Thanks shaofeng, I set kylin.rest.servers=localhost:8081,localhost:8082, but final String[] nodes = KylinConfig.getInstanceFromEnv().getRestServers(); only returns the first element "localhost:8081" for nodes, it's weird and incorrect, thus, I want to look into the code final String[] nodes = KylinConfig.getInstanceFromEnv().getRestServers(); but after the 1st this morning, I cannot hit this breakpoint anymore later~
2015-05-08 15:41 GMT+08:00 Shi, Shaofeng <[email protected]>: > I’m trying to reproduce in my sandbox, will let you know if have any > findings. > > On 5/8/15, 3:23 PM, "dong wang" <[email protected]> wrote: > > >hi, > >when I debug the following java codes , the breakpoint can be hit "final > >BroadcastEvent broadcastEvent = broadcastEvents.takeFirst();" but, it > >cannot hit the bp of the line "final String[] nodes = > >KylinConfig.getInstanceFromEnv().getRestServers();", do anyone how to > >make it hit? > > > > private Broadcaster() { > > Executors.newSingleThreadExecutor().execute(new Runnable() { > > @Override > > public void run() { > > final String[] nodes = > >KylinConfig.getInstanceFromEnv().getRestServers(); > > if (nodes == null || nodes.length < 1) {//TODO if the node > >count is greater than 1, it means it is a cluster > > logger.warn("There is no available rest server; check > >the 'kylin.rest.servers' config"); > > return; > > } > > final List<RestClient> restClients = Lists.newArrayList(); > > for (String node : nodes) { > > restClients.add(new RestClient(node)); > > } > > final ExecutorService wipingCachePool = > >Executors.newFixedThreadPool(restClients.size()); > > while (true) { > > try { > > final BroadcastEvent broadcastEvent = > >broadcastEvents.takeFirst(); > > > >2015-05-07 21:07 GMT+08:00 dong wang <[email protected]>: > > > >> @qianhao, > >> 1, I will debug the code if available > >> 2, just now, as observed, after "REFRESH" a cube, even though I click > >> "Reload Metadata" under "Admin" tab on the WEB of the "QUERY" kylin > >> instances, the cube status cannot be syncd with the "ALL" kylin > >> instance(but if restarting the "QUERY" kylin instances, "QUERY" > >>instances > >> can sync the FRESH cube result info), another thing is that it works OK > >>for > >> "BUILD" a cube if just reloading the metadata without restarting the > >> instance, thus,I guess, the above 2 correct URL put methods will not > >>work > >> for "REFERSH" as well > >> > >> 2015-05-07 20:17 GMT+08:00 Zhou, Qianhao <[email protected]>: > >> > >>> Thanks for the research, the root cause should be the wrong URL > >>>redirect > >>> > >>> 2015年5月7日 18:59于 dong wang <[email protected]>写道： > >>> @qianzhao, as tested, another discovery is that if I use the URL like: > >>> > >>> > >>> > http://test.abc.com:8082/kylin/api/cache/cube/lbs_map_active_user_fact/u > >>>pdate > >>> , > >>> the cube status of the "QUERY" kylin instance also works SUCCESSFULLY > >>> > >>> thus, the problem should be something wrong with the communication > >>>between > >>> "ALL" mode instance and "QUERY" mode instances > >>> > >>> 2015-05-07 18:12 GMT+08:00 dong wang <[email protected]>: > >>> > >>> > @qianhao, @shaofeng, > >>> > > >>> > today, I tried to mockup a PUT request as you mentioned above with > >>>the > >>> > following different URLs: > >>> > 1, http://test.abc.com:8082/cube/lbs_map_new_user_fact/update > >>> > > >>> > nothing happened, even more, no log information received in the kylin > >>> > instance log file deployed with port 8082 > >>> > > >>> > 2, http://test.abc.com:8082/kylin/api/cache/all/all/update > >>> > > >>> > with