)]}'
{"/PATCHSET_LEVEL":[{"author":{"_account_id":38368,"name":"Christian Ohanaja","display_name":"Christian Ohanaja","email":"cohanaja@nvidia.com","username":"cohanaja"},"change_message_id":"19d1c4a96569936fc4504da5e393423a8f73f0bd","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":3,"id":"8ac566fc_b629ccda","updated":"2026-09-29 21:18:48.000000000","message":"I want to split off the interesting merge_items_not_in bits from the improvement patch, understand them better, make sure they don\u0027t suck, and clean them up for actual review in a separate patch. I think it\u0027ll help a really good bit in prod actually, but I want to first improve the logging we have around the replicator so that we know when we hit the rsync branch of choose_replication_mode.\n\nThis uses the force_replication_mode patch from a long while ago as a base, mostly because it\u0027s an easy way to test rsync only. Ideally once we can iron out the kinks involved with rsync we can do something really cool and swap out usync entirely maybe theoretically hopefully in an ideal world","commit_id":"cdabd059b00788f5ce7e44d1dc9a30c69b2de874"},{"author":{"_account_id":38368,"name":"Christian Ohanaja","display_name":"Christian Ohanaja","email":"cohanaja@nvidia.com","username":"cohanaja"},"change_message_id":"a1365e3e79eed11d00613421ddcac06465ba55e2","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":3,"id":"c3457a65_ff5d9458","updated":"2026-09-29 21:20:32.000000000","message":"There\u0027s also this prior sync_points patch I submitted but didn\u0027t really push on,\n\nhttps://review.opendev.org/c/openstack/swift/+/1006491\n\nI think the conditions where that one leads to an improvement in a sensible way are rarer than what we\u0027d get from this patch. They can be used in tandem though, and I\u0027m going to run some additional benchmarks with both this + that one and see if that ekes out even better numbers","commit_id":"cdabd059b00788f5ce7e44d1dc9a30c69b2de874"}]}