hbase-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Henry JunYoung KIM <henry.jy...@gmail.com>
Subject Re: how to delete a zombie table permanently?
Date Fri, 10 Aug 2012 08:33:50 GMT
Hi, JM.

I followed procedures you mentioned. 

here is it.

from zookeeper, I deleted it 

---------------
$] rmr /hbase/table/*
$] ls /hbase/table
Node does not exist: /hbase/table
---------------

and then, I checked my hbase cluster.

---------------
$> hbase hbck 
,,,,
Number of empty REGIONINFO_QUALIFIER rows in .META.: 0
Exception in thread "main" java.lang.NullPointerException
	at org.apache.hadoop.hbase.zookeeper.ZKTable.getDisabledOrDisablingTables(ZKTable.java:397)
	at org.apache.hadoop.hbase.util.HBaseFsck$1.connect(HBaseFsck.java:996)
	at org.apache.hadoop.hbase.util.HBaseFsck$1.connect(HBaseFsck.java:991)
	at org.apache.hadoop.hbase.client.HConnectionManager.execute(HConnectionManager.java:360)
	at org.apache.hadoop.hbase.util.HBaseFsck.loadDisabledTables(HBaseFsck.java:991)
	at org.apache.hadoop.hbase.util.HBaseFsck.onlineConsistencyRepair(HBaseFsck.java:360)
	at org.apache.hadoop.hbase.util.HBaseFsck.onlineHbck(HBaseFsck.java:383)
	at org.apache.hadoop.hbase.util.HBaseFsck.main(HBaseFsck.java:3121)
---------------

OMG.

ok, let's refresh master's cache which is from zk.

---------------
$> hadoop-hbase-master restart
$> hbase hbck
…
ERROR: Found inconsistency in table blog
Summary:
  -ROOT- is okay.
    Number of regions: 1
    Deployed on:  search-ddm-test10,60020,1344586183254
  .META. is okay.
    Number of regions: 1
    Deployed on:  search-ddm-test6,60020,1344586183295
Table blog is inconsistent.
    Number of regions: 3
    Deployed on:  search-ddm-test5,60020,1344586183269 search-ddm-test8,60020,1344586183257
search-ddm-test9,60020,1344586183260
8 inconsistencies detected.
Status: INCONSISTENT
---------------

now, it shows up the different situation for me. (thanks)

from hbase shell, I could check the existence of the tables I want to delete. 

---------------
hbase(main):002:0> list
TABLE                                                                                    
                                                                         
blog                                                                                     
                                                                         
sidx                                                                                     
                                                                         
2 row(s) in 0.0180 seconds
---------------
try to fix it again. 

---------------
$> hbase hbck -fix
…
Number of empty REGIONINFO_QUALIFIER rows in .META.: 0
12/08/10 17:29:54 DEBUG util.HBaseFsck: There are 2 region info entries
Summary:
  -ROOT- is okay.
    Number of regions: 1
    Deployed on:  search-ddm-test10,60020,1344586183254
  .META. is okay.
    Number of regions: 1
    Deployed on:  search-ddm-test6,60020,1344586183295
0 inconsistencies detected.
Status: OK
---------------

hbase shell says also

---------------
hbase(main):009:0> list
TABLE                                                                                    
                                                                         
0 row(s) in 0.0160 seconds
---------------


not easy procedures to fix it  for me. 
anyway, finally, in the conclusion, I fixed it. 


2012. 8. 9., 오후 9:00, Jean-Marc Spaggiari <jean-marc@spaggiari.org> 작성:

> Hi Henry,
> 
> I faced the same issue not so long time ago...
> 
> Can you take a look at what you have in zookeeper under /hbase/table ?
> 
> If you table is there, that's why you see it on the list. Simply
> remove it from zookeeper.
> 
> You can also take a look there: https://issues.apache.org/jira/browse/HBASE-6294
> 
> JM
> 
> 2012/8/9, henry.kim <henry.jykim@gmail.com>:
>> hi, hbase users.
>> 
>> I got a problem when I am testing coprocessors which is released at base
>> 0.92.1.
>> 
>> here is the hbase shell outputs
>> 
>> ----------------
>> hbase(main):001:0> truncate 'blog'
>> Truncating 'blog' table (it may take a while):
>> 
>> ERROR: Unknown table blog!
>> 
>> Here is some help for this command:
>>  Disables, drops and recreates the specified table.
>> 
>> 
>> hbase(main):002:0> list
>> TABLE
>> 
>> 
>> blog
>> 
>> 
>> counter
>> 
>> 
>> sidx
>> 
>> 
>> 3 row(s) in 15.3610 seconds
>> ----------------
>> 
>> yes, table 'blog' is a zombie I think.
>> 
>> there was some big stress from coprocessors which is using
>> RegionServerObserver interfaces.
>> 
>> how could I fix this situation?


Mime
View raw message