zookeeper-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ted Dunning <ted.dunn...@gmail.com>
Subject Re: disk full
Date Thu, 23 Jun 2011 16:32:44 GMT
On Wed, Jun 22, 2011 at 7:15 PM, Donna Li <donna.li@utstar.com> wrote:


>        After I make space, the zookeeper server also can not start
> normally. Following log is printed. It shows "Last transaction was partial".
> I must clear the data (/data/version-2), then the server can restart
> normally.
>

If clearing the data is acceptable, then it sounds like you have an answer.

Can somebody else comment on this error?  Will the server refuse to start
here?

Donna,  is this the only server in the cluster that had disk full?  Or did
all of them have disk full?

If only one had the disk full, then wiping the data or even just the last
log file should be fine since the current data will be recovered from other
members of the cluster.


>    Under the disk full situation, if the cluster is unavailable, the TCP
> connection between the server and the client would break. So when the
> cluster comes back normally, the client can not change to connected status.
>    I want to know if this is the limitation that zookeeper can not meet
> disk full.
>

I don't understand the question.

Are you talking about one server out of several failing?

If so, then having that server come back should cause no problems at all.

Also, your sentence that starts "I want to know ..." is something I don't
quite understand.  Can you say it in different words?


> -----邮件原件-----
> 发件人: Ted Dunning [mailto:ted.dunning@gmail.com]
> 发送时间: 2011年6月11日 14:25
> 收件人: user@zookeeper.apache.org
> 主题: Re: disk full
>
> Zookeeper provides guarantees of durability and reliability.  When the
> disk is full, transactions cannot be persisted to disk and thus these
> guarantees cannot be made.
>
> If you restart ZK after making space, then it should continue
> reliably.  One way to do this is to get rid of old transaction logs
> that are no longer needed.  Another is to plan your needs more
> precisely so conflicts on space do not occur.  Sometimes this is quite
> difficult which is one reason that some people opt for dedicated
> machines for critical services like this.
>
> On Sat, Jun 11, 2011 at 7:04 AM, Donna Li <donna.li@utstar.com> wrote:
> > Hi,all:
> >
> >
> >
> > What is the impact of disk full to zookeeper? Why restart zookeeper can
> > not resolve the problem? I must clear the data of the zookeeper and
> > restart. The print error is:
> >
> >
> >
> >
> >
> > 2011-06-09 11:44:29,204 - INFO  [main:QuorumPeerConfig@90] - Reading
> > configuration from:
> > /usr/local/rss/zookeeper/tool/zookeeper-3.3.2/bin/../conf/zoo.cfg
> >
> > 2011-06-09 11:44:29,211 - WARN  [main:QuorumPeerConfig@266] -
> > Non-optimial configuration, consider an odd number of servers.
> >
> > 2011-06-09 11:44:29,211 - INFO  [main:QuorumPeerConfig@310] - Defaulting
> > to majority quorums
> >
> > 2011-06-09 11:44:29,223 - INFO  [main:QuorumPeerMain@119] - Starting
> > quorum peer
> >
> > 2011-06-09 11:44:29,239 - INFO  [main:NIOServerCnxn$Factory@143] -
> > binding to port 0.0.0.0/0.0.0.0:2181
> >
> > 2011-06-09 11:44:29,251 - INFO  [main:QuorumPeer@818] - tickTime set to
> > 2000
> >
> > 2011-06-09 11:44:29,252 - INFO  [main:QuorumPeer@829] -
> > minSessionTimeout set to -1
> >
> > 2011-06-09 11:44:29,252 - INFO  [main:QuorumPeer@840] -
> > maxSessionTimeout set to -1
> >
> > 2011-06-09 11:44:29,252 - INFO  [main:QuorumPeer@855] - initLimit set to
> > 5
> >
> > 2011-06-09 11:44:29,266 - INFO  [main:FileSnap@82] - Reading snapshot
