Yan Soares Couto d8e04c872e Fix sharding logic for time shards ... Summary: I was rolling out edenapi uploads and realised they jumped up quicker than I expected. I went to take a look, and it turns out a lot of our migrations are happening incorrectly. Not that much, though. When generating the config, we generate a shard (a 0-99 number), and use that number for all sharding operations: percentage rollouts, timed rollouts, and group decision. The problem is it means those three migrations are always correlated. This diff fixes that by using independent shard numbers for percentage migrations and time migrations. Reviewed By: DurhamG Differential Revision: D32434107 fbshipit-source-id: e5a44969aab421b9c430999c633d58f5ba19191a		2021-11-15 11:33:08 -08:00
..
fs	don't return eden-related error from low-level unix socket wrapper	2021-11-15 04:14:37 -08:00
hg-server	fix bug in long commands logging where FFI boundary expects int and gets float	2021-11-15 08:11:21 -08:00
integration	prjfs: do not bail on read-only callbacks	2021-11-10 14:56:09 -08:00
locale
mononoke	mononoke: add sqlblob_gc integration test	2021-11-15 09:42:31 -08:00
scm	Fix sharding logic for time shards	2021-11-15 11:33:08 -08:00
test_support	test_support: canonicalize the temporary directory path	2021-08-16 16:08:45 -07:00
test-data	fix fsck snapshot integration tests	2021-07-14 16:20:04 -07:00
.gitignore
Eden.project.toml