@lemon @ascrutae I change the proto file path for preparing 6.0, but didn't change .net namespace. But I think if you sync the protocol submodule, you still need some changes.

* #1207

Return true when segment parse throw exception to avoid parse the segment over and over again which read from buffer file.

* no message

* #1202

1. Determine the log is enabled for the DEBUG level before printing message.
2. Make the columns initialize to be static attribute.
3. Topology build optimize: Cache query result to avoid repeating queries.

* #1202

Add elasticsearch batch process setting into application.yml.

* #1202

Fixed check style error.

* #1202

Use XContentFactory to build source to insert into elasticsearch.

* #1206

1. Support list data type attribute in streaming data.
2. Support multiple service names in segment duration index.

* Fixed the conflict.

* #1206

Add TODO description to notify linjiaqi.

* #1206

Remove @NotNull

* Add Zipkin receiver

* #1124

Add heart beat time into service name index to filter non-active service name.

* Filter service name by heart beat time when query service name count.

* #1124
1. Add application id to be a query condition of searchService method.
2. Filter service name by given time which calculated by the minute metric TTL setting. ( CurrentTimeStamp - minuteMetricTTL )

* Fixed the auto test failure.

#1124

Add folder lock to collector's buffer folder.

Module class rename to ModuleDefine, Module is the JDK9 class. It is conflict with our Module class. (#1168)

Delete buffer directory when collector restart.

Segment end time: maximum end time in all span

#1149

* Add non-merge persistence worker for segment, segment_duration, global_trace persistence.

* omitted class.

* Disable _all field.

* no message

* Refactor the old ComponentDefine and ComponentsDefine to be based on file for better extenable.

* Support don't uplink componentId, but componentName in segment.

* Fix CI.

* Move component id exchange logic from SpanDecorator to SpanIdExchanger.

* Fixed the compile error cause of class construct changes.

* Provide short database column name.

* Compile successful

* Attribute name change from source to target.

* Provide highPerformanceMode setting for short column name speciality.

* Fixed a bug that persistent timer started multiple times.
Running about 10 minute, the result of collector performance test is above 1w TPS when using MacBook 2017.

* Refactor the collector configuration initialization.

#1047

Add contains point method instead instance of for decide span listener class type.
Transform minute time bucket just once.

#1032

* 1. Add the performance collection point.
2. Report formatter.

#1032

* Add a new cache implementation by caffeine.

#1032

* Performance optimization:
Make all the loggers class variable to have the static modifier.

* Time stamp format to time bucket performance optimization.

#1032

* 1. Add the performance collection point.
2. Report formatter.

#1032

* Add a new cache implementation by caffeine.

#1032

* Performance optimization:
Make all the loggers class variable to have the static modifier.

2. Report formatter.

#1032

* Collector instrument agent.

* Make the agent.jar output to the /agent folder in collector package.

* Try to add metric annotation.

* Change the instrument ways.

* Reformat.

* Finish the agent codes. Wait for @peng-yongsheng 's metric requirements.

* Debug new instrument.

* Fix a detect bug.

* Finish the instrument

* Fix compile issue.

* Fixed #920

* Sync sub module.

2. Setting the src span type when agent register service name.
3. Revert the topology build logic.

2. Add column named service_name_keyword into service_name elastic search table by the type of keyword.

Use listener factory to create new instance when create SegmentParser instance. Because of the listener is stateful.