)]}'
{"/PATCHSET_LEVEL":[{"author":{"_account_id":2271,"name":"Michael Still","email":"mikal@stillhq.com","username":"mikalstill"},"change_message_id":"083b891c6138c820bd01f997838577a6a2c14140","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":1,"id":"294e4124_c799f2b0","updated":"2026-10-24 03:24:24.000000000","message":"A question for neutron-lib reviewers before this goes further. The fix for LP #1802369 made neutron_lib/__init__.py import neutron_lib.db.api unconditionally, so that the ORM event listeners  exist before any model class is defined. This change keeps that guarantee for anything that uses \nneutron_lib.db.model_base, because model_base now imports the DB API and registers the BASEV2 listener itself. Everything else gets the DB API on first use.\n\nThe risk is out-of-tree code that defines or uses models without importing model_base first. For that  code, the listeners would no longer be there. Is lazy registration acceptable, or do you know of consumers that rely on the import side effect?\n\nFor scale: with the three neutron changes that move the agents\u0027 own DB API imports, this saves 4 to 6 MB in each of the OVN, DHCP, L3 and metadata agents. SQLAlchemy and oslo.db still load through neutron_lib.context, so this isn\u0027t the whole DB stack.","commit_id":"5c0eca659a4a3feffae555f738eb2426b8eb4813"},{"author":{"_account_id":5890,"name":"Doug Goldstein","email":"cardoe@cardoe.com","username":"cardoe"},"change_message_id":"a32a28bbf346510b9d167862aab7528edb2d9ecb","unresolved":true,"context_lines":[],"source_content_type":"","patch_set":2,"id":"c5d72429_afcf9e9a","updated":"2026-10-27 20:49:22.000000000","message":"So this really doesn\u0027t depend on the devstack changes to monitor performance. But this is a good change never the less. So can we remove that depend and land this?","commit_id":"ab63a1a3c1a24218fb881b96435df9ecdec0e1fe"},{"author":{"_account_id":16688,"name":"Rodolfo Alonso","email":"ralonsoh@redhat.com","username":"rodolfo-alonso-hernandez"},"change_message_id":"1d38eb4af5169608dea4ef791a0d46c4411a49cf","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":2,"id":"ce980218_b19ec0a6","updated":"2026-10-27 10:49:06.000000000","message":"Waiting for experimental queue, but this patch looks OK","commit_id":"ab63a1a3c1a24218fb881b96435df9ecdec0e1fe"},{"author":{"_account_id":16688,"name":"Rodolfo Alonso","email":"ralonsoh@redhat.com","username":"rodolfo-alonso-hernandez"},"change_message_id":"cd5f82906148979f30c60dae00d9338c6441b8dc","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":2,"id":"ef4d035f_764c3ae5","updated":"2026-10-27 10:41:55.000000000","message":"check experimental","commit_id":"ab63a1a3c1a24218fb881b96435df9ecdec0e1fe"},{"author":{"_account_id":2271,"name":"Michael Still","email":"mikal@stillhq.com","username":"mikalstill"},"change_message_id":"a08fc064307187e2c50a333c2c9d31979b10150b","unresolved":true,"context_lines":[],"source_content_type":"","patch_set":2,"id":"001bb5f6_d57b0671","in_reply_to":"c5d72429_afcf9e9a","updated":"2026-10-28 04:29:33.000000000","message":"I have no desire to fight about it, but the depends-on is there because otherwise we can\u0027t measure the impact of this change (and its little friends). So its not the hill I am going to die on and I\u0027ll remove it if you really really insist, but wouldn\u0027t it just be better to land the devstack change so we can measure these things and then get on with our lives?","commit_id":"ab63a1a3c1a24218fb881b96435df9ecdec0e1fe"}]}