-
Notifications
You must be signed in to change notification settings - Fork 110
Garbage collection does not seem to reduce LevelDB datadir folder size #1113
Comments
So, a bit of research:
This should essentially call the compact on the entire range of the This is the output of a monitor loop that checks the datadir size every 30 seconds:
with compaction:
At certain points, major deltas were noticed in the compaction branch data store size (from A bug where the number wrong metric was sent to InfluxDB has been submitted to upstream (grafana dashboards were showing no chunks evicted upon GC): ethereum/go-ethereum#19102 In the case that the compaction call is indeed correct - the results are inconclusive so far and I am not sure about the benefits of the actual compaction call upon GCing. I will report more with correct GC metrics merged |
closing this as there's no apparent fix at the moment and localstore size does not leak but saturates at a certain level. let's revisit this topic later on |
If you start a node with store size of 500, datadir should be about 2MB.
However after multiple uploads and garbage collection runs, I can see that actually datadir gets to 30MB and beyond.
We need to investigate why this is happening and make sure that
--store.size
param, which is used to limit the disk space used by LevelDB, actually works.The text was updated successfully, but these errors were encountered: