lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Bowden Wise <wi...@acm.org>
Subject Re: Lock Errors within JBoss Environment
Date Wed, 19 Dec 2012 12:24:31 GMT
Ian, I did see the warnings against using the NativeFSLockFactory on NFS.
I know the the volume is mounted, but I am not sure it is NFS. If I run my
indexing process outside of JBoss I do not see the lock errors, which made
me think it was not using NFS. But I am not sure.

If I don't use NativeFSLockFactory, how does one do "lock-less" commits on
NFS?

Thanks
Bowden

On Tue, Dec 18, 2012 at 3:26 PM, Ian Lea <ian.lea@gmail.com> wrote:

> Is the index on NFS?  There are words in the javadocs warning against
> using NativeFSLockFactory on NFS.
>
>
> --
> Ian.
>
>
> On Tue, Dec 18, 2012 at 8:02 PM, Bowden Wise <wiseb@acm.org> wrote:
> > Hi Andrew:
> >
> > Thanks for the reply; I am glad to here our approach is also being used
> out
> > there.
> >
> > In our case, there is always only 1 writer. I have added debug output to
> my
> > log and can see that the index writer is getting closed each time. And
> yet,
> > I am still seeing the error. I am doing another test run to see what
> > happens.
> >
> > Bowden
> >
> >
> >
> > On Tue, Dec 18, 2012 at 2:13 PM, Andrew Hecox <ahecox@redhat.com> wrote:
> >
> >> On 12/18/2012 01:59 PM, Bowden Wise wrote:
> >>
> >>> Hello,
> >>>
> >>> I have been getting the following lock error when attempting to open an
> >>> index writer to add new documents to an index.
> >>>
> >>> org.apache.lucene.store.**LockObtainFailedException
> >>> Lock obtain timed out:
> >>> NativeFSLock@/opt/shared/data/**CTXTMNG/PAC_INDEX/lucene/aero/**
>  >>> prod/index/write.lock
> >>>
> >>>
> >>> Our ingest/index update process is started via a Quartz scheduler
> process
> >>> that runs daily; the Quartz scheduler is built into the JBoss
> >>> server that the application is installed in. I have seen the related
> posts
> >>> with respect to the NFS file system. However,
> >>> I have tested the same code outside of JBoss on the same file system
> and
> >>> do
> >>> not see the error appear. So I suspect
> >>> that this is not an NFS issue but something particular to the JBoss
> >>> environment or JVM, etc.
> >>>
> >>> Has anyone run a Lucene index update process using Quartz and JBoss?
> >>>
> >>
> >> yes, we run on a very similar stack -- JBoss 4.3, Lucene 3.4, Quartz to
> >> wake up our pollers.
> >>
> >> Not sure I have any words of wisdom, but I did ask the guy who wrote
> that
> >> code and he guessed "either they are having two writers running at the
> same
> >> time, or their writer is not closing the index when it's done writing."
> >>
> >> HTH,
> >>
> >> -Ah
> >>
> >>
> >> Any words of wisdom or best practices?
> >>>
> >>> >
> >>
> >>> Could this be the NFS issue but it manifests itself under JBoss but not
> >>> when running outside?
> >>>
> >>> Lucene version 3.0.1
> >>> JBoss 4.3.0
> >>> Java 1.6
> >>> Sun OS 5.10
> >>> Quartz Scheduler V.1.5.2
> >>>
> >>> Any thoughts appreciated!
> >>> Bowden
> >>>
> >>>
> >>
> >> --
> >> Andrew Hecox
> >> Supervisor, Software Engineering
> >> Subscription Engineering
> >> 没有压力,就没有动力
> >>
> >>
> ------------------------------**------------------------------**---------
> >> To unsubscribe, e-mail: java-user-unsubscribe@lucene.**apache.org<
> java-user-unsubscribe@lucene.apache.org>
> >> For additional commands, e-mail: java-user-help@lucene.apache.**org<
> java-user-help@lucene.apache.org>
>  >>
> >>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
> For additional commands, e-mail: java-user-help@lucene.apache.org
>
>

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