Brian Zhou

Read latency can sometimes go up to 1 second

6

**Describe the bug** When we start the local Pravega standalone, and perform `readNextEvent()` after `writeEvent()`, very rarely(under 0.1% in our test) it cannot return the event inside the 1000-millisecond timeout...

Extra warning log when call getReaderGroup that not yet created

8

**Describe the bug** Pravega used to persist the idempotence of the call `createReaderGroup`, so flink connector used to have those part of code like this in the checkpoint hook. ```...

**Problem description** To avoid the log4j-core jars with vulnerability [CVE-2021-44228](https://nvd.nist.gov/vuln/detail/CVE-2021-44228) being downloaded during the compilation of the project, we need to update the Flink dependency version. We need to update...

**Problem description** It could be better if we have a central place to store all the client configurations, but now we are separating them in each of the source/sink builder....

Partial recovery in FLIP-27 source

7

**Problem description** In the new FLIP-27 source, we still only support a global recovery that relies on the Flink's externally induced source reader API(added in [FLINK-20270](https://issues.apache.org/jira/browse/FLINK-20270)) + Pravega checkpoint. This...

**Problem description** Pravega has an open source implementation of the Keycloak client authentication, see https://github.com/pravega/pravega-keycloak. It's better to have a document to show how we set the configurations to talk...

**Problem description** In [FLIP-167](https://cwiki.apache.org/confluence/display/FLINK/FLIP-167%3A+Watermarks+for+Sink+API), a new watermark API is introduced in both the new and old Sink API We can - Change our current watermark implementation in `SinkFunction` implementation. We...

**Problem description** Investigate the possibility of the upsert table support using Pravega's Key value table feature. **Problem location** Table API **Suggestions for an improvement**

**Problem description** Create and publish a separate jar for SQL usage, name as `pravega-connectors-flink-sql-1.12_2.12` ref: https://issues.apache.org/jira/browse/FLINK-11026 **Problem location** Build, github actions, and document **Suggestions for an improvement**

**Problem description** We use a two-phase commit algorithm with Flink checkpoint and Pravega transactional writer to implement the end to end exactly-once feature, see #5 in detail. In the second...