Solr 4.10.3没有在新集群启动时进行领导者选举 - 挂起节点

时间:2015-01-17 22:03:06

标签: solr apache-zookeeper solrcloud

我最近将我们的一项服务从4.10.1升级到4.10.3,并注意到,在启动新群集时,所有节点都会停留在' down'州。查看日志并将它们与我以前的版本进行比较,我还注意到集群从未进入领导者选举。

背景 我们使用start.jar使用相当简单的启动过程,并依赖于自动核心发现。我们在其他主机上运行了3个zookeeper实例,并在不同的主机上启动了1个或2个Solr实例来测试新版本。 ZK实例中包含的所有配置文件都是最新的solr 4.10.3标准。 Solr home是默认值(/ solr)。除了solrconfig.xml中的solr-match-version之外,我们的4.10.1和4.10.3部署之间没有区别。版本之间的所有其他配置更改都不存在。

core.properties: 这些设置是不必要的,因为一切都与我们的目录结构相匹配,但无论如何它都在这里。给定属性在命令行中指定为ttPoiMDS(您将在下面的日志中注明)。

name ${our.index.name:}
collection ${our.index.name:}

Solr log: 以下是我们日志的输出。

INFO 2015-01-17 16 39 56 028 jetty-8.1.10.v20130312
INFO 2015-01-17 16 39 56 060 Opened /logs/access_log
INFO 2015-01-17 16 39 56 069 Deployment monitor /install_dir/solr-4.10.3/contexts at interval 0
INFO 2015-01-17 16 39 56 079 Deployable added: /install_dir/solr-4.10.3/contexts/solr-jetty-context.xml
INFO 2015-01-17 16 39 57 710 NO JSP Support for /solr, did not find org.apache.jasper.servlet.JspServlet
INFO 2015-01-17 16 39 57 987 SolrDispatchFilter.init()
INFO 2015-01-17 16 39 58 006 JNDI not configured for solr (NoInitialContextEx)
INFO 2015-01-17 16 39 58 007 solr home defaulted to 'solr/' (could not find system property or JNDI)
INFO 2015-01-17 16 39 58 007 new SolrResourceLoader for directory: 'solr/'
INFO 2015-01-17 16 39 58 232 Loading container configuration from /install_dir/solr-4.10.3/solr/solr.xml
INFO 2015-01-17 16 39 58 363 Config-defined core root directory: /install_dir/solr-4.10.3/solr
INFO 2015-01-17 16 39 58 371 New CoreContainer 971279180
INFO 2015-01-17 16 39 58 371 Loading cores into CoreContainer [instanceDir=solr/]
INFO 2015-01-17 16 39 58 396 Setting socketTimeout to: 0
INFO 2015-01-17 16 39 58 396 Setting urlScheme to: null
INFO 2015-01-17 16 39 58 403 Setting connTimeout to: 0
INFO 2015-01-17 16 39 58 403 Setting maxConnectionsPerHost to: 20
INFO 2015-01-17 16 39 58 403 Setting corePoolSize to: 0
INFO 2015-01-17 16 39 58 403 Setting maximumPoolSize to: 2147483647
INFO 2015-01-17 16 39 58 403 Setting maxThreadIdleTime to: 5
INFO 2015-01-17 16 39 58 403 Setting sizeOfQueue to: -1
INFO 2015-01-17 16 39 58 403 Setting fairnessPolicy to: false
INFO 2015-01-17 16 39 58 563 Creating UpdateShardHandler HTTP client with params: socketTimeout=0&connTimeout=0&retry=false
INFO 2015-01-17 16 39 58 565 SLF4J impl is org.slf4j.impl.Log4jLoggerFactory
INFO 2015-01-17 16 39 58 566 Registering Log Listener [Log4j (org.slf4j.impl.Log4jLoggerFactory)]
INFO 2015-01-17 16 39 58 568 Host Name: 
INFO 2015-01-17 16 39 58 568 Zookeeper client=our_zk_instance:<port_num>,...
INFO 2015-01-17 16 39 58 625 Waiting for client to connect to ZooKeeper
INFO 2015-01-17 16 39 58 663 Watcher org.apache.solr.common.cloud.ConnectionManager@76cdaff7 name:ZooKeeperConnection Watcher:our_zk_instance:<port_num> got event WatchedEvent state:SyncConnected type:None path:null path:null type:None
INFO 2015-01-17 16 39 58 663 Client is connected to ZooKeeper
INFO 2015-01-17 16 39 58 726 Register node as live in ZooKeeper:/live_nodes/<our_ip>:8983_solr
INFO 2015-01-17 16 39 58 732 makePath: /live_nodes/<our_ip>:8983_solr
INFO 2015-01-17 16 39 58 748 Overseer (id=null) closing
INFO 2015-01-17 16 39 58 763 I am going to be the leader <our_ip>:8983_solr
