I think I remember that you had some issues with the old APIs in 0.20,
but AFAIK it should be backward compatible (but with deprecation).

J-D

On Tue, Jul 13, 2010 at 2:13 AM, Yabo Xu <[email protected]> wrote:
> OK, thanks. The thing that delays us on upgrading is the API change. We have
> a bunch of old applications that sits on 0.19.1 API.
>
> Best,
> Arber
>
>
> On Tue, Jul 13, 2010 at 11:17 AM, Jean-Daniel Cryans 
> <[email protected]>wrote:
>
>> Variety of reasons, without evidences (master's log for example) I
>> can't tell exactly. Also, since 0.19 does contain a lot less
>> reliability fixes than 0.20, especially 0.20.5
>>
>> As a comparison, our 20 nodes production cluster is serving real-time
>> data 24/7 without that kind of issue. We're running on the latest cdh2
>> and HBase 0.20 + a couple of home-brewed patches that serves our own
>> particular usage of HBase.
>>
>> J-D
>>
>> On Mon, Jul 12, 2010 at 6:46 PM, Yabo Xu <[email protected]> wrote:
>> > Thanks, J-D.
>> >
>> > This morning I found the data block was automatically deleted. But that
>> > block was indeed there. And because there was not much traffic on the
>> test
>> > cluster, so it seems more possibility goes to the double assignment issue
>> > you mentioned.
>> >
>> > Just curious how does that occur? We may not want to restart every time
>> to
>> > address this issue.
>> >
>> > Thanks again.
>> >
>> > Best,
>> > Arber
>> >
>> >
>> > On Tue, Jul 13, 2010 at 12:21 AM, Jean-Daniel Cryans <
>> [email protected]>wrote:
>> >
>> >> > file=/hbase/-ROOT-/70236052/info/mapfiles/3687060941742211902/data
>> >>
>> >> Can you get the data of that file in HDFS? If so, then it could be an
>> >> xciever problem
>> >> (http://wiki.apache.org/hadoop/Hbase/Troubleshooting#A5). If not, then
>> >> there could be a double assignment issue and restarting the cluster
>> >> would take care of it (since it's only a test env).
>> >>
>> >> Seeing that you aren't using a 0.20 release (since we stopped using
>> >> mapfiles in 0.20), I can only recommend upgrading to 0.20.5
>> >>
>> >> J-D
>> >>
>> >> On Mon, Jul 12, 2010 at 2:36 AM, Yabo Xu <[email protected]>
>> wrote:
>> >> > Hi there:
>> >> >
>> >> > On an internal testing cluster with 3 nodes, when I run "flush '.META'
>> "
>> >> on
>> >> > the hbase shell, it gets the following "Can not obtain block" error. I
>> >> > checked around, and many posts say that it might be due to the crash
>> of
>> >> some
>> >> > datanodes. But in my case, i checked the UI, all nodes appears to be
>> >> fine.
>> >> > Any other possibilities?
>> >> >
>> >> > Error details pasted below. Any help is appreciated!
>> >> >
>> >> > Best,
>> >> > Arber
>> >> >
>> >> > hbase(main):001:0> flush '.META'
>> >> > 10/07/12 17:29:30 WARN client.HConnectionManager$TableServers: Testing
>> >> for
>> >> > table existence threw exception
>> >> > org.apache.hadoop.hbase.client.RetriesExhaustedException: Trying to
>> >> contact
>> >> > region server null for region , row '', but failed after 5 attempts.
>> >> > Exceptions:
>> >> > java.io.IOException: java.io.IOException: Could not obtain block:
>> >> > blk_-80326634570231114_202750
>> >> > file=/hbase/-ROOT-/70236052/info/mapfiles/3687060941742211902/data
>> >> >    at
>> >> >
>> >>
>> org.apache.hadoop.hdfs.DFSClient$DFSInputStream.chooseDataNode(DFSClient.java:1707)
>> >> >    at
>> >> >
>> >>
>> org.apache.hadoop.hdfs.DFSClient$DFSInputStream.blockSeekTo(DFSClient.java:1535)
>> >> >    at
>> >> >
>> org.apache.hadoop.hdfs.DFSClient$DFSInputStream.read(DFSClient.java:1662)
>> >> >    at
>> >> >
>> org.apache.hadoop.hdfs.DFSClient$DFSInputStream.read(DFSClient.java:1592)
>> >> >    at java.io.DataInputStream.readInt(DataInputStream.java:370)
>> >> >    at
>> >> >
>> >>
>> org.apache.hadoop.hbase.io.SequenceFile$Reader.readRecordLength(SequenceFile.java:1909)
>> >> >    at
>> >> >
>> >>
>> org.apache.hadoop.hbase.io.SequenceFile$Reader.next(SequenceFile.java:1939)
>> >> >    at
>> >> >
>> >>
>> org.apache.hadoop.hbase.io.SequenceFile$Reader.next(SequenceFile.java:1844)
>> >> >    at
>> >> >
>> >>
>> org.apache.hadoop.hbase.io.SequenceFile$Reader.next(SequenceFile.java:1890)
>> >> >    at org.apache.hadoop.hbase.io.MapFile$Reader.next(MapFile.java:544)
>> >> >    at
>> >> >
>> >>
>> org.apache.hadoop.hbase.regionserver.HStore.rowAtOrBeforeFromMapFile(HStore.java:1723)
>> >> >    at
>> >> >
>> >>
>> org.apache.hadoop.hbase.regionserver.HStore.getRowKeyAtOrBefore(HStore.java:1695)
>> >> >    at
>> >> >
>> >>
>> org.apache.hadoop.hbase.regionserver.HRegion.getClosestRowBefore(HRegion.java:1089)
>> >> >    at
>> >> >
>> >>
>> org.apache.hadoop.hbase.regionserver.HRegionServer.getClosestRowBefore(HRegionServer.java:1555)
>> >> >    at sun.reflect.GeneratedMethodAccessor6.invoke(Unknown Source)
>> >> >    at
>> >> >
>> >>
>> sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:25)
>> >> >    at java.lang.reflect.Method.invoke(Method.java:597)
>> >> >    at
>> org.apache.hadoop.hbase.ipc.HBaseRPC$Server.call(HBaseRPC.java:632)
>> >> >    at
>> >> >
>> org.apache.hadoop.hbase.ipc.HBaseServer$Handler.run(HBaseServer.java:912)
>> >> >
>> >>
>> >
>>
>

Reply via email to