You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@hbase.apache.org by Иван <m1...@mail.ru> on 2008/09/18 20:24:28 UTC

Starting HBase from an old snapshot

The situation is quite simple: I'm just trying to launch an HBase instance from a hbase directory in HDFS which was remaining inactive for a some period of time (in fact the HBase was just running from a different directory on the same HDFS, it was something like a backup). But for some reasons there has been some problems in launching it, which resulted in some exceptions, mostly like this:

2008-09-18 12:35:53,995 INFO org.apache.hadoop.ipc.Server: IPC Server handler 2 on 60020 caught: java.nio.channels.ClosedChannelException
        at sun.nio.ch.SocketChannelImpl.ensureWriteOpen(SocketChannelImpl.java:125)
        at sun.nio.ch.SocketChannelImpl.write(SocketChannelImpl.java:294)
        at org.apache.hadoop.ipc.Server$Responder.processResponse(Server.java:594)
        at org.apache.hadoop.ipc.Server$Responder.doRespond(Server.java:654)
        at org.apache.hadoop.ipc.Server$Handler.run(Server.java:917)

and this:

2008-09-18 12:53:35,097 WARN org.apache.hadoop.hbase.regionserver.HRegionServer: Processing message (Retry: 1)
java.net.SocketTimeoutException: timed out waiting for rpc response
        at org.apache.hadoop.ipc.Client.call(Client.java:559)
        at org.apache.hadoop.hbase.ipc.HbaseRPC$Invoker.invoke(HbaseRPC.java:230)
        at $Proxy0.regionServerReport(Unknown Source)
        at org.apache.hadoop.hbase.regionserver.HRegionServer.run(HRegionServer.java:314)
        at java.lang.Thread.run(Thread.java:595)

The snapshot was made while cleanly stopping of HBase and copying (or maybe moving, I'm not really sure) it's folder to another location in DFS.
Even web interface of HBase have failed to launch, but it seems that that HBase is doing something when it tries to launch: assigning some regions, creating/removing some blocks in HDFS, etc. But finally it fails...

Does anyone have any ideas about the reasons of such a behavior?

P.S.: Hadoop 0.17.1, HBase 0.2.0, Debian Etch

Thanks,
Ivan Blinkov

Re[2]: Starting HBase from an old snapshot

Posted by Иван <m1...@mail.ru>.
Sure, it is.


Thanks,
Ivan Blinkov


-----Original Message-----
From: "Billy Pearson" <sa...@pearsonwholesale.com>
To: hbase-user@hadoop.apache.org
Date: Thu, 18 Sep 2008 13:40:59 -0500
Subject: Re: Starting HBase from an old snapshot

> 
> Was the old directory made form the same version of hbase as the current one 
> you are trying to use?
> 
> Billy
> 
> "Иван" <m1...@mail.ru> wrote in message 
> news:E1KgOB2-0007Or-00.m11-mail-ru@f161.mail.ru...
> > The situation is quite simple: I'm just trying to launch an HBase instance 
> > from a hbase directory in HDFS which was remaining inactive for a some 
> > period of time (in fact the HBase was just running from a different 
> > directory on the same HDFS, it was something like a backup). But for some 
> > reasons there has been some problems in launching it, which resulted in 
> > some exceptions, mostly like this:
> >
> > 2008-09-18 12:35:53,995 INFO org.apache.hadoop.ipc.Server: IPC Server 
> > handler 2 on 60020 caught: java.nio.channels.ClosedChannelException
> >        at 
> > sun.nio.ch.SocketChannelImpl.ensureWriteOpen(SocketChannelImpl.java:125)
> >        at sun.nio.ch.SocketChannelImpl.write(SocketChannelImpl.java:294)
> >        at 
> > org.apache.hadoop.ipc.Server$Responder.processResponse(Server.java:594)
> >        at 
> > org.apache.hadoop.ipc.Server$Responder.doRespond(Server.java:654)
> >        at org.apache.hadoop.ipc.Server$Handler.run(Server.java:917)
> >
> > and this:
> >
> > 2008-09-18 12:53:35,097 WARN 
> > org.apache.hadoop.hbase.regionserver.HRegionServer: Processing message 
> > (Retry: 1)
> > java.net.SocketTimeoutException: timed out waiting for rpc response
> >        at org.apache.hadoop.ipc.Client.call(Client.java:559)
> >        at 
> > org.apache.hadoop.hbase.ipc.HbaseRPC$Invoker.invoke(HbaseRPC.java:230)
> >        at $Proxy0.regionServerReport(Unknown Source)
> >        at 
> > org.apache.hadoop.hbase.regionserver.HRegionServer.run(HRegionServer.java:314)
> >        at java.lang.Thread.run(Thread.java:595)
> >
> > The snapshot was made while cleanly stopping of HBase and copying (or 
> > maybe moving, I'm not really sure) it's folder to another location in DFS.
> > Even web interface of HBase have failed to launch, but it seems that that 
> > HBase is doing something when it tries to launch: assigning some regions, 
> > creating/removing some blocks in HDFS, etc. But finally it fails...
> >
> > Does anyone have any ideas about the reasons of such a behavior?
> >
> > P.S.: Hadoop 0.17.1, HBase 0.2.0, Debian Etch
> >
> > Thanks,
> > Ivan Blinkov
> > 
> 
>

Re: Starting HBase from an old snapshot

Posted by Billy Pearson <sa...@pearsonwholesale.com>.
Was the old directory made form the same version of hbase as the current one 
you are trying to use?

