Attenders: Peter, Chia-Chuan, Pony, Hao, Minyu, LanSong, Jianghua, Wei-Chiu, Sammi
- Minyu reported a dead loop issue in SCM delete transaction iterating. Minyu will file a JIRA later for this issue. - About HDDS-12322 <https://issues.apache.org/jira/browse/HDDS-12322> "EC read and reconstruct fail when datanode in maintenance", after further investigation, it's an expected behavior that datanode in maintenance mode will be treated as valid data source before the datanode state is changed to DEAD. Minyu will add more details to the JIRA. - Hadoop community has started the discussion of 3.4.2 release. 3.4.2 will include the patch for the DomainSocket library improvement required for Ozone short-circuit read feature. - Xi proposed a new solution for Ozone client fat jar third party dependency shade management. Currently Ozone shades explicitly included packages for third party package. Xi proposed that we can explicitly exclude some packages, and shade the rest, to avoid update the shade instruction every time there is no dependency introduced. - Wei-Chiu shared the status of OM RocksDB compaction investigation, to reduce the OM operation latency caused by massive delete tombstones in RocksDB. - Peter would like to contribute to the SCM safemode improvement, HDDS-11693 <https://issues.apache.org/jira/browse/HDDS-11693>. - Xi shared the idea of "Job worker"( https://docs.google.com/document/d/1EFJI6l76sIwwK7_f5Hs_tgt3mELXTlP0XJWpJKUkEy4/edit?tab=t.0#heading=h.byww9e6uslm9) which will act as a independent service, do things such as, initiate the data replicate factor covering, data storage class converting, etc. - Ivan proposed this HDDS-12307 <https://issues.apache.org/jira/browse/HDDS-12307> "Realtime Cross-Region Bucket Replication" discussion.
