| Summary: | Session persistence mechanism is not triggered when Virgo restarts | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Product: | [RT] Virgo | Reporter: | Brad Chen <brad5620732> | ||||||
| Component: | tooling | Assignee: | Project Inbox <virgo-inbox> | ||||||
| Status: | CLOSED FIXED | QA Contact: | |||||||
| Severity: | normal | ||||||||
| Priority: | P3 | CC: | eclipse, glyn.normington, milesg78, mlippert | ||||||
| Version: | unspecified | ||||||||
| Target Milestone: | 1.0.1.RELEASE | ||||||||
| Hardware: | All | ||||||||
| OS: | All | ||||||||
| Whiteboard: | |||||||||
| Bug Depends on: | |||||||||
| Bug Blocks: | 368785 | ||||||||
| Attachments: |
|
||||||||
|
Description
Brad Chen
Created attachment 198471 [details] Test web bundle. Test web bundle is attached. The test servlet may be accessed at http://localhost:8080/sess/Test. If the bundle is unloaded first before Virgo is shutdown, the session value would be persisted correctly. Otherwise, no session values would be persisted. Forgot to note that for the test web bundle, the session storage points to /tmp/sess-store.ser. When testing on Windows and that /tmp doesn't exist, first change the pathname in META-INF/context.xml. I tried hard and I cannot reproduce this on M05 or at HEAD. Some clarifications in case we are doing different things: in step 5 I either use the shutdown.sh script or Ctrl-C to shut down Virgo and then I restart it *not* specifying -clean. After the shutdown, I can see the file /tmp/sess-store.ser as expected. The session SID and value printed out by the web app before and after shutdown are identical. I even ran debugged through org.apache.catalina.session.StandardManager to check that the behaviour is correct and I can see it persisting the session data during shutdown. I am running on Mac OS X 10.6.7 with Java 1.6.0_26, although I doubt that is relevant. Sorry for the confusion. After some more testing, session persistence appears to work correctly when Virgo is launched from the command line. When it is started using the Virgo Tooling inside STS, the issue can be reproduced when Virgo is restarted using the Restart command. Therefore, this may be a Virgo Tooling bug. I will attach the Eclipse project for convenience. Created attachment 200003 [details]
Eclipse project to reproduce the bug
Thanks for the clarification. Targetting 3.5 on the assumption that a tooling fix will be needed. I re-tested on the latest tooling snapshot and the problem is no longer recreatable. A number of fixes have gone in since this was last tested, so I presume it is fixed. I tried with VTS 3.0.0.M05 which was what I used to reproduce the problem originally and with VTS 3.6.0.RELEASE. I guess FIXED is more appropriate than WORKSFORME as it definitely used to fail. |