* Try adjustment.
* Remove heartbeat.
* Make heartbeat back to work. And default threadpool size more sense.
* Make L2 to less than before.
* Make instance heartbeat works.
* Try L1 aggregation thread = core * 2 * 2.
* Support to register different host/port at cluster coordinator. Only for Zookeeper and Consul
* Provide document for internalComHost and internalComPort
* Reset port to default.
* The prototype of topN sql worker.
* Add scope and new manual dispatcher.
* no message
* Finish topN persistence codes. Not test yet. And query have not added.
* Finish the top n database statement persistent.
* Support different slow thresholds for different db types.
* Fix wrong db default threshold
* Finish new query protocol binding.
* Provide query empty implementation and sync ui.
* Finish all codes, hope it works :P
I will run the tests tonight.
* Fix
* Fix
* Fix a startup issue.
* Add time bucket to persistent.
* Fix wrong query result column name.
* Fix Database query.
* Fix checkstyle.
* Fix wrong order logic, and add a test case to verify, to fix https://github.com/apache/incubator-skywalking/pull/2239#discussion_r255948249
* Fixed an error that took a non-protocol uri string as a java.net.URL construct parameter.
* Fixed an error that took a non-protocol uri string as a java.net.URL construct parameter.
* Fixed checkstyle error.
fixed CPU UsagePercent Error
The cpu usage storage longitude should be one percent, not one in ten thousand. When the storage unit is one ten thousandth, as long as the usage rate does not exceed 100%. Stored values are less than one. When the current query is forced to be converted to a long type, it will always be 0. This is not correct.
* Add Database source
* Add DatabaseSource source and database_* script for oal
* Enable Database_Indicator
* Add file license
* Rename DatabaseService to ClientDatabase. In the future, we may have real database server monitoring.
* Add query database from H2
* refactor
* update submodule
* refactor
* Fix style
* Fix: unknown tag error
* fix checkstyle
* Optimize the grpc channel management.
* Rename old version concepts
* Notify when rebuild connection.
* Remove notify in new channel created.
* Remove println.
* 1. Sleep 500 milliseconds after a batch re-call finish.
2. Re-create register lock index when the system property named debug is setting.
3. Get the register inventory before lock to avoid increment the sequence but not use.
4. Return the exchange flag after all the references and spans in one segment parsed to reduce the number of segment parse.
5. Put the not exchanged segment into a collection then try to exchange no more than 10 times because of the exchange is asynchronous.
6. Cache the segment object to avoid repeated deserialization.
#2185
* #2185
1. Sleep 500 milliseconds after a batch re-call finish.
2. Re-create register lock index when the system property named debug is setting.
3. Get the register inventory before lock to avoid increment the sequence but not use.
* fix no language attr value in mesh (#2182)
* Fixed the inventory register lock invalid bug. (#2184)
* #2183
Merge commit into master.
* Make the h2 register lock same as mysql.
* Fix#2186