fix(reserve): evict just enough chunks to reach the capacity #4549
Conversation
istae
changed the title
| @@ -335,6 +360,14 @@ func (r *Reserve) EvictBatchBin( | |||
| return 0, err | |||
| } | |||
|
|
|||
| // evict oldest chunks first | |||
| sort.Slice(evicted, func(i, j int) bool { | |||
| return evicted[i].BinID < evicted[j].BinID | |||
There was a problem hiding this comment.
Color me not fully grokking this, but how is sorting by BinID getting us the "oldest chunks first" as the comment states?
There was a problem hiding this comment.
binID is an unique, incremented value that is assigned to each chunk in the reserve. So by removing the chunks with the lowest binID, we can at least bring some order (and not just randomly) as to which chunks should be removed first.
There was a problem hiding this comment.
So binID has nothing to do with the pullsync bins, nor anything else within bee that is called bin? And since it is unique, I would gather that each "bin" contains but a single chunk? Bizarre.
One of my personal software development mantras is "Naming can make or break a project".
There was a problem hiding this comment.
the reserve has "bins", which correspond to the proximity orders of the chunks (bin 0 = po 0, etc). so a "binID" is the unique ID of a chunk in the bin.
pullsync gets chunks from a peer using the peer's binIDs (so called bin intervals) for each bin/po.
istae
force-pushed
the
localstore-cleanup
branch
from
32ec942 to
657f7ab
Compare
Checklist
closes #4538
Description
Open API Spec Version Changes (if applicable)
Motivation and Context (Optional)
Related Issue (Optional)
Screenshots (if appropriate):