INFO 2015-01-17 16 39 58 767 makePath: /overseer_elect/leader
INFO 2015-01-17 16 39 58 778 Overseer (id=93155855038349317-<our_ip>:8983_solr-n_0000000002) starting
INFO 2015-01-17 16 39 58 869 Starting OverseerAutoReplicaFailoverThread autoReplicaFailoverWorkLoopDelay=10000 autoReplicaFailoverWaitAfterExpiration=30000 autoReplicaFailoverBadNodeExpiration=60000
INFO 2015-01-17 16 39 58 901 Process current queue of collection creations
INFO 2015-01-17 16 39 58 905 Updating cluster state from ZooKeeper... 
INFO 2015-01-17 16 39 58 937 Starting to work on the main queue
INFO 2015-01-17 16 39 58 975 Looking for core definitions underneath /install_dir/solr-4.10.3/solr
INFO 2015-01-17 16 39 58 989 Found core ttPoiMDS in /install_dir/solr-4.10.3/solr/ttPoiMDS/
INFO 2015-01-17 16 39 58 989 Found 1 core definitions
INFO 2015-01-17 16 39 58 995 publishing core=ttPoiMDS state=down collection=ttPoiMDS
INFO 2015-01-17 16 39 58 995 numShards not found on descriptor - reading it from system property
INFO 2015-01-17 16 39 59 002 look for our core node name
INFO 2015-01-17 16 39 59 002 waiting to find shard id in clusterstate for ttPoiMDS
INFO 2015-01-17 16 39 59 002 Check for collection zkNode:ttPoiMDS
INFO 2015-01-17 16 39 59 003 LatchChildWatcher fired on path: /overseer/queue state: SyncConnected type NodeChildrenChanged
INFO 2015-01-17 16 39 59 004 Collection zkNode exists
INFO 2015-01-17 16 39 59 004 Load collection config from:/collections/ttPoiMDS
INFO 2015-01-17 16 39 59 008 path=/collections/ttPoiMDS configName=ttPoiMDS specified config exists in ZooKeeper
INFO 2015-01-17 16 39 59 008 new SolrResourceLoader for directory: '/install_dir/solr-4.10.3/solr/ttPoiMDS/'
INFO 2015-01-17 16 39 59 013 Update state numShards=1 message={
    "operation":"state",
    "numShards":"1",
    "shard":null,
    "roles":null,
    "state":"down",
    "core":"ttPoiMDS",
    "collection":"ttPoiMDS",
    "node_name":"<our_ip>:8983_solr",
    "base_url":"http://<our_ip>:8983/solr"}
INFO 2015-01-17 16 39 59 013 node=core_node2 is already registered
INFO 2015-01-17 16 39 59 013 shard=shard1 is already registered
INFO 2015-01-17 16 39 59 042 A cluster state change: WatchedEvent state:SyncConnected type:NodeDataChanged path:/clusterstate.json, has occurred - updating... (live nodes size: 1)
INFO 2015-01-17 16 39 59 079 Adding specified lib dirs to ClassLoader
INFO 2015-01-17 16 39 59 080 Adding 'file:/install_dir/solr-4.10.3/lib/mq/solr-search-ahead-2.0.0.jar' to classloader
INFO 2015-01-17 16 39 59 236 Using Lucene MatchVersion: 4.10.3
INFO 2015-01-17 16 39 59 400 Loaded SolrConfig: solrconfig.xml
INFO 2015-01-17 16 39 59 412 Reading Solr Schema from /configs/ttPoiMDS/schema.xml
INFO 2015-01-17 16 39 59 451 [ttPoiMDS] Schema name=poiMDS
INFO 2015-01-17 16 39 59 841 user.dir=/install_dir/solr-4.10.3
INFO 2015-01-17 16 39 59 841 SolrDispatchFilter.init() done
INFO 2015-01-17 16 39 59 870 Deployable added: /install_dir/solr-4.10.3/contexts/ns-jetty-context.xml
INFO 2015-01-17 16 39 59 943 NO JSP Support for /, did not find org.apache.jasper.servlet.JspServlet
INFO 2015-01-17 16 39 59 997 Started SocketConnector@0.0.0.0:8983

如果我设法错过了什么,请告诉我。这是我怀疑的,因为很难相信某些东西已经进入代码库。它确实看起来像ZkController类已经有了很大的改变。

任何帮助将不胜感激,谢谢!

修改

我尝试使用外部ZooKeeper主机和自动核心发现,使用分发包中包含的示例部署,并遇到了同样的问题。

0 个答案:

没有答案