[
https://issues.jboss.org/browse/ISPN-10423?page=com.atlassian.jira.plugin...
]
Tristan Tarrant updated ISPN-10423:
-----------------------------------
Sprint: DataGrid Sprint #31, DataGrid Sprint #32, DataGrid Sprint #33, DataGrid Sprint
#34, DataGrid Sprint #35, DataGrid Sprint #36, DataGrid Sprint #37 (was: DataGrid Sprint
#31, DataGrid Sprint #32, DataGrid Sprint #33, DataGrid Sprint #34, DataGrid Sprint #35,
DataGrid Sprint #36)
RocksDB ReadOptions memory leak
-------------------------------
Key: ISPN-10423
URL:
https://issues.jboss.org/browse/ISPN-10423
Project: Infinispan
Issue Type: Bug
Components: Loaders and Stores
Affects Versions: 10.0.0.Beta4, 9.4.15.Final
Reporter: Ryan Emerson
Assignee: Ryan Emerson
Priority: Major
Fix For: 10.0.0.Beta5, 9.4.16.Final
Currently our rocksdb store impl creates a new ReadOptions instance per iterator created.
As per the RocksDB we should be calling {{dispose()}} on this once it goes out of scope,
so that the c++ underlying impl can free the memory. Currently we're not doing this,
resulting in a memory leak.
As we only ever use the same ReadOptions for all iterators, we should simply re-use a
singleton for all iterators and then dispose of this on shutdown.
--
This message was sent by Atlassian Jira
(v7.13.8#713008)