this API, It works OK, the cube status is flushed SUCCESSFULLY, > >>> log > >>> > information received as shown below: > >>> > > >>> > [http-bio-8082-exec-7]:[2015-05-07 > >>> > > >>> > >>>17:54:57,469][INFO][org.apache.kylin.rest.controller.CacheController.wip > >>>eCache(CacheController.java: > >>> > 64)] - wipe cache type: ALL event:UPDATE name:all > >>> > [http-bio-8082-exec-7]:[2015-05-07 > >>> > > >>> > >>>17:54:57,488][DEBUG][org.apache.kylin.metadata.MetadataManager.reloadAll > >>>SourceTable(MetadataManager. > >>> > java:244)] - Reloading SourceTable from folder > >>> > kylin_metadata(key='/table')@kylin_metadata@hbase > >>> > [http-bio-8082-exec-7]:[2015-05-07 > >>> > > >>> > >>>17:54:57,533][DEBUG][org.apache.kylin.metadata.MetadataManager.reloadAll > >>>SourceTable(MetadataManager. > >>> > java:253)] - Loaded 8 SourceTable(s) > >>> > [http-bio-8082-exec-7]:[2015-05-07 > >>> > > >>> > >>>17:54:57,534][DEBUG][org.apache.kylin.metadata.MetadataManager.reloadAll > >>>SourceTableExd(MetadataManag > >>> > er.java:201)] - Reloading SourceTable exd info from folder > >>> > kylin_metadata(key='/table_exd')@kylin_metadata@hbase > >>> > [http-bio-8082-exec-7]:[2015-05-07 > >>> > > >>> > >>>17:54:57,562][DEBUG][org.apache.kylin.metadata.MetadataManager.reloadAll > >>>SourceTableExd(MetadataManag > >>> > er.java:210)] - Loaded 8 SourceTable EXD(s) > >>> > [http-bio-8082-exec-7]:[2015-05-07 > >>> > > >>> > >>>17:54:57,562][DEBUG][org.apache.kylin.metadata.MetadataManager.reloadAll > >>>DataModel(MetadataManager.ja > >>> > va:291)] - Reloading DataModel from folder > >>> > kylin_metadata(key='/model_desc')@kylin_metadata@hbase > >>> > [http-bio-8082-exec-7]:[2015-05-07 > >>> > > >>> > >>>17:54:57,599][DEBUG][org.apache.kylin.metadata.MetadataManager.reloadAll > >>>DataModel(MetadataManager.ja > >>> > va:305)] - Loaded 8 DataModel(s) > >>> > [http-bio-8082-exec-7]:[2015-05-07 > >>> > > >>> > >>>17:54:57,604][INFO][org.apache.kylin.rest.service.BasicService.resetOLAP > >>>DataSources(BasicService.jav > >>> > a:90)] - resetOLAPDataSources is called. > >>> > > >>> > 3, currently, since we deploy 3 kylin instance on the same machine > >>>with > >>> > different ports, and I set export KYLIN_HOME=/path_to_kylin at the > >>>very > >>> > beginning of the file bin/kylin.sh since we cannot set a global env > >>> > KYLIN_HOME > >>> > > >>> > 2015-05-07 17:50 GMT+08:00 Tao Wong <[email protected]>: > >>> > > >>> >> Cube status on query mode instance still can not change > >>>automatically. > >>> >> > >>> >> the all mode node seem to broadcast the query mode nodes,however > >>>,the > >>> >> query > >>> >> node run with error as above and below: > >>> >> > >>> >> Caused by: java.lang.ClassNotFoundException: > >>> >> org.apache.kylin.query.schema.OLAPSchemaFactory > >>> >> 733 at java.net.URLClassLoader$1.run(URLClassLoader.java:366) > >>> >> 734 at java.net.URLClassLoader$1.run(URLClassLoader.java:355) > >>> >> 735 at java.security.AccessController.doPrivileged(Native > >>>Method) > >>> >> 736 at > >>>java.net.URLClassLoader.findClass(URLClassLoader.java:354) > >>> >> 737 at java.lang.ClassLoader.loadClass(ClassLoader.java:425) > >>> >> 738 at > >>> sun.misc.Launcher$AppClassLoader.loadClass(Launcher.java:308) > >>> >> 739 at java.lang.ClassLoader.loadClass(ClassLoader.java:358) > >>> >> 740 at java.lang.Class.forName0(Native Method) > >>> >> 741 at java.lang.Class.forName(Class.java:191) > >>> >> 742 at > >>> >> org.apache.calcite.model.ModelHandler.visit(ModelHandler.java:209) > >>> >> > >>> >> 2015-05-07 14:17 GMT+08:00 