jackrabbit-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Bouchet Jean-Luc" <jean-luc.bouc...@sib.fr>
Subject Some errors happen with a JackRabbit 1.4.2 clustered WebDav Server and a frontal APACHE web server/ load balancer
Date Thu, 27 Mar 2008 17:59:56 GMT
 My purpose is to obtain an high availability solution for document
management with JackRabbit and APACHE web server in active/active mode.
I'm trying to have à Jackrabbit webdav server in cluster mode ( 2 nodes)
using APACHE as frontal web server with mod_jk and APACHE load balancer.
The load balancing between my too cluster nodes works fine with APACHE web
But when I try to make some "intensive" ( 4 webdav clients threads) calls to
APACHE frontal  (instead of jackrabbit's http frontal connector), it happens
that some PROPFIND calls return in 404 errors. 
I finally understand that the problem was caused by the Jackrabbit cache (or
caches). It seems that the synchronisation of clusters cache with the
journal works asynchronously. PROPFIND calls (for example after a MKCOL)
fail when my node's cache is not refreshed quikly enough (MKCOL on node1 and
PROPFIND for the same entity but APACHE web server load balancer sends it to
404 errors dont't happen if i put a very very short syncDelay
(syncDelay="1") in reposityory.xml, but my ORACLE database server is used
too CPU intensively (probably because of journal's calls in database). 
Does somebody know a solution to bypass or to solve this problem? Can I turn
off jackrabbit's cache for example ? I don't find anything about it in the
JL Bouchet

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