> > /usr/local/rss/zookeeper/data/version-2/snapshot.7000011ff
> >
> > 2011-06-09 11:44:29,301 - ERROR [main:Util@238] - Last transaction was
> > partial.
> >
> > 2011-06-09 11:44:29,302 - FATAL [main:QuorumPeer@399] - Unable to load
> > database on disk
> >
> > java.io.EOFException
> >
> >        at java.io.DataInputStream.readInt(DataInputStream.java:375)
> >
> >        at
> > org.apache.jute.BinaryInputArchive.readInt(BinaryInputArchive.java:63)
> >
> >        at
> > org.apache.zookeeper.server.persistence.FileHeader.deserialize(FileHeade
> > r.java:65)
> >
> >        at
> > org.apache.zookeeper.server.persistence.FileTxnLog$FileTxnIterator.inStr
> > eamCreated(FileTxnLog.java:508)
> >
> >        at
> > org.apache.zookeeper.server.persistence.FileTxnLog$FileTxnIterator.creat
> > eInputArchive(FileTxnLog.java:527)
> >
> >        at
> > org.apache.zookeeper.server.persistence.FileTxnLog$FileTxnIterator.goToN
> > extLog(FileTxnLog.java:493)
> >
> >        at
> > org.apache.zookeeper.server.persistence.FileTxnLog$FileTxnIterator.next(
> > FileTxnLog.java:575)
> >
> >        at
> > org.apache.zookeeper.server.persistence.FileTxnSnapLog.restore(FileTxnSn
> > apLog.java:145)
> >
> >        at
> > org.apache.zookeeper.server.ZKDatabase.loadDataBase(ZKDatabase.java:197)
> >
> >        at
> > org.apache.zookeeper.server.quorum.QuorumPeer.start(QuorumPeer.java:397)
> >
> >        at
> > org.apache.zookeeper.server.quorum.QuorumPeerMain.runFromConfig(QuorumPe
> > erMain.java:143)
> >
> >        at
> > org.apache.zookeeper.server.quorum.QuorumPeerMain.initializeAndRun(Quoru
> > mPeerMain.java:103)
> >
> >        at
> > org.apache.zookeeper.server.quorum.QuorumPeerMain.main(QuorumPeerMain.ja
> > va:76)
> >
> > 2011-06-09 11:44:29,303 - FATAL [main:QuorumPeerMain@87] - Unexpected
> > exception, exiting abnormally
> >
> > java.lang.RuntimeException: Unable to run quorum server
> >
> >        at
> > org.apache.zookeeper.server.quorum.QuorumPeer.start(QuorumPeer.java:400)
> >
> >        at
> > org.apache.zookeeper.server.quorum.QuorumPeerMain.runFromConfig(QuorumPe
> > erMain.java:143)
> >
> >        at
> > org.apache.zookeeper.server.quorum.QuorumPeerMain.initializeAndRun(Quoru
> > mPeerMain.java:103)
> >
> >        at
> > org.apache.zookeeper.server.quorum.QuorumPeerMain.main(QuorumPeerMain.ja
> > va:76)
> >
> > Caused by: java.io.EOFException
> >
> >        at java.io.DataInputStream.readInt(DataInputStream.java:375)
> >
> >        at
> > org.apache.jute.BinaryInputArchive.readInt(BinaryInputArchive.java:63)
> >
> >        at
> > org.apache.zookeeper.server.persistence.FileHeader.deserialize(FileHeade
> > r.java:65)
> >
> >        at
> > org.apache.zookeeper.server.persistence.FileTxnLog$FileTxnIterator.inStr
> > eamCreated(FileTxnLog.java:508)
> >
> >        at
> > org.apache.zookeeper.server.persistence.FileTxnLog$FileTxnIterator.creat
> > eInputArchive(FileTxnLog.java:527)
> >
> >        at
> > org.apache.zookeeper.server.persistence.FileTxnLog$FileTxnIterator.goToN
> > extLog(FileTxnLog.java:493)
> >
> >        at
> > org.apache.zookeeper.server.persistence.FileTxnLog$FileTxnIterator.next(
> > FileTxnLog.java:575)
> >
> >        at
> > org.apache.zookeeper.server.persistence.FileTxnSnapLog.restore(FileTxnSn
> > apLog.java:145)
> >
> >        at
> > org.apache.zookeeper.server.ZKDatabase.loadDataBase(ZKDatabase.java:197)
> >
> >        at
> > org.apache.zookeeper.server.quorum.QuorumPeer.start(QuorumPeer.java:397)
> >
> >        ... 3 more
> >
> >
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message