4 Commits

Author	SHA1	Message	Date
Jarek Kowalski	b2b5e6fb97	fix(repository): fixed 'unable to get content data and info' (#1844) ... We were incorrectly assigning the same content cache key regardless of the storage format of the content (compression format, etc.) This means that if a content has both compressed and non-compressed entry in the index, it would sometimes get cached incorrectly and the subsequent reads would fail. Clearing the cache would fix the issue - this change appends format-specific suffix to cache keys, so that clearing the cache is not needed. Huge thanks to Mark Derricutt who helped get to the bottom of this on Slack. Fix #1843	2022-03-22 20:39:36 -07:00
Jarek Kowalski	6ba2f586ad	feat(repository): ensure we don't run parallel fetches for the same blob IDs (#1835) ... * feat(repository): ensure we don't run parallel fetches for the same blob IDs Fixed a bunch of test flakes. * fixed race condition, limit the size of mutex cache by using LRU	2022-03-18 21:27:06 -07:00
Jarek Kowalski	991c08e4b4	chore(repository): switched from opencensus to directly exporting prometheus metrics (#1831)	2022-03-17 23:39:36 -07:00
Jarek Kowalski	c4ab086d5f	chore(repository): streamlined internal content cache API (#1828) ... moved from ./repo/content to ./internal/cache	2022-03-15 23:57:04 -07:00