Release proposal: Nightly v0.6

[BohuTANG]

Summary Please add v0.6 label to your issue if you want to release it in 6 weeks, release name: v0.6-nightly, get in the train now :hand:

Checklist

Stable checklist

• [x] #2561
• [x] HTTP handler API works with all the queries which databend support(DONE)
• [x] #1434

Experimental checklist

• [x] #2162
• [x] #2627
• [x] #2679
• [x] #2868
• [x] #2526
• [x] #2894

Meta Service checklist

• [x] #2667
• [x] #2422

Testing checklist

• [x] #1866

What Databend nightly v0.6 can do?

• You can experience with ontime test suite with fuse table engine(local/s3)

Archive releases

• [x] #2257

Reference

What's Databend release channels? Nightly v0.6 is part of our Roadmap 2021 Community website: https://databend.rs

Thank

Thank you everyone for all the issues and PRs that will be resulted in this release.

[flaneur2020]

do we need include the plan of databend-operator in this checklist? it's in a stand-alone repository

[ZhiHanZ]

do we need include the plan of databend-operator in this checklist? it's in a stand-alone repository

I think so, we should add at least two things:

1. refactor databend-operator for meta service and DAL storage
2. add test infra to test databend-operator on different cloud providers( AWS, GKE etc)

it would be nice to have the following experimental features:

1. monitoring demo(add prometheus to databend_cluster)
2. Autoscaling with keda
3. multi-cluster scheduling with admiralty or karmada etc.

[ZhiHanZ]

Plus, we should also support bendctl integration with operator for basic logging, debugging and visualization supports

[ZhiHanZ]

Besides in CI and e2e test part, we should add the following tests to guarantee our OOS quality.

1. regular stateful test for local and Object storage(mocking)
2. Test kubernetes operator integration part and demo part before release get published during stable channel release.
3. Periodically test the kubernetes operator and helm part reside in different providers and clusters (perhaps once a day)

Not sure about when would us ready to run TPC-H benchmarks on large cluster, but we could prepare it on test infra.

1. Cluster size and estimated cost plan
2. test infra CI scripts to run TPC-H on GKE or graviton EKS
3. TPC-H dataset preparation and query part

[flaneur2020]

adding a system.columns table with a SHOW COLUMNS table syntax: https://github.com/datafuselabs/databend/issues/2608

[flaneur2020]

HTTP API with polling based incremental body response & progress metadata

[flaneur2020]

a new databend-playground with a modern SQL IDE

[drmingdrmer]

Update metasrv tasks

[BohuTANG]

Add Experimental checklist:

• [ ] #2679

[youngsofun]

@BohuTANG

HTTP handler API works with all the queries which databend support(need issue)

It seems no extra work is required to support "all the queries".

So I only need to add tests for each interpreter?

By the way, why insert result is "0 rows affected" now?

[BohuTANG]

@BohuTANG

HTTP handler API works with all the queries which databend support(need issue)

It seems no extra work is required to support "all the queries".

So I only need to add tests for each interpreter?

Ok.

By the way, why insert result is "0 rows affected" now?

I guess there is something wrong with the status(http handler), we could check it from the MySQL client.

[BohuTANG]

Progress report:

• [x] #3181
• [x] #3143

Once these two issues are finshed, we are ready to release v0.6 :rocket:

Note Other unfinished tasks will be rolled to the next version(Nightly v0.7)

[BohuTANG]

This release is ready and will be released as scheduled :rocket: