simplify multi-topic stream

update comments

…ta-tolerant

Don't override HighWaterMarks on bad metadata

apache#27)

In large clusters with large metadata size the UpdateMetadataRequest::toString can generate
really large strings. If there are n/w issue these strings are logged resulting in lots of these
string to be generated and cause high memory usage.

apache#27)

In large clusters with large metadata size the UpdateMetadataRequest::toString can generate
really large strings. If there are n/w issue these strings are logged resulting in lots of these
string to be generated and cause high memory usage.

…sult (apache#27)

TICKET =
LI_DESCRIPTION = 
In large clusters with large metadata size the UpdateMetadataRequest::toString can generate
really large strings. If there are n/w issue these strings are logged resulting in lots of these
string to be generated and cause high memory usage.

EXIT_CRITERIA = MANUAL ["Occurs on large clusters only hence limiting this to LinkedIn branch only"]

…sult (apache#27)

TICKET =
LI_DESCRIPTION = 
In large clusters with large metadata size the UpdateMetadataRequest::toString can generate
really large strings. If there are n/w issue these strings are logged resulting in lots of these
string to be generated and cause high memory usage.

EXIT_CRITERIA = MANUAL ["Occurs on large clusters only hence limiting this to LinkedIn branch only"]

…sult (apache#27)

TICKET =
LI_DESCRIPTION =
In large clusters with large metadata size the UpdateMetadataRequest::toString can generate
really large strings. If there are n/w issue these strings are logged resulting in lots of these
string to be generated and cause high memory usage.

EXIT_CRITERIA = MANUAL ["Occurs on large clusters only hence limiting this to LinkedIn branch only"]

…sult (apache#27)

TICKET =
LI_DESCRIPTION =
In large clusters with large metadata size the UpdateMetadataRequest::toString can generate
really large strings. If there are n/w issue these strings are logged resulting in lots of these
string to be generated and cause high memory usage.

EXIT_CRITERIA = MANUAL ["Occurs on large clusters only hence limiting this to LinkedIn branch only"]

…sult (apache#27)

TICKET =
LI_DESCRIPTION =
In large clusters with large metadata size the UpdateMetadataRequest::toString can generate
really large strings. If there are n/w issue these strings are logged resulting in lots of these
string to be generated and cause high memory usage.

EXIT_CRITERIA = MANUAL ["Occurs on large clusters only hence limiting this to LinkedIn branch only"]

…sult (apache#27)

TICKET =
LI_DESCRIPTION =
In large clusters with large metadata size the UpdateMetadataRequest::toString can generate
really large strings. If there are n/w issue these strings are logged resulting in lots of these
string to be generated and cause high memory usage.

EXIT_CRITERIA = MANUAL ["Occurs on large clusters only hence limiting this to LinkedIn branch only"]

CSMDS-200: ART trigger on PR (apache#27)

CSMDS-141: CSM CI build for kafka (apache#19)

- Add new admin api findCoordinator
- When create mirror topic it should find the coordinator first then send the create topic request

```
./bin/kafka-topics.sh --createMirror --topic test --bootstrap-server localhost:9094 --remote-bootstrap-server localhost:9092 --topic-id  RI3uACW1QM2Exahq-CATrA --link my-link
The findCoordinator response is: FindCoordinatorResponseData(throttleTimeMs=0, errorCode=0, errorMessage='', nodeId=0, host='', port=0, coordinators=[Coordinator(key='my-link', nodeId=1, host='localhost', port=9094, errorCode=0, errorMessage='')])
Found coordinator 1 for link my-link.
Node info: localhost:9094 (id: 1 rack: null isFenced: false)
Creating topic test using bootstrap server localhost:9094.
Created topic test.
```