| Summary: | memory leak when doing successive redeploy of webappContext due to not invalidated session in server | ||
|---|---|---|---|
| Product: | [RT] Jetty | Reporter: | Ludovic Praud <ludovic.praud> |
| Component: | server | Assignee: | Greg Wilkins <gregw> |
| Status: | RESOLVED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | P3 | CC: | gevatter.tod, jesse.mcconnell, jetty-inbox, ludovic.praud |
| Version: | 7.4.1 | ||
| Target Milestone: | 7.2.x | ||
| Hardware: | PC | ||
| OS: | Linux | ||
| Whiteboard: | |||
|
Description
Ludovic Praud
if your embedding jetty for your unit tests you could call the invalidate yourself just a thought if this is a major issue for you, or are you using jetty in another way for this sort of testing? It is not a major issue for me because I found a workaround by changing the server sessionIdManager before starting another webappContext : server.setSessionIdManager(new HashSessionIdManager()); I agree this is not the way I use jetty on production environment but still, it would be cool to prevent such OOM because it always cost time to find where it come from. I've reworked the hash session managers to a) always remove the sessions on doStop b) use weak references from the session id manager (just in case). This is checked in at r3123. I would really appreciate if you could test this ASAP so we can see if it can go into 7.4.1 I've tested 7.4.1-SNAPSHOT, removing my workaround before of course. It works great. I stressed it a little and every thing work smoothly. The webappContext loaded classes are well garbaged each time the webappContext is undeployed. You were really fast, thanks. Just for making some advertizing, I went through this leak when working with this plugin http://code.google.com/p/mycila/wiki/PluginJetty which uses jetty as embedded container for webapp integration testing. *** Bug 336449 has been marked as a duplicate of this bug. *** |