)]}'
{"specs/kilo/create-volume-persistence-taskflow-support-volume.rst":[{"author":{"_account_id":1297,"name":"Joshua Harlow","email":"jxharlow@godaddy.com","username":"harlowja"},"change_message_id":"4f67e7f4d5b633388d64fecbb9560a2f757b2480","unresolved":false,"context_lines":[{"line_number":27,"context_line":"  possibility of deleting. Use of reset-state or manual DB cleanup is required"},{"line_number":28,"context_line":"  to delete the volume."},{"line_number":29,"context_line":"- CreateVolumeFromSpecTask isn\u0027t idempotent (call to driver will fail when"},{"line_number":30,"context_line":"  started second time), so it will crash when restarted."},{"line_number":31,"context_line":"- Context object passed between tasks isn\u0027t serialized, so workflow saving"},{"line_number":32,"context_line":"  fails."},{"line_number":33,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"1a930d6b_45bfeed0","line":30,"updated":"2015-01-20 23:36:49.000000000","message":"This enters the area/boundary of drivers and having the driver be idepoment; I\u0027m not sure if most are, something to think about I guess... If we can have the thing that calls the driver be idepoment that will be a good start (and later deal with the drivers that aren\u0027t idepoment).","commit_id":"bdc7a43c3f515ab45981610acbba6fdd7ffde921"},{"author":{"_account_id":1297,"name":"Joshua Harlow","email":"jxharlow@godaddy.com","username":"harlowja"},"change_message_id":"4f67e7f4d5b633388d64fecbb9560a2f757b2480","unresolved":false,"context_lines":[{"line_number":44,"context_line":"idempotent. Easiest idea here is to check in the task if the volume was already"},{"line_number":45,"context_line":"updated in the DB and when this is true do not call the driver. This however"},{"line_number":46,"context_line":"won\u0027t address a case when service dies when driver is creating the volume. In"},{"line_number":47,"context_line":"the future it will be needed to check if volume exist on driver level."},{"line_number":48,"context_line":""},{"line_number":49,"context_line":"Context object will be serialized using to_dict() method before passing into"},{"line_number":50,"context_line":"workflow and unserialized (from_dict) in task\u0027s execute method as needed."}],"source_content_type":"text/x-rst","patch_set":1,"id":"1a930d6b_a5cdf288","line":47,"updated":"2015-01-20 23:36:49.000000000","message":"Agreed, this will likely be the way forward until drivers can do a better job.","commit_id":"bdc7a43c3f515ab45981610acbba6fdd7ffde921"},{"author":{"_account_id":170,"name":"Mike Perez","email":"thingee@gmail.com","username":"thingee"},"change_message_id":"57240b8be8d1a05e2294fff4bcd93c96efbc9476","unresolved":false,"context_lines":[{"line_number":22,"context_line":""},{"line_number":23,"context_line":"Problems in Volume workflow:"},{"line_number":24,"context_line":""},{"line_number":25,"context_line":"- When cinder-volume dies inside the flow before or at the beggining of"},{"line_number":26,"context_line":"  CreateVolumeFromSpecTask, volume is stuck in creating state forever without"},{"line_number":27,"context_line":"  possibility of deleting. Use of reset-state or manual DB cleanup is required"},{"line_number":28,"context_line":"  to delete the volume."}],"source_content_type":"text/x-rst","patch_set":2,"id":"da86d52c_32265bbb","line":25,"updated":"2015-02-10 06:21:34.000000000","message":"/beggining/beginning/","commit_id":"5b8474073781404682ac9d147aed1314b9b6f9fc"},{"author":{"_account_id":170,"name":"Mike Perez","email":"thingee@gmail.com","username":"thingee"},"change_message_id":"57240b8be8d1a05e2294fff4bcd93c96efbc9476","unresolved":false,"context_lines":[{"line_number":42,"context_line":""},{"line_number":43,"context_line":"To make the flow resumable CreateVolumeFromSpecTask will need to be made more"},{"line_number":44,"context_line":"idempotent. Easiest idea here is to check in the task if the volume was already"},{"line_number":45,"context_line":"updated in the DB and when this is true do not call the driver. This however"},{"line_number":46,"context_line":"won\u0027t address a case when service dies when driver is creating the volume. In"},{"line_number":47,"context_line":"the future it will be needed to check if volume exist on driver level."},{"line_number":48,"context_line":""}],"source_content_type":"text/x-rst","patch_set":2,"id":"da86d52c_52290f88","line":45,"updated":"2015-02-10 06:21:34.000000000","message":"I don\u0027t know of a certain way you can determine to not call the driver purely on the volume entry being updated. What are you look at in the entry to determine that?","commit_id":"5b8474073781404682ac9d147aed1314b9b6f9fc"},{"author":{"_account_id":170,"name":"Mike Perez","email":"thingee@gmail.com","username":"thingee"},"change_message_id":"57240b8be8d1a05e2294fff4bcd93c96efbc9476","unresolved":false,"context_lines":[{"line_number":44,"context_line":"idempotent. Easiest idea here is to check in the task if the volume was already"},{"line_number":45,"context_line":"updated in the DB and when this is true do not call the driver. This however"},{"line_number":46,"context_line":"won\u0027t address a case when service dies when driver is creating the volume. In"},{"line_number":47,"context_line":"the future it will be needed to check if volume exist on driver level."},{"line_number":48,"context_line":""},{"line_number":49,"context_line":"Context object will be serialized using to_dict() method before passing into"},{"line_number":50,"context_line":"workflow and unserialized (from_dict) in task\u0027s execute method as needed."}],"source_content_type":"text/x-rst","patch_set":2,"id":"da86d52c_7220d39b","line":47,"updated":"2015-02-10 06:21:34.000000000","message":"I like this idea of asking the driver for this information. It will also require a bit of communication and coordinating with all the driver maintainers. IMO though, without this piece, this is useless to implement in cinder-volume.","commit_id":"5b8474073781404682ac9d147aed1314b9b6f9fc"}]}
