Add cuckoo-based drop cache #567
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
MikeGoldsmith
added
type: enhancement
collect/cache/cuckoo.go
Outdated
Comment on lines
55
to
63
| if c.firstTime && dropFactor > 0.5 { | ||
| c.mut.Lock() | ||
| defer c.mut.Unlock() | ||
| c.firstTime = false | ||
| } | ||
| // if the current one is full, cycle the filters | ||
| if dropFactor > 0.99 { | ||
| c.mut.Lock() | ||
| defer c.mut.Unlock() |
There was a problem hiding this comment.
Could this result in a deadlock? We could acquire a lock on line 56, defer unlock on line 57 and then try to acquire another lock on line 62. Will the lock be released as part of the if statement scope closing?
There was a problem hiding this comment.
Whoa -- good catch. The lock will NOT be released and this can cause a deadlock. Will fix!
There was a problem hiding this comment.
Fixed in latest update, please re-review.
ghost
pushed a commit
to opsramp/tracing-proxy
that referenced
this pull request
Jul 5, 2024
## Which problem is this PR solving? Before this change, Refinery used a circular LRU cache to retain a record for every trace; this cache is hardcoded to 5x the configured cache size, and does not change when the cache is resized. This is a relatively small number, and for long-lived traces, it might mean that late spans look like new traces, and therefore might get a different sampling decision -- which would result in missing spans in Honeycomb. honeycombio#561 abstracted the SampleCache interface to prepare for other implementations. This uses it to provide a new cache design. ## Short description of the changes This design implements a "cuckoo filter" cache for dropped traces, which can store the dropped trace information much more efficiently (about 4 bytes per trace as compared to about 200 bytes for kept traces). - Adds a CuckooTraceChecker type that implements a 2-stage cuckoo filter for tracking recently-used trace IDs over time. - Implements the SampleCache interface with a CuckooSentCache, which uses the existing LRU for kept traces, and a CuckooTraceChecker for dropped traces. - Implements a new configuration block for caches to allow users to opt into the cuckoo cache and control it for their needs, but is still backwards compatible. - Adds documentation to the config_complete file. - Adds additional metrics for tracking the cuckoo cache size
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Which problem is this PR solving?
Currently, Refinery uses a circular LRU cache to retain a record for every trace; this cache is hardcoded to 5x the configured cache size, and does not change when the cache is resized.
This is a relatively small number, and for long-lived traces, it might mean that late spans look like new traces, and therefore might get a different sampling decision -- which would result in missing spans in Honeycomb.
#561 abstracted the SampleCache interface to prepare for other implementations. This uses it to provide a new cache design.
Short description of the changes
This design implements a "cuckoo filter" cache for dropped traces, which can store the dropped trace information much more efficiently (about 4 bytes per trace as compared to about 200 bytes for kept traces).