two"},{"line_number":64,"context_line":"arrays are connected with FC or iSCSI fabric."},{"line_number":65,"context_line":"* We can migrate a remote LUN to a local LUN and meanwhile the remote(source)"},{"line_number":66,"context_line":"LUN is writable and readable with the migration task is undergoing, after the"},{"line_number":67,"context_line":"migration is completed, the local(target) LUN is exactly the same as the"},{"line_number":68,"context_line":"remote(source) LUN and no data will be lost."},{"line_number":69,"context_line":""},{"line_number":70,"context_line":"To enable one array to take over other array\u0027s volume, we should allow one"},{"line_number":71,"context_line":"driver to call other drivers\u0027 interfaces directly or indirectly. These two"}],"source_content_type":"text/x-rst","patch_set":24,"id":"1aa78d24_64916b47","line":68,"range":{"start_line":66,"start_character":67,"end_line":68,"end_character":44},"in_reply_to":"1aa78d24_1613b5b9","updated":"2016-07-06 03:27:57.000000000","message":"We can use VPLEX to move data between different arrays.  However I\u0027m still not clear how it can fit into this design.","commit_id":"5758603751eef6562d6c2bad2fb3befe428d0c5a"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"3dbe2e8efc0ff99082d679bd587feb3ff5516dbb","unresolved":false,"context_lines":[{"line_number":63,"context_line":"* One array can take over other array\u0027s LUN as a remote LUN if these two"},{"line_number":64,"context_line":"arrays are connected with FC or iSCSI fabric."},{"line_number":65,"context_line":"* We can migrate a remote LUN to a local LUN and meanwhile the remote(source)"},{"line_number":66,"context_line":"LUN is writable and readable with the migration task is undergoing, after the"},{"line_number":67,"context_line":"migration is completed, the local(target) LUN is exactly the same as the"},{"line_number":68,"context_line":"remote(source) LUN and no data will be lost."},{"line_number":69,"context_line":""},{"line_number":70,"context_line":"To enable one array to take over other array\u0027s volume, we should allow one"},{"line_number":71,"context_line":"driver to call other drivers\u0027 interfaces directly or indirectly. These two"}],"source_content_type":"text/x-rst","patch_set":24,"id":"1aa78d24_1613b5b9","line":68,"range":{"start_line":66,"start_character":67,"end_line":68,"end_character":44},"in_reply_to":"1aa78d24_4abac824","updated":"2016-07-05 06:36:54.000000000","message":"There is no generic way, and as I said in this spec, this is a feature many vendors have been supported. We should take advantage of this feature to implement this migration. For EMC, the feature of “Open Replicator hot pull” on VMAX is the right one we need here. Thanks!","commit_id":"5758603751eef6562d6c2bad2fb3befe428d0c5a"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"53f638e70dd60546c5c0abfa83ce93a956429fe1","unresolved":false,"context_lines":[{"line_number":63,"context_line":"* One array can take over other array\u0027s LUN as a remote LUN if these two"},{"line_number":64,"context_line":"arrays are connected with FC or iSCSI fabric."},{"line_number":65,"context_line":"* We can migrate a remote LUN to a local LUN and meanwhile the remote(source)"},{"line_number":66,"context_line":"LUN is writable and readable with the migration task is undergoing, after the"},{"line_number":67,"context_line":"migration is completed, the local(target) LUN is exactly the same as the"},{"line_number":68,"context_line":"remote(source) LUN and no data will be lost."},{"line_number":69,"context_line":""},{"line_number":70,"context_line":"To enable one array to take over other array\u0027s volume, we should allow one"},{"line_number":71,"context_line":"driver to call other drivers\u0027 interfaces directly or indirectly. These two"}],"source_content_type":"text/x-rst","patch_set":24,"id":"1aa78d24_e76e54c8","line":68,"range":{"start_line":66,"start_character":67,"end_line":68,"end_character":44},"in_reply_to":"1aa78d24_6b09a231","updated":"2016-07-04 12:51:28.000000000","message":"Good question! If we migrate from lvm to vendor bachend and vendor driver implemented \"migrate_volume_target\", vendor driver should take over the corresponding lvm volume, start the migration task on vendor backend and return \"async\" as \"True\" and \"moved\" as \"True\". Then volume manager will change the volume[\u0027host\u0027] to the vendor backend. Later IO will be send to the volume on vendor backend, not the lvm. Please see the code implementation for more details: https://review.openstack.org/#/c/334923/.","commit_id":"5758603751eef6562d6c2bad2fb3befe428d0c5a"},{"author":{"_account_id":10628,"name":"Peter Wang","email":"peter.wang13@emc.com","username":"peter_wang"},"change_message_id":"21a9763d18620d05010ee14e4fff212b5778b4d1","unresolved":false,"context_lines":[{"line_number":63,"context_line":"* One array can take over other array\u0027s LUN as a remote LUN if these two"},{"line_number":64,"context_line":"arrays are connected with FC or iSCSI fabric."},{"line_number":65,"context_line":"* We can migrate a remote LUN to a local LUN and meanwhile the remote(source)"},{"line_number":66,"context_line":"LUN is writable and readable with the migration task is undergoing, after the"},{"line_number":67,"context_line":"migration is completed, the local(target) LUN is exactly the same as the"},{"line_number":68,"context_line":"remote(source) LUN and no data will be lost."