)]}'
{"specs/xena/snapshot-attached-volumes.rst":[{"author":{"_account_id":4523,"name":"Eric Harney","email":"eharney@redhat.com","username":"eharney"},"change_message_id":"4a9d0094fd577a84a2533d6495ba41644b3461c3","unresolved":true,"context_lines":[{"line_number":57,"context_line":"REST API impact"},{"line_number":58,"context_line":"---------------"},{"line_number":59,"context_line":""},{"line_number":60,"context_line":"snapshot create will succeed for in-use volumes w/o force flag added"},{"line_number":61,"context_line":""},{"line_number":62,"context_line":"Security impact"},{"line_number":63,"context_line":"---------------"}],"source_content_type":"text/x-rst","patch_set":1,"id":"12c8f248_a6c0798a","line":60,"updated":"2021-06-10 18:40:10.000000000","message":"Since Gorka raised this in the code review -- this should define what happens if someone passes force\u003dFalse to the new API microversion which allows snapshotting in-use volumes by default.\n\nThere are two options:\n1) The force flag has no effect and is ignored, since under the new model there isn\u0027t anything that requires \"force\" when snapshotting an in-use volume\n\n2) We assume that \"force\" was basically short for \"snapshot in-use volumes\" before, and now interpret \"force\u003dFalse\" to mean \"I purposefully _don\u0027t_ want to snapshot in-use volumes\".  This means that someone passing \"force\u003dFalse\" doesn\u0027t get the new behavior of snapshotting volumes that are in-use.\n\nI would prefer (1) since it\u0027s straightforward and seems to make sense.","commit_id":"6da287ef83f0012d5621c1ce6d3c58ce057f51ab"},{"author":{"_account_id":4523,"name":"Eric Harney","email":"eharney@redhat.com","username":"eharney"},"change_message_id":"1d0829d5355a68fd5f71869eb48e7a13d6caac43","unresolved":true,"context_lines":[{"line_number":57,"context_line":"REST API impact"},{"line_number":58,"context_line":"---------------"},{"line_number":59,"context_line":""},{"line_number":60,"context_line":"snapshot create will succeed for in-use volumes w/o force flag added"},{"line_number":61,"context_line":""},{"line_number":62,"context_line":"Security impact"},{"line_number":63,"context_line":"---------------"}],"source_content_type":"text/x-rst","patch_set":1,"id":"f7e899b2_54c5e4db","line":60,"in_reply_to":"12c8f248_a6c0798a","updated":"2021-06-10 20:14:13.000000000","message":"There\u0027s also option 3: don\u0027t allow force\u003dFalse at all on the new microversions, which is reasonable since surely few people ever pass that.","commit_id":"6da287ef83f0012d5621c1ce6d3c58ce057f51ab"},{"author":{"_account_id":5314,"name":"Brian Rosmaita","email":"rosmaita.fossdev@gmail.com","username":"brian-rosmaita"},"change_message_id":"af7e88cb64f7e74e9969e891b791078a5a95bddc","unresolved":true,"context_lines":[{"line_number":57,"context_line":"REST API impact"},{"line_number":58,"context_line":"---------------"},{"line_number":59,"context_line":""},{"line_number":60,"context_line":"snapshot create will succeed for in-use volumes w/o force flag added"},{"line_number":61,"context_line":""},{"line_number":62,"context_line":"Security impact"},{"line_number":63,"context_line":"---------------"}],"source_content_type":"text/x-rst","patch_set":1,"id":"e6f6bff1_407c0f4b","line":60,"in_reply_to":"f7e899b2_54c5e4db","updated":"2021-06-19 01:20:53.000000000","message":"I think what we should do is:\n\n- the new model is that there is no such thing as \"forcing\" a snapshot.  If you ask for a snapshot, we\u0027ll give you one (and if crash-consistency isn\u0027t good enough for you, you should take appropriate action with your volume before creating a snapshot)\n\n- we need to accept \"force\u003dtrue\" in the query string because people will want to run scripts in multiple clouds, some of which will have too low a max microversion to have the new behavior available\n\n- if someone passes \"force\u003dfalse\" to the new mv, I think we should reject the request immediately at the API layer (i.e., don\u0027t even check to see if the volume is attached), because the new model is that there is no such thing as a forced snapshot.  This will have the side effect of satisfying the use case of passing \"force\u003dfalse\" on the old model when you didn\u0027t want to snap a volume if it was currently attached.  (It will fail all the time, but the key thing is that someone passing \u0027force\u003dfalse\u0027 will never get the surprise of having a successful snapshot creation, but wind up with a (merely) crash-consistent snapshot, which isn\u0027t what they wanted.)","commit_id":"6da287ef83f0012d5621c1ce6d3c58ce057f51ab"},{"author":{"_account_id":5314,"name":"Brian Rosmaita","email":"rosmaita.fossdev@gmail.com","username":"brian-rosmaita"},"change_message_id":"2093352a63accbc6bb4029360bd6b61cffac8410","unresolved":true,"context_lines":[{"line_number":120,"context_line":"Documentation Impact"},{"line_number":121,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":122,"context_line":""},{"line_number":123,"context_line":"* Minimal"},{"line_number":124,"context_line":""},{"line_number":125,"context_line":""},{"line_number":126,"context_line":"References"}],"source_content_type":"text/x-rst","patch_set":1,"id":"fcf47621_8f294071","line":123,"updated":"2021-06-22 12:57:56.000000000","message":"You don\u0027t need to update unless you need to push a new patch set, but it would be good to have a memo here that (as agreed at the Xena PTG) documentation about snapshots being crash-consistent should be added to the api-ref where end users will see it.","commit_id":"6da287ef83f0012d5621c1ce6d3c58ce057f51ab"}]}