Bugzilla – Full Text Bug Listing
| Summary: | Very slow startup | ||
|---|---|---|---|
| Product: | [Eclipse Project] Platform | Reporter: | Jerome Lanneluc <jerome_lanneluc> |
| Component: | CVS | Assignee: | platform-cvs-inbox <platform-cvs-inbox> |
| Status: | RESOLVED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | P3 | CC: | andre_weinand, eclipse, jcompagner, jeffmcaffer, pascal, wassim.melhem |
| Version: | 3.0 | ||
| Target Milestone: | 3.0 M9 | ||
| Hardware: | PC | ||
| OS: | Windows 2000 | ||
| Whiteboard: | |||
|
Description
Jerome Lanneluc
Closing the CVS Resources History view before exiting and restarting results in a normal startup time. Pascal, could the fix for mutli-thread plugin activation be slowing us down that much? Each of the following message MESSAGE Threads "Thread[main,6,main]" and "Worker-1" are concurrently trying to start bundle org.eclipse.team.cvs.ui. Thread "Thread[main,6,main]" was trying to load "org.eclipse.team.internal.ccvs. appears once we have waited for 5000ms for a plugin to finish starting up. Therefore it is normal that it slows down the startup. Beyond the fact that it is slow, the message indicates a problem in the startup sequence because two threads are trying to activate the same plugin. This message indicates that the startup of the plugin "org.eclipse.team.cvs.ui" that was initiated by "worker-1" is taking more than 5000ms, and that the main thread (including the UI) is waiting for this guy to be done before continuing. If it was able to continue without warning, you would not be sure of the state in which your plugin would be. Let me know if I can help you. *** Bug 59441 has been marked as a duplicate of this bug. *** Pascal, since you offered, it would be a great help if you could also log the stacktrace of the blocking thread (in our case, the main thread). I know that John did a similar thing for the deadlock case since it is the blocking thread that is much more interesting then the blocked thread. This would make pinpointing the problem a lot easier, especially for race conditions such as this. I suspect this may be related to the hack we added to compensate for the bug before it was fixed in Core. I'll try to reproduce when time permits. Michael, the only thread I can dump the stack trace of is the second thread trying to start the plugin. In this case it is "Main". The fix has been released in HEAD. *** Bug 59577 has been marked as a duplicate of this bug. *** I have removed the code we had to ensure proper plugin loading since that is now handled by core. This should fix the problem. *** Bug 59758 has been marked as a duplicate of this bug. *** *** Bug 59857 has been marked as a duplicate of this bug. *** |