Cassandra Community, For those of you who don't follow the dev@ list, the Cassandra developers have frozen trunk for 4.0 ( https://lists.apache.org/thread.html/18c76129a4fe6785a51dad7500e04ee13a407a7f7ac5c8f9a3d83c87@%3Cdev.cassandra.apache.org%3E and https://lists.apache.org/thread.html/494c3ced9e83ceeb53fa127e44eec6e2588a01b769896b25867fd59f@%3Cdev.cassandra.apache.org%3E ) - no new features will go into trunk until we're confident enough to cut a beta.
There are currently 303 new bug-fixes and features waiting to be tested. A number of developers (and companies that employ them) have already started testing, but there's a LOT of changes, so I'm asking for your help: if you run Cassandra in production, and you have an opportunity to try out the pre-release-candidate trunk, PLEASE give it a shot. Here are some quick suggestions: - Please DON'T use it on production, mission critical workloads. There are a LOT of changes, some of them may cause data correctness, consistency issues, or availability problems (crashes, etc). We expect to find them before release, but we don't want YOU to find them on projects you care about. Please don't test this in prod unless you really know what you're doing. - Please DO use it with mock / simulated / shadow workloads that match your environment. For example, the project doesn't have access to large scale AWS/GCP/Azure clusters or every operating system or every version of Java - we test as much as we can, but if you care about using the ec2 snitch or plan on running on Windows, please take a few minutes and spin up a cluster. - Please DO try upgrading from the version you're currently running. If it's not easy to do, please open a JIRA. Remember to read NEWS.txt. - Please DO try out java11 if you're able. - Please DO tell us about performance regressions in your workloads - this is especially valuable if you have before/after metrics showing a change in behavior. - Please DO share your problems - especially if you find a problem. Please open a JIRA and add the label "4.0-pre-rc-bugs". We'll have new labels for alpha/beta/RCs. - Please DO communicate on the dev@ list if you see something you suspect may be a bug, but you're not sure. - Please DONT reply-all to this email with any bug you find. - Please DONT use this as an opportunity to ask for new features or enhancements you want to land in 4.0. If you're feeling especially adventurous, try out some of the new features - things like streaming entire sstables for bootstrap ( https://github.com/apache/cassandra/blob/trunk/conf/cassandra.yaml#L799-L809 ) probably speeds up bootstrapping on clusters using LCS by a meaningful amount, and there's a whole new audit log ( https://github.com/apache/cassandra/blame/trunk/conf/cassandra.yaml#L1218-L1227 ) and a new "group commitlog" for folks currently using the especially durable batch commitlog ( https://github.com/apache/cassandra/blame/trunk/conf/cassandra.yaml#L380-L382 ) . There's also a new ability to run multiple instances on different ports using the same IP - if you're running in containers, you probably want to try this out ( https://github.com/apache/cassandra/blame/trunk/conf/cassandra.yaml#L433 ) If you write software that interacts with Cassandra (e.g, you run a metrics collection SaaS company), now would be a great time to think about making sure the APIs you rely on still exist. For some, you may find there are newer, easier access methods - we've added a whole new easily-queryable CQL interface to some of the metrics you likely care about ( see, for example, https://issues.apache.org/jira/browse/CASSANDRA-14523 ). If you have questions, please ask! - Jeff