If not set cluster id in formatting the Namenode, is there a policy in hdfs to guarantee the even of distributing DataNodes into different Namespace, or just randomly?
2013/7/4 Azuryy Yu <[email protected]> > Additional, > > If these are two new clusters, then on each namenode, using "hdfs namenode > -format -clusterID yourID" > > But if you want to upgrade these two clusters from NonHA to HA, then using > "bin/start-dfs.sh -upgrade -clusterID yourID" > > > > On Thu, Jul 4, 2013 at 3:14 PM, Azuryy Yu <[email protected]> wrote: > >> >> This is because you don't use the same clusterID. all data nodes and >> namenodes should use the same clusterID. >> >> >> On Thu, Jul 4, 2013 at 3:12 PM, Bing Jiang <[email protected]>wrote: >> >>> Hi, all >>> >>> We try to use hadoop-2.0.5-alpha, using two namespaces, one is for hbase >>> cluster, and the other one is for common use.At the same time, we use >>> Quorum Journal policy as HA. >>> >>> GS-CIX-SEV0001, GS-CIX-SEV0002, namenodes in hbasecluster namespace >>> >>> GS-CIX-SEV0003, GS-CIX-SEV0004, namenodes in commoncluster namespace. >>> >>> GS-CIX-SEV0001~GS-CIX-SEV0008 , 8 machines used as Datanode >>> >>> After launching the hdfs cluster all, there is something which makes me >>> confused, that each namespace has half of the datanodes. >>> >>> NameNode 'GS-CIX-SEV0004:9100' >>> >>> Started: Thu Jul 04 10:28:00 CST 2013 >>> Version: 2.0.5-alpha, 1488459 >>> Compiled: 2013-06-01T04:05Z by jenkins from branch-2.0.5-alpha >>> Cluster ID: CID-15c48d78-2137-4c6e-aacf-0edbf2bb3db7 >>> Block Pool ID: BP-1792015895-10.100.2.3-1372904504940 >>> >>> >>> Browse the filesystem >>> NameNode Logs >>> Go back to DFS home >>> >>> ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ >>> Live Datanodes : 4 >>> >>> >>> Last Admin Configured Used Non DFS Remaining >>> Used Used Remaining Block Block Pool Failed >>> Node Contact State Capacity (GB) Used (GB) >>> (%) (%) (%) Blocks Pool Used (%)> Volumes >>> (GB) >>> (GB) Used (GB) Blocks >>> GS-CIX-SEV0001 1 In Service 888.07 0.00 116.04 772.03 >>> 0.00 ┌────────────┐ 86.93 0 0.00 0.00 0 >>> >>> └────────────┘ >>> GS-CIX-SEV0002 1 In Service 888.07 0.00 135.50 752.57 >>> 0.00 ┌────────────┐ 84.74 0 0.00 0.00 0 >>> >>> └────────────┘ >>> GS-CIX-SEV0005 1 In Service 888.07 0.00 97.61 790.46 >>> 0.00 ┌────────────┐ 89.01 0 0.00 0.00 0 >>> >>> └────────────┘ >>> GS-CIX-SEV0006 1 In Service 888.07 0.00 122.30 765.77 >>> 0.00 ┌────────────┐ 86.23 0 0.00 0.00 0 >>> >>> └────────────┘ >>> >>> >>> Another Namespace's NameNode: >>> >>> NameNode 'GS-CIX-SEV0001:9100' >>> >>> Started: Thu Jul 04 10:19:03 CST 2013 >>> Version: 2.0.5-alpha, 1488459 >>> Compiled: 2013-06-01T04:05Z by jenkins from branch-2.0.5-alpha >>> Cluster ID: CID-1a53483d-000e-4726-aef1-f500bedb1df6 >>> Block Pool ID: BP-1142418822-10.100.2.1-1372904314309 >>> >>> >>> Browse the filesystem >>> NameNode Logs >>> Go back to DFS home >>> >>> ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ >>> Live Datanodes : 4 >>> >>> >>> Last Admin Configured Used Non DFS Remaining >>> Used Used Remaining Block Block Pool Failed >>> Node Contact State Capacity (GB) Used (GB) >>> (%) (%) (%) Blocks Pool Used (%)> Volumes >>> (GB) >>> (GB) Used (GB) Blocks >>> GS-CIX-SEV0003 0 In Service 888.07 0.00 150.54 737.53 >>> 0.00 ┌────────────┐ 83.05 0 0.00 0.00 0 >>> >>> └────────────┘ >>> GS-CIX-SEV0004 0 In Service 888.07 0.00 177.22 710.85 >>> 0.00 ┌────────────┐ 80.04 0 0.00 0.00 0 >>> >>> └────────────┘ >>> GS-CIX-SEV0007 0 In Service 888.07 0.00 62.91 825.16 >>> 0.00 ┌────────────┐ 92.92 0 0.00 0.00 0 >>> >>> └────────────┘ >>> GS-CIX-SEV0008 0 In Service 888.07 0.00 125.25 762.82 >>> 0.00 ┌────────────┐ 85.90 0 0.00 0.00 0 >>> >>> >>> >>> And check the DN(GS-CIX-SEV0001)'s log, it prints like this: >>> 2013-07-04 10:34:51,699 FATAL >>> org.apache.hadoop.hdfs.server.datanode.DataNode: Initialization failed for >>> block pool Block pool BP-1142418822-10.100.2.1-1372904314309 (storage id >>> DS-1677272131-10.100.2.1-50010-1372905291690) service to GS-CIX-SEV0001/ >>> 10.100.2.1:9100 >>> java.io.IOException: Inconsistent storage IDs. Name-node returned >>> DS811369792. Expecting DS-1677272131-10.100.2.1-50010-1372905291690 >>> at >>> org.apache.hadoop.hdfs.server.datanode.DataNode.bpRegistrationSucceeded(DataNode.java:731) >>> at >>> org.apache.hadoop.hdfs.server.datanode.BPOfferService.registrationSucceeded(BPOfferService.java:308) >>> at >>> org.apache.hadoop.hdfs.server.datanode.BPServiceActor.register(BPServiceActor.java:632) >>> at >>> org.apache.hadoop.hdfs.server.datanode.BPServiceActor.connectToNNAndHandshake(BPServiceActor.java:225) >>> at >>> org.apache.hadoop.hdfs.server.datanode.BPServiceActor.run(BPServiceActor.java:664) >>> at java.lang.Thread.run(Thread.java:662) >>> >>> It is proved that one datanode has been required to attached to only one >>> namespace? >>> >>> Any views about it will be thankful. >>> >>> Regards~ >>> >>> >>> -- >>> Bing Jiang >>> Tel：(86)134-2619-1361 >>> weibo: http://weibo.com/jiangbinglover >>> BLOG: http://blog.sina.com.cn/jiangbinglover >>> National Research Center for Intelligent Computing Systems >>> Institute of Computing technology >>> Graduate University of Chinese Academy of Science >>> >> >> > -- Bing Jiang Tel：(86)134-2619-1361 weibo: http://weibo.com/jiangbinglover BLOG: http://blog.sina.com.cn/jiangbinglover National Research Center for Intelligent Computing Systems Institute of Computing technology Graduate University of Chinese Academy of Science
