This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.

Bug 188435 - java.io.IOException: wrong path at org.netbeans.modules.masterfs.watcher.WindowsNotifier.addWatch(WindowsNotifier.java:303)
Summary: java.io.IOException: wrong path at org.netbeans.modules.masterfs.watcher.Win...
Status: RESOLVED INVALID
Alias: None
Product: platform
Classification: Unclassified
Component: Filesystems (show other bugs)
Version: 6.x
Hardware: PC Windows 7
: P2 normal with 2 votes (vote)
Assignee: Jaroslav Tulach
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2010-07-07 09:31 UTC by aDaneInSpain
Modified: 2010-12-08 06:35 UTC (History)
3 users (show)

See Also:
Issue Type: DEFECT
Exception Reporter:


Attachments
Console dump (14.64 KB, text/plain)
2010-07-16 09:54 UTC, aDaneInSpain
Details
Console dump as of Build 20101014000 (51.62 KB, text/plain)
2010-10-14 18:08 UTC, XyploatKyrt
Details
\.netbeans\dev\var\log\messages.log (112.53 KB, application/octet-stream)
2010-10-14 18:39 UTC, XyploatKyrt
Details
Message log 201012070001 (26.36 KB, text/plain)
2010-12-07 10:56 UTC, XyploatKyrt
Details

Note You need to log in before you can comment on or make changes to this bug.
Description aDaneInSpain 2010-07-07 09:31:18 UTC
When using a "code template" while the IDE is "scanning projects" the IDE stops responding and never completes the project scanning. The only solution appears to be to use the task manager to kill the IDE and restart it.

This bug has been replicated several times on two separate installations/PCs and appears to only happen on Windows 7.
Comment 1 Peter Pis 2010-07-16 09:30:34 UTC
Could you please provide a thread dump in such a case? Thanks. 
http://wiki.netbeans.org/GenerateThreadDump
Comment 2 aDaneInSpain 2010-07-16 09:54:53 UTC
Created attachment 100878 [details]
Console dump

As requested here is the console dump
Comment 3 Peter Pis 2010-07-16 13:10:41 UTC
Reassigning for further evaluation.
Comment 4 aDaneInSpain 2010-08-27 09:30:41 UTC
This bug is really annoying and more or less renders Netbeans unusable for anyone using code templates on Windows 7. Our organisation is considering changing editor because of this, which I really want to avoid. What is a typical time span for fixing a bug?
Comment 5 Tomas Zezula 2010-08-30 09:35:13 UTC
The stack trace is not complete the "Parsing & Indexing Loop" is missing among the threads. This is the thread which holds the lock. But the problem is in CSL org.netbeans.modules.csl.editor.codetemplates.GsfCodeTemplateProcessor.initParsing which calls Parsing API blocking call in AWT.
Comment 6 aDaneInSpain 2010-08-30 09:49:24 UTC
Let me know if there is any further information I can help gather.
Comment 7 Tomas Zezula 2010-08-30 13:06:25 UTC
Fixed jet-main 7230951beca3
Comment 8 Quality Engineering 2010-08-31 03:22:00 UTC
Integrated into 'main-golden', will be available in build *201008310001* on http://bits.netbeans.org/dev/nightly/ (upload may still be in progress)
Changeset: http://hg.netbeans.org/main/rev/7230951beca3
User: Tomas Zezula <tzezula@netbeans.org>
Log: #188435:Netbeans stops responding when using a code template while scanning projects on Windows 7
Comment 9 aDaneInSpain 2010-09-15 09:16:42 UTC
This issue is not fixed. Netbeans still crashes in Nightly Build 201009150000 when using a code template while project scanning.
Comment 10 David Strupl 2010-10-14 14:46:54 UTC
Without new thread dump + .netbeans/<version>/var/log/messages.log we cannot do much with this report. Please reopen only after adding the requested information. Thanks.
Comment 11 XyploatKyrt 2010-10-14 18:08:10 UTC
Created attachment 102415 [details]
Console dump as of Build 20101014000

Like aDaneInSpain I've been experiencing this since I moved to Windows 7 just under a year ago. This is from my office computer, but the same happens with any version of Netbeans 6.x up to now on my home computer with a 64 bit Windows and JDK installed so the problem appears to be quite widespread.
Comment 12 XyploatKyrt 2010-10-14 18:39:09 UTC
Created attachment 102417 [details]
\.netbeans\dev\var\log\messages.log

Complete messages.log as requested
Comment 13 David Strupl 2010-10-14 19:06:19 UTC
Thanks for the information. I can start looking at it now.
Comment 14 David Strupl 2010-10-14 19:10:34 UTC
The attached messages.log file is rather crucial for this problem. Passing to the correct team that can do something about it.
Comment 15 Petr Nejedly 2010-10-15 10:03:42 UTC
It seems you're using UNC paths to access your files. This is unsupported by the filesystem watches (Windows themselves refuse to register UNC paths for listening, while happily listening for the same files mapped through a drive letter).
You can solve your problem either by moving the files on a local drive (which could improve the IDE performance tremendously, the log also contains cases where the IDE complained about slow FS access) or at least mapping the server as a drive letter and opening the projects that way.

Still, this issue is valid, the IDE should both complain politely and just ignore such files from FS watching.
Comment 16 Jaroslav Tulach 2010-12-06 12:36:12 UTC
I am not sure you can attribute the original problem to IOException("wrong path"). First of all, this exception is only informational (it was always and after ergonomics#536aa1e515f2). Moreover the original report appeared way sooner than Petr did any native FS listener changes.

Passing back.
Comment 17 David Strupl 2010-12-06 16:16:17 UTC
I apologize for the ping-pong but please reconsider when reading the last thread dump where we clearly call
        at org.netbeans.modules.masterfs.filebasedfs.fileobjects.FolderObj.getChildren(FolderObj.java:137)
        at org.netbeans.modules.parsing.impl.indexing.FileObjectCrawler.collect(FileObjectCrawler.java:179)

It is no longer a deadlock or something but it takes forever to finish this due to the exceptions - if you look where you catch the exception and what you do after you will see that there is something very time consuming going on (recursive descent of some kind) ....
Comment 18 Jaroslav Tulach 2010-12-07 10:17:54 UTC
OK, if it is not a deadlock, but something is "just" slow, I need profiling snapshot:
http://wiki.netbeans.org/FitnessViaPartnership
Comment 19 XyploatKyrt 2010-12-07 10:56:07 UTC
Created attachment 103662 [details]
Message log 201012070001
Comment 20 XyploatKyrt 2010-12-07 10:56:50 UTC
For what it's worth, this issue can be reproduced in all versions of Netbeans PHP I've tried so far. Here's a dump from build 201012070001.

It appears to happen when using Netbeans on a computer running any version of Windows 7, accessing a project stored on a server shared with Samba on CentOS or using Windows XP's built-in file sharing (obviously it affects more configurations but these are the ones I've verified through my own experience).

It doesn't matter if you use \\windows\network\path or map the path as a "Network Drive", the issue is the same. 

When the application locks up, the window goes black and I have left it for over an hour before on a project with many files after which I ended up killing the .exe from the Task Manager.
Comment 21 Quality Engineering 2010-12-08 06:35:02 UTC
Integrated into 'main-golden', will be available in build *201012080001* on http://bits.netbeans.org/dev/nightly/ (upload may still be in progress)
Changeset: http://hg.netbeans.org/main/rev/536aa1e515f2
User: Jaroslav Tulach <jtulach@netbeans.org>
Log: #188435: Use only informational severity level. Native FS listener need not be supported all the time.