HBASE-28436 has been resolved. HBASE-28425 is ready for reviewing, where we allow specifying the peer cluster with connection uri when adding a replication peer, PTAL if you have interest.
But there is still a missing part in the plan, the map reduce jobs, where we still only support the zk based cluster key if we want to write to another HBase cluster... Please see HBASE-28565 for more details. Thanks. Duo Zhang <zhang...@apache.org> 于2024年3月31日周日 21:37写道： > > 3.0.0-beta-2 will be the last beta release for 3.0.0 release line, so > it should include all the features we want to include in 3.0.0. > > In 3.0.0, we want to remove all the persistent data on zookeeper, and > also remove zookeeper from our public API as much as possible. For > example, we have implemented table base replication queue storage, to > replace the zk based replication queue storage. > > But recently we found out that, we still only accept zk based cluster > key when adding a replication peer, so we filed HBASE-28425[1] to > address it. And then we found out that, we need to introduce something > like a connection URI/URL for a hbase cluster, as Nick's email in the > dev list[2]. So we filed HBASE-28436[3] and also a PR[4] for it. > > This is an very important/critical change for HBase, especially if we > ship this feature out, it could never be changed in the future, or at > least, very hard to change it without breaking our users. So if you > are interested, please comment on the jira issue, github PR, or just > reply here, about your suggestions, especially how we should we define > the scheme part of the URI/URL. > > Notice that, Phoenix already shipped a similiar feature out[5], which > could be an example. > > Suggestions are welcomed. > > Thanks. > > 1. https://issues.apache.org/jira/browse/HBASE-28425 > 2. https://lists.apache.org/thread/98wqlkqvlnmpx3r7yrg9mw4pqz9ppofh > 3. https://issues.apache.org/jira/browse/HBASE-28436 > 4. https://github.com/apache/hbase/pull/5770 > 5. https://phoenix.apache.org/classpath_and_url.html
