Community
Participate
Working Groups
Build Identifier: 20100617-1415 Importing from General -> File System hangs and becomes unresponsive on Linux. Occurs on Ubuntu 9.04 Reproducible: Always Steps to Reproduce: 1. File -> Import 2. Select General - > File System 3. Select any File 4. Select Finish
Please provide backtraces of eclipse hung: http://wiki.eclipse.org/How_to_report_a_deadlock What version of Eclipse core.resources is this?
I'm not able to see the problem locally. Does it also happen on Windows or Mac? Note that this may be a problem in UI, not Resources.
Are there recursive links in the folder? There was a bug fixed in 3.6 for that. Can you reproduce this in the recent 3.7 builds?
Created attachment 188360 [details] kill -3 dump Stack dump.
core.resources version is 20100526-0737 ... I now believe this is limited to importing from a symbolic-linked directory pointing to a mounted file system... Specifically, my situation is importing from this directory: lrwxrwxrwx 1 megan megan 16 2011-02-01 17:57 mhome -> /media/psf/Home/ Where... [~]$ mount [...] none on /media/psf type prl_fs (rw,nosuid,nodev,sync,noatime,share) [...] And yes, the mounted file system is a Parallels filesystem (this particular machine is a Parallels virtual machine). However, this same hang up has occurred on a differing mounted file system with another one of our users (not running in a VM). I will get more information as soon as possible on the particulars of their configuration. I have NOT been able to reproduce this with Indigo, and I have not been able to reproduce with another mounted filesystem type (e.g. NFS).
This is really NOT_ECLIPSE. The hang is happening in java.io.*. We've seen this in the automounter directory when the automounter is under stress (or crashed!). Eclipse may be more stressful on filers than other programs. However a lock up in standard file I/O routines isn't something we can fix.
James, If it's not manifesting in 3.7, using the same JRE (see Megan's last comment above), how can it not be somehow related to Eclipse?
(In reply to comment #7) > If it's not manifesting in 3.7, using the same JRE (see Megan's last comment > above), how can it not be somehow related to Eclipse? Random chance. The top of the stack is part of a call-stack reachable by both versions of eclipse: at java.io.UnixFileSystem.getBooleanAttributes0(Native Method) at java.io.UnixFileSystem.getBooleanAttributes(UnixFileSystem.java:243) at java.io.File.isDirectory(File.java:771) ... Megan's already indicated that there might be some issue with that mount directory. While it could be a JVM issue, more likely it's a OS / Parallels issue. Of course if you think there's something else going wrong in eclipse you'll need to provide more backtraces using jstack or similar. By NOT_ECLIPSE I mean the provided backtrace doesn't implicate eclipse.
Created attachment 188380 [details] Another Stack Dump Alright, I was able to replicate this on a different machine (non-VM), RedHat 5 and on a non-mounted directory. In addition to that, I was also able to unmount the Parallels disk and still get the same problem (in a non-mounted directory) for my VM of Ubuntu. Attached is the prior scenario, nevertheless it does eventually come back up, but it takes awhile (> 2 mins). This was on importing a ~400KB file. It is Tom Crockett's belief that this issue is never a true deadlock and it always recovers, but only after a considerable amount of time, and during this time the UI becomes unresponsive (painting the screen with dialog box).
I would like to add that this issue isn't only on relatively large files. Still occurs on small files (few KB).
Megan, please try a 3.6.1 (or 3.6.2 RC3) build. Unlike Prakash mentioned in comment 3, the important fix went into 3.6.1 and I'm pretty sure that this bug is a duplicate of bug 321505.
I've tried reproducing this issue with 3.6.1 and to the best of my ability cannot. I've replicated the same procedure as I did in 3.6.0 to the T and it works just fine. So I would say that this was in fact a duplicate of the bug you've mentioned, and it appears to have been resolved. I vote to close this and mark as duplicate. I appreciate it.
*** This bug has been marked as a duplicate of bug 334334 ***
This is not a duplicate of bug 334334, it's a duplicate of bug 321505, isn't it?
(In reply to comment #14) > This is not a duplicate of bug 334334, it's a duplicate of bug 321505, isn't > it? Correct. *** This bug has been marked as a duplicate of bug 321505 ***
(In reply to comment #15) > (In reply to comment #14) > > This is not a duplicate of bug 334334, it's a duplicate of bug 321505, isn't > > it? > Correct. Oops! don't know where I got that number from! Thanks Dani for correctly marking it.
Bug 334334 arises under similar conditions to this one (importing a file into the workspace), so it was initially marked as related although it now appears not to be.