You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@jackrabbit.apache.org by Simon Oldham <em...@simonoldham.com> on 2008/06/01 16:41:28 UTC

Can't make clustering work. 24 hours before I have to ditch Jackrabbit.

Hi,

I am working on a project using jackrabbit 1.4 and have run into problems
when trying to set up clustering. I have posted the details here
http://www.simonoldham.com/jackrabbit/clustering-issue/  (so that the
configs are nicely navigable in your browser). I have read the wiki and a
number of newsgroup postings but neither seem to address the problem that I
am encountering.

I have until then end of tomorrow to make clustering work, otherwise we are
going to have to ditch Jackrabbit and develop a bespoke solution (which I
don't want to do having got so far along the line).

I hope you can help.

Thanks in advance,

Si

Re: Can't make clustering work. 24 hours before I have to ditch Jackrabbit.

Posted by Dominique Pfister <do...@day.com>.
Hi Simon,

I wouldn't share the repository homes over NFS; e.g. the search index
might get invalid or even corrupted if more than one cluster node
makes changes simultaneously.

You are already using a shared database (pointing to node1.local:3306)
which is good, so please consider changing the repository homes on
node1 and node2 to some local directory. With this change, your
problems might very well go away.

Kind regards
Dominique

On 6/1/08, Simon Oldham <em...@simonoldham.com> wrote:
> Hi,
>
>  I am working on a project using jackrabbit 1.4 and have run into problems
>  when trying to set up clustering. I have posted the details here
>  http://www.simonoldham.com/jackrabbit/clustering-issue/  (so that the
>  configs are nicely navigable in your browser). I have read the wiki and a
>  number of newsgroup postings but neither seem to address the problem that I
>  am encountering.
>
>  I have until then end of tomorrow to make clustering work, otherwise we are
>  going to have to ditch Jackrabbit and develop a bespoke solution (which I
>  don't want to do having got so far along the line).
>
>  I hope you can help.
>
>  Thanks in advance,
>
>  Si
>

Re: Can't make clustering work. 24 hours before I have to ditch Jackrabbit.

Posted by Sébastien Launay <se...@anyware-tech.com>.
Hi Simon,

You are using MySQL v5 but what MySQL engine are you using ?

 From my experience, i was only able to set up a cluster with
InnoDB engine because MyISAM does not support transaction
and therefore do not lock the journal (which may explains your
error) and i encounter problems with NDBCluster because of read
locks with blobs.

Jackrabbit does not specify the engine to use when creating the
tables so you inherit the default engine.

The default engine can be retrieved on my.cnf in the property
default-table-type.
Using command SHOW TABLE STATUS you can find what engine
is used for each table.

--
Sébastien LAUNAY

Simon Oldham wrote:
> Hi,
>
> I am working on a project using jackrabbit 1.4 and have run into problems
> when trying to set up clustering. I have posted the details here
> http://www.simonoldham.com/jackrabbit/clustering-issue/  (so that the
> configs are nicely navigable in your browser). I have read the wiki and a
> number of newsgroup postings but neither seem to address the problem that I
> am encountering.
>
> I have until then end of tomorrow to make clustering work, otherwise we are
> going to have to ditch Jackrabbit and develop a bespoke solution (which I
> don't want to do having got so far along the line).
>
> I hope you can help.
>
> Thanks in advance,
>
> Si
>
>