What about commenting out this piece of outdated code in SolrServlet:
} catch (Throwable e) {
SolrException.log(log,e);
sendErr(500, SolrException.toStr(e), request, response);
}For instance, SUN Java 5 not necessarily has resources to output OutOfMemoryError including stack trace; only JRockit can do it... I understand that historically SOLR developers tried to implement full power of HTTP, but let's be more pragmatic...
