hbase-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Duo Zhang (JIRA)" <j...@apache.org>
Subject [jira] [Resolved] (HBASE-22439) 重启hbase,特定节点有几率出现Num.Regions=0的情况
Date Fri, 17 May 2019 07:54:00 GMT

     [ https://issues.apache.org/jira/browse/HBASE-22439?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Duo Zhang resolved HBASE-22439.
-------------------------------
    Resolution: Invalid

Please subscribe and send email to user@hbase.apache.org to ask.

And I think this is a question, not a bug. If you restart the whole cluster, it is fine that
some region servers do not hold any regions, the load balancer will balance regions later.

And please use English if possible. Not all the people here can read Chinese...
==== Chinese below ====

直接订阅user@hbase.apache.org然后发邮件问吧。

看描述,我感觉这个不是bug,重启整个集群确实有可能有region server上没有任何region,打开balancer过一会儿就均匀了。

另外,请尽量用英文提问和回复,社区里不是所有人都能看懂中文。。。

> 重启hbase,特定节点有几率出现Num.Regions=0的情况
> ---------------------------------
>
>                 Key: HBASE-22439
>                 URL: https://issues.apache.org/jira/browse/HBASE-22439
>             Project: HBase
>          Issue Type: Bug
>          Components: regionserver
>    Affects Versions: 1.4.9
>         Environment: centos 6.9  
> zk 4.6.11
> hadoop2.7.7
> hbase 1.4.9
>            Reporter: sylvanas
>            Priority: Major
>              Labels: hbase
>
> 每次启动hbase,特定节点(总磁盘是其他节点50%,但是占用率基本位置在25%左右)有几率出现Num.Regions=0的现象(出现几率大于50%),regions被其他节点分摊,整个过程master、regionserver均为发现任何有价值的日志。
> 最终regtion=0所在机器输入日志如下:
> {code:java}
> //代码占位符
> ...
> 2019-05-17 10:38:29,894 INFO [regionserver/gladslave3/10.86.10.103:16020] regionserver.ReplicationSourceManager:
Current list of replicators: [gladslave4,16020,1558060698511, gladslave2,16020,1558060697919,
gladslave3,16020,1558060700753, gladslave1,16020,1558060697694] other RSs: [gladslave4,16020,1558060698511,
gladslave2,16020,1558060697919, gladslave3,16020,1558060700753, gladslave1,16020,1558060697694]
> 2019-05-17 10:38:29,936 INFO [SplitLogWorker-gladslave3:16020] regionserver.SplitLogWorker:
SplitLogWorker gladslave3,16020,1558060700753 starting
> 2019-05-17 10:38:29,936 INFO [regionserver/gladslave3/10.86.10.103:16020] regionserver.HeapMemoryManager:
Starting HeapMemoryTuner chore.
> 2019-05-17 10:38:29,939 INFO [regionserver/gladslave3/10.86.10.103:16020] regionserver.HRegionServer:
Serving as gladslave3,16020,1558060700753, RpcServer on gladslave3/10.86.10.103:16020, sessionid=0x401e4be60870092
> 2019-05-17 10:38:30,467 INFO [regionserver/gladslave3/10.86.10.103:16020] quotas.RegionServerQuotaManager:
Quota support disabled
> 2019-05-17 10:38:35,699 INFO [regionserver/gladslave3/10.86.10.103:16020] wal.FSHLog:
WAL configuration: blocksize=128 MB, rollsize=121.60 MB, prefix=gladslave3%2C16020%2C1558060700753,
suffix=, logDir=hdfs://haservice/hbase/WALs/gladslave3,16020,1558060700753, archiveDir=hdfs://haservice/hbase/oldWALs
> 2019-05-17 10:38:37,122 INFO [regionserver/gladslave3/10.86.10.103:16020] wal.FSHLog:
Slow sync cost: 369 ms, current pipeline: []
> 2019-05-17 10:38:37,123 INFO [regionserver/gladslave3/10.86.10.103:16020] wal.FSHLog:
New WAL /hbase/WALs/gladslave3,16020,1558060700753/gladslave3%2C16020%2C1558060700753.1558060715708
> {code}
>  
> 但是,修改日志级别为debug后,Num.Regions=0的情况再也没有发生过(重复启动测试20次为发现问题),但是,每次重启hbase,各regionserver的regions数量发生变化,并不是上一次停止时的数量。
>  
>  



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Mime
View raw message