},{"line_number":69,"context_line":""},{"line_number":70,"context_line":"To enable one array to take over other array\u0027s volume, we should allow one"},{"line_number":71,"context_line":"driver to call other drivers\u0027 interfaces directly or indirectly. These two"}],"source_content_type":"text/x-rst","patch_set":24,"id":"1aa78d24_4abac824","line":68,"range":{"start_line":66,"start_character":67,"end_line":68,"end_character":44},"in_reply_to":"1aa78d24_e76e54c8","updated":"2016-07-04 14:44:34.000000000","message":"I already read your code before I left above comments.\n\nActually, My concern is about how a vendor driver moves the existing data on the volume being migrated to the volume created on vendor backend. Is there a generic way to do this?\n\nThanks\nPeter","commit_id":"5758603751eef6562d6c2bad2fb3befe428d0c5a"},{"author":{"_account_id":11904,"name":"Sean McGinnis","email":"sean.mcginnis@gmail.com","username":"SeanM"},"change_message_id":"58c963f04b74aa2604918995244ebd77829a79c4","unresolved":false,"context_lines":[{"line_number":68,"context_line":"remote(source) LUN and no data will be lost."},{"line_number":69,"context_line":""},{"line_number":70,"context_line":"To enable one array to take over other array\u0027s volume, we should allow one"},{"line_number":71,"context_line":"driver to call other drivers\u0027 interfaces directly or indirectly. These two"},{"line_number":72,"context_line":"drivers are from two independent backends, can be managed by one single volume"},{"line_number":73,"context_line":"node or two different volume nodes."},{"line_number":74,"context_line":""}],"source_content_type":"text/x-rst","patch_set":24,"id":"1aa78d24_3053143b","line":71,"updated":"2016-07-12 01:16:19.000000000","message":"So to do this I am assuming you are planning on passing in a driver instance into the target driver instance so it is able to call methods on the source target. This raises some questions for me. Will the manager take care of calling do_setup and necessary initialization? I also have some concerns about operation locking with two different driver instances for the same backend running.","commit_id":"5758603751eef6562d6c2bad2fb3befe428d0c5a"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"542955749c02baa9fceb3adc2fb2f4fec736cb37","unresolved":false,"context_lines":[{"line_number":68,"context_line":"remote(source) LUN and no data will be lost."},{"line_number":69,"context_line":""},{"line_number":70,"context_line":"To enable one array to take over other array\u0027s volume, we should allow one"},{"line_number":71,"context_line":"driver to call other drivers\u0027 interfaces directly or indirectly. These two"},{"line_number":72,"context_line":"drivers are from two independent backends, can be managed by one single volume"},{"line_number":73,"context_line":"node or two different volume nodes."},{"line_number":74,"context_line":""}],"source_content_type":"text/x-rst","patch_set":24,"id":"1aa78d24_a2468cbd","line":71,"in_reply_to":"1aa78d24_3053143b","updated":"2016-07-12 08:42:36.000000000","message":"The manager needn\u0027t take care of calling do_setup and initialization, because if we try to migrate a volume between two different backends, we assume that these two backends are already running so already been called do_setup and initialized. If not so, even the current host-copy way will fail.\n\nFor the interfaces like initialize_connection() and delete_volume(), it\u0027s the driver who implement these interfaces determine whether a lock is need. Currently in volume manager the caller is not response for the operation locking of driver interface. So I think here we needn\u0027t do the operation locking.","commit_id":"5758603751eef6562d6c2bad2fb3befe428d0c5a"},{"author":{"_account_id":10628,"name":"Peter Wang","email":"peter.wang13@emc.com","username":"peter_wang"},"change_message_id":"3b4e3866079ecf2bbcdc2be92a2fdf6a5501fea3","unresolved":false,"context_lines":[{"line_number":119,"context_line":"   migrate_volume() routine is end now. If migrate_volume_target() returns"},{"line_number":120,"context_line":"   True, go to step 4."},{"line_number":121,"context_line":""},{"line_number":122,"context_line":"#. Call target backend driver\u0027s complete_migration_target() to monitor the"},{"line_number":123,"context_line":"   undergoing migration, and do some cleaup after the migration is totally"},{"line_number":124,"context_line":"   completed."},{"line_number":125,"context_line":""}],"source_content_type":"text/x-rst","patch_set":24,"id":"1aa78d24_0fbaa923","line":122,"updated":"2016-07-04 07:36:04.000000000","message":"Does third party driver need to monitor the migration status in a looping call, or complete_migartion_target itself is a looping call?","commit_id":"5758603751eef6562d6c2bad2fb3befe428d0c5a"},{"author":{"_account_id":13203,"name":"Wilson Liu","email":"295988511@qq.com","username":"Bob-OpenStack"},"change_message_id":"53f638e70dd60546c5c0abfa83ce93a956429fe1","unresolved":false,"context_lines":[{"line_number":119,"context_line":"   migrate_volume() routine is end now. If migrate_volume_target() returns"},{"line_number":120,"context_line":"   True, go to step 4."},{"line_number":121,"context_line":""},{"line_number":122,"context_line":"#. Call target backend driver\u0027s complete_migration_target() to monitor the"},{"line_number":123,"context_line":"   undergoing migration, and do some cleaup after the migration is totally"},{"line_number":124,"context_line":"   completed."},{"line_number":125,"context_line":""}],"source_content_type":"text/x-rst","patch_set":24,"id":"1aa78d24_4718081e","line":122,"in_reply_to":"1aa78d24_0fbaa923","updated":"2016-07-04 12:51:28.000000000","message":"complete_migartion_target itself is not a looping call, vendor drivers should be response to monitor the migration status in a looping call.","commit_id":"5758603751eef6562d6c2bad2fb3befe428d0c5a"}]}