Billy

"Иван" <m1...@mail.ru> wrote in message 
news:E1KgOB2-0007Or-00.m11-mail-ru@f161.mail.ru...
> The situation is quite simple: I'm just trying to launch an HBase instance 
> from a hbase directory in HDFS which was remaining inactive for a some 
> period of time (in fact the HBase was just running from a different 
> directory on the same HDFS, it was something like a backup). But for some 
> reasons there has been some problems in launching it, which resulted in 
> some exceptions, mostly like this:
>
> 2008-09-18 12:35:53,995 INFO org.apache.hadoop.ipc.Server: IPC Server 
> handler 2 on 60020 caught: java.nio.channels.ClosedChannelException
>        at 
> sun.nio.ch.SocketChannelImpl.ensureWriteOpen(SocketChannelImpl.java:125)
>        at sun.nio.ch.SocketChannelImpl.write(SocketChannelImpl.java:294)
>        at 
> org.apache.hadoop.ipc.Server$Responder.processResponse(Server.java:594)
>        at 
> org.apache.hadoop.ipc.Server$Responder.doRespond(Server.java:654)
>        at org.apache.hadoop.ipc.Server$Handler.run(Server.java:917)
>
> and this:
>
> 2008-09-18 12:53:35,097 WARN 
> org.apache.hadoop.hbase.regionserver.HRegionServer: Processing message 
> (Retry: 1)
> java.net.SocketTimeoutException: timed out waiting for rpc response
>        at org.apache.hadoop.ipc.Client.call(Client.java:559)
>        at 
> org.apache.hadoop.hbase.ipc.HbaseRPC$Invoker.invoke(HbaseRPC.java:230)
>        at $Proxy0.regionServerReport(Unknown Source)
>        at 
> org.apache.hadoop.hbase.regionserver.HRegionServer.run(HRegionServer.java:314)
>        at java.lang.Thread.run(Thread.java:595)
>
> The snapshot was made while cleanly stopping of HBase and copying (or 
> maybe moving, I'm not really sure) it's folder to another location in DFS.
> Even web interface of HBase have failed to launch, but it seems that that 
> HBase is doing something when it tries to launch: assigning some regions, 
> creating/removing some blocks in HDFS, etc. But finally it fails...
>
> Does anyone have any ideas about the reasons of such a behavior?
>
> P.S.: Hadoop 0.17.1, HBase 0.2.0, Debian Etch
>
> Thanks,
> Ivan Blinkov
>

Re: Starting HBase from an old snapshot

Posted by stack <st...@duboce.net>.
Ivan:

This should work -- or at least, its worked for me in the past as long 
as the shutdown was clean (as you say your's was).  The errors below 
don't help much.  Perhaps retry with DEBUG enabled and send us over a 
log (See FAQ for how to enable DEBUG).

Thanks,
St.Ack


Иван wrote:
> The situation is quite simple: I'm just trying to launch an HBase instance from a hbase directory in HDFS which was remaining inactive for a some period of time (in fact the HBase was just running from a different directory on the same HDFS, it was something like a backup). But for some reasons there has been some problems in launching it, which resulted in some exceptions, mostly like this:
>
> 2008-09-18 12:35:53,995 INFO org.apache.hadoop.ipc.Server: IPC Server handler 2 on 60020 caught: java.nio.channels.ClosedChannelException
>         at sun.nio.ch.SocketChannelImpl.ensureWriteOpen(SocketChannelImpl.java:125)
>         at sun.nio.ch.SocketChannelImpl.write(SocketChannelImpl.java:294)
>         at org.apache.hadoop.ipc.Server$Responder.processResponse(Server.java:594)
>         at org.apache.hadoop.ipc.Server$Responder.doRespond(Server.java:654)
>         at org.apache.hadoop.ipc.Server$Handler.run(Server.java:917)
>
> and this:
>
> 2008-09-18 12:53:35,097 WARN org.apache.hadoop.hbase.regionserver.HRegionServer: Processing message (Retry: 1)
> java.net.SocketTimeoutException: timed out waiting for rpc response
>         at org.apache.hadoop.ipc.Client.call(Client.java:559)
>         at org.apache.hadoop.hbase.ipc.HbaseRPC$Invoker.invoke(HbaseRPC.java:230)
>         at $Proxy0.regionServerReport(Unknown Source)
>         at org.apache.hadoop.hbase.regionserver.HRegionServer.run(HRegionServer.java:314)
>         at java.lang.Thread.run(Thread.java:595)
>
> The snapshot was made while cleanly stopping of HBase and copying (or maybe moving, I'm not really sure) it's folder to another location in DFS.
> Even web interface of HBase have failed to launch, but it seems that that HBase is doing something when it tries to launch: assigning some regions, creating/removing some blocks in HDFS, etc. But finally it fails...
>
> Does anyone have any ideas about the reasons of such a behavior?
>
> P.S.: Hadoop 0.17.1, HBase 0.2.0, Debian Etch
>
> Thanks,
> Ivan Blinkov
>