Actually just yesterday @Siyao Meng <sm...@cloudera.com> discovered a new rocksdb crash issue. As far as I am aware it is a bug in Ozone not rocksdb.
@Siyao Meng <sm...@cloudera.com> or @Ritesh Shukla <rit...@cloudera.com> would you like to share more details? On Tue, Nov 1, 2022 at 1:22 AM Hui Fei <feihui.u...@gmail.com> wrote: > Thanks for sharing, Wei-Chiu! > > Wei-Chiu Chuang <weic...@apache.org> 于2022年11月1日周二 12:32写道： > > > > > > > Thanks all contributors. > > > Does is solve datanode crashing and ozone manager memory leak issues? > > could > > > anyone share experiences? > > > > > > > We believe we're addressed most if not all RocksDB memory leaks in OM, > SCM, > > DataNode and client. > > > > For what is worth, our internal branch is still using RocksDB 6 and thus > we > > can't verify that RocksDB 7 resolves the SIGSEGV crash problem. > > For now, increasing number of container cache seems to workaround the > > problem. > > >