Tao Wong <[email protected]>: > >>> >> > >>> >> > After I build a new segment. the kylin.log on the query mode node. > >>> >> > > >>> >> > kylin.log: > >>> >> > > >>> >> > [http-bio-8082-exec-1]:[2015-05-07 > >>> >> > > >>> >> > >>> > >>>13:53:13,681][DEBUG][org.apache.kylin.rest.service.AdminService.getConfi > >>>gAsString(AdminService > >>> >> > .java:91)] - Get Kylin Runtime Config > >>> >> > 3261 [http-bio-8082-exec-1]:[2015-05-07 > >>> >> > > >>> >> > >>> > >>>13:56:08,971][INFO][org.apache.kylin.metadata.project.ProjectL2Cache.loa > >>>dCache(ProjectL2Cache. > >>> >> > java:152)] - Loading L2 project cache for user > >>> >> > 3262 [http-bio-8082-exec-1]:[2015-05-07 > >>> >> > > >>> >> > >>> > >>>13:56:08,974][INFO][org.apache.kylin.rest.controller.TableController.get > >>>HiveTables(TableContro > >>> >> > ller.java:74)] - Return all table metadata in 3 seconds > >>> >> > 3263 [http-bio-8082-exec-4]:[2015-05-07 > >>> >> > >>> > >>>13:59:03,657][ERROR][org.apache.kylin.rest.controller.BasicController.ha > >>>ndleError(BasicControl > >>> >> > ler.java:52)] - > >>> >> > 3264 java.lang.RuntimeException: Error instantiating > >>> >> > JsonCustomSchema(name=USER) > >>> >> > 3265 at > >>> >> > org.apache.calcite.model.ModelHandler.visit(ModelHandler.java:217) > >>> >> > 3266 at > >>> >> > > >>> >> > >>> > >>>org.apache.calcite.model.JsonCustomSchema.accept(JsonCustomSchema.java:3 > >>>6) > >>> >> > 3267 at > >>> >> > org.apache.calcite.model.ModelHandler.visit(ModelHandler.java:144) > >>> >> > 3268 at > >>> >> > org.apache.calcite.model.ModelHandler.<init>(ModelHandler.java:87) > >>> >> > 3269 at > >>> >> > org.apache.calcite.jdbc.Driver$1.onConnectionInit(Driver.java:97) > >>> >> > 3270 at > >>> >> > > >>> >> > >>> > >>>org.apache.calcite.avatica.UnregisteredDriver.connect(UnregisteredDriver > >>>.java:145) > >>> >> > 3271 at > >>> java.sql.DriverManager.getConnection(DriverManager.java:571) > >>> >> > 3272 at > >>> java.sql.DriverManager.getConnection(DriverManager.java:187) > >>> >> > 3273 at > >>> >> > > >>> >> > >>> > >>>org.springframework.jdbc.datasource.DriverManagerDataSource.getConnectio > >>>nFromDriverManager(DriverManagerDataSource.java:17 > >>> >> > 3) > >>> >> > 3274 at > >>> >> > > >>> >> > >>> > >>>org.springframework.jdbc.datasource.DriverManagerDataSource.getConnectio > >>>nFromDriver(DriverManagerDataSource.java:164) > >>> >> > 3275 at > >>> >> > > >>> >> > >>> > >>>org.springframework.jdbc.datasource.AbstractDriverBasedDataSource.getCon > >>>nectionFromDriver(AbstractDriverBasedDataSource.ja > >>> >> > va:149) > >>> >> > 3276 at > >>> >> > > >>> >> > >>> > >>>org.springframework.jdbc.datasource.AbstractDriverBasedDataSource.getCon > >>>nection(AbstractDriverBasedDataSource.java:119) > >>> >> > > >>> >> > 2015-05-07 9:15 GMT+08:00 Shi, Shaofeng <[email protected]>: > >>> >> > > >>> >> >> Dong, you can download a build directly from > >>> >> >> https://kylin.incubator.apache.org/download/; > >>> >> >> > >>> >> >> On 5/7/15, 7:26 AM, "周千昊" <[email protected]> wrote: > >>> >> >> > >>> >> >> >Hi, dong > >>> >> >> > #1 you don't have to add "http://"; prefix for the > >>> >> >> kylin.rest.servers > >>> >> >> >config, > >>> >> >> > according to the log, the Broadcaster event has been > >>> >> >> >successfully > >>> >> >> >created, however there maybe some network issues when "ALL" > >>>server > >>> >> trying > >>> >> >> >to reach "QUERY" server, can you try to mockup a PUT request > >>>like > >>> >> >> >http://test.abc.com:8081/cube/lbs_map_new_user_fact/update to > >>>see > >>> if > >>> >> the > >>> >> >> >network is OK? > >>> >> >> > #2 yes you can git clone the "0.7.1-staging" branch to > >>>build > >>> the > >>> >> >> >binary > >>> >> >> > > >>> >> >> >dong wang <[email protected]>于2015年5月7日周四 上午12:10写道： > >>> >> >> > > >>> >> >> >> 1, please help to review the above 2 response first > >>> >> >> >> 2, in kylin.properties,currently we set kylin.rest.servers= > >>> >> >> >> test.abc.com:8080 > >>> >> >> >> ,test.abc.com:8081,test.abc.com:8082 and test.abc.com:8080 > >>>are > >>> >> the > >>> >> >> >>"ALL" > >>> >> >> >> instance, the other 2 are the "QUERY" kylin instances > >>> >> >> >> should we set with the 'http://' prefix like: > >>> >> kylin.rest.servers= > >>> >> >> >> > >>> >> >> >>http://test.abc.com:8080,http://test.abc.com:8081, > >>> >> >> http://test.abc.com:808 > >>> >> >> >>2 > >>> >> >> >> > >>> >> >> >> 2015-05-06 23:40 GMT+08:00 dong wang <[email protected] > >: > >>> >> >> >> > >>> >> >> >> > @shaofeng, another question is that can I git clone the > >>>branch > >>> >> >> >> > "0.71-staging"( > >>> >> https://github.com/KylinOLAP/Kylin/tree/0.7.1-staging > >>> >> >> ) > >>> >> >> >>to > >>> >> >> >> > get the latest code to build the binary? > >>> >> >> >> > > >>> >> >> >> > 2015-05-06 23:38 GMT+08:00 dong wang > >>><[email protected]>: > >>> >> >> >> > > >>> >> >> >> >> I double checked one piece of the above log of the "ALL" > >>>mode > >>> >> kylin > >>> >> >> >> >> instance: > >>> >> >> >> >> > >>> >> >> >> >> > >>> >> >> >> >> > >>> >> >> >> > >>> >> >> > >>> >> >> > >>> >> > >>> > >>>>>17:36:21,910][INFO][org.apache.kylin.common.restclient.Broadcaster$1.r > >>>>>un( > >>> >> >> >>Broadcaster.java:71)] > >>> >> >> >> >> - new broadcast event:BroadcastEvent{type=cube, name=lbs_ma > >>> >> >> >> >> p_new_user_fact, action=update} > >>> >> >> >> >> 2553 [pool-7-thread-10]:[2015-05-06 > >>> >> >> >> >> > >>> >> >> >> >> > >>> >> >> >> > >>> >> >> > >>> >> >> > >>> >> > >>> > >>>>>17:36:21,911][DEBUG][org.apache.kylin.common.persistence.ResourceStore > >>>>>.pu > >>> >> >> >>tResource(ResourceStore.java:171)] > >>> >> >> >> >> - Saving resource /execute_output/f03d3c3c-68 > >>> >> >> >> >> f4-4593-b01d-2778fcaf098b-15 (Store kylin_metadata@hbase) > >>> >> >> >> >> 2554 [pool-7-thread-10]:[2015-05-06 > >>> >> >> >> >> > >>> >> >> >> >> > >>> >> >> >> > >>> >> >> > >>> >> >> > >>> >> > >>> > >>>>>17:36:21,921][DEBUG][org.apache.kylin.common.persistence.ResourceStore > >>>>>.pu > >>> >> >> >>tResource(ResourceStore.java:171)] > >>> >> >> >> >> - Saving resource /execute_output/f03d3c3c-68 > >>> >> >> >> >> f4-4593-b01d-2778fcaf098b-15 (Store kylin_metadata@hbase) > >>> >> >> >> >> 2555 [http-bio-8081-exec-5]:[2015-05-06 > >>> >> >> >> >> > >>> >> >> >> >> > >>> >> >> >> > >>> >> >> > >>> >> >> > >>> >> > >>> > >>>>>17:36:21,922][INFO][org.apache.kylin.rest.controller.CacheController.w > >>>>>ipe > >>> >> >> >>Cache(CacheController.java:63)] > >>> >> >> >> >> - wipe cache type: CUBE event:UPDATE name:lb > >>> >> s_map_new_user_fact > >>> >> >> >> >> > >>> >> >> >> >> > >>> >> >> >> >> from the first line and the last line, it seems that it > >>>tries > >>> to > >>> >> >> call > >>> >> >> >> >> sync something related to 'rest', however, when I check > >>>the > >>> >> >> >>kylin.log > >>> >> >> >> for > >>> >> >> >> >> each of the "QUERY" mode kylin instances, no related log > >>> reponse > >>> >> to > >>> >> >> >>the > >>> >> >> >> >> rest sync > >>> >> >> >> >> > >>> >> >> >> > > >>> >> >> >> > > >>> >> >> >> > >>> >> >> > >>> >> >> > >>> >> > > >>> >> > >>> > > >>> > > >>> > >> > >> > >
