[
https://issues.apache.org/jira/browse/SLING-3441?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13923774#comment-13923774
]
Konrad Windszus edited comment on SLING-3441 at 3/7/14 10:52 AM:
-----------------------------------------------------------------
Wouldn't we run into the same problems which are described in SLING-3285?
Currently the old resolvers are not removed quickly enough from the heap due to
JCR-2768. Therefore there is a drawback when creating a lot of resource
resolvers on demand.
was (Author: kwin):
Wouldn't we run into the same problems which are described in SLING-3285.
Currently the old resolvers are not removed quickly enough from the heap due to
JCR-2768. Therefore there is a drawback when creating a lot of resource
resolvers on demand.
> Avoid shared resource resolver usage
> ------------------------------------
>
> Key: SLING-3441
> URL: https://issues.apache.org/jira/browse/SLING-3441
> Project: Sling
> Issue Type: Improvement
> Components: Servlets
> Affects Versions: Servlets Resolver 2.3.0
> Reporter: Carsten Ziegeler
> Assignee: Carsten Ziegeler
> Fix For: Servlets Resolver 2.3.2
>
>
> The servlet resolver uses a single shared rsource resolver to resolve
> scripts. Resource resolvers are not thread safe and therefore should not be
> used at the same time from different threads.
> Apart from that, this creates a bottleneck as the repository implementations
> (Jackrabbit, Oak) synchronize all access in this case - which then ultimately
> synchronises all requests. This is even more problematic as the servlet
> resolver is hit alot during a single request
--
This message was sent by Atlassian JIRA
(v6.2#6252)
