Mount creates infinite loop directory structure

[GoogleCodeExporter]

What steps will reproduce the problem?
1.Mount a bucket with lots of files,  without using use_cache
2.cd to the mounted directory
3.see that no files exist there, but create a file, and see it exists in both 
the local directory and has also appear in the AWS console
4.umount the directory
5.Mount again using use_cache
6.See that no files exist there, but notice in AWS that a directory with the 
same name as the bucket has been created, and when enter it, it appears to be a 
self-referential link back to the root bucket directory

I then tried to ls the mounted directory, which hung.  I tried to kill, then 
kill -9 the ls process, but it refused to kill it.  I then tried to umount the 
directory, which failed.  I then tried to kill the s3fs process, which turned 
it into a [defunct] state.  Now I have two zombine processes when even a kill 
-9 cannot stop!  And perhaps s3fs is in the background busily creating infinite 
loops of directory structure in my bucket.  I cannot delete the directory in 
AWS console, as it tells me there is an "error".

What is the expected output? What do you see instead?
To see the files that exist in the bucket

What version of the product are you using? On what operating system?
s3fs 1.3.5
fuse 2.8.5
Linux titan.diyaudio.com 2.6.18-164.6.1.el5 #1 SMP Tue Nov 3 16:12:36 EST 2009 
x86_64 x86_64 x86_64 GNU/Linux

Please provide any additional information below.
Appears I will need to restart my box to kill the s3fs process, and contact 
amazon support to assist with clearing my bucket.

Original issue reported on code.google.com by jeremy.c...@gmail.com on 17 Mar 2011 at 8:13

[GoogleCodeExporter]

Sorry, I should have added more here:

What is the expected output? What do you see instead?

I would have expected to see the files that are in the bucket as soon as I 
mounted the bucket.  However they were not there.  It was empty, except for a 
file I created using touch.

When I went to AWS console to verify the touched file exist, I saw that it did, 
but also that a folder existed with the same name as the bucket.  Opening that 
folder, the content of the folder is the same as the contents of the bucket 
root.  Opening the folder of the same name as the bucket, inside the folder 
that has the same name of the bucket, just descends me another level - it 
appears as if a symlink was created with the same name of the bucket, that 
points back to the bucket root.

Original comment by jeremy.c...@gmail.com on 17 Mar 2011 at 8:22

[GoogleCodeExporter]

It appears that the difference is that you didn't use use_cache the first time, 
but you did the second time?

Does it hang if you either use (or don't use) use_cache consistently?

Can you run s3fs from the command line with -f and -d options and share the 
output and syslog entries?

Also, your kernel version is a bit old (relatively speaking) -- I don't know 
for sure, but that might have something to do with it. 

Original comment by dmoore4...@gmail.com on 25 Mar 2011 at 4:44

[GoogleCodeExporter]

The original submitter of this issue needs to respond if any further action 
will be taken on this issue.  ...otherwise this issue will be closed.

Original comment by dmoore4...@gmail.com on 7 Apr 2011 at 2:42

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

I'm sorry but I've abandoned s3fs due to this issue and don't have the time to 
re-visit this.  I've already got many hoops I need to go through to be able to 
remove the directories as even Amazon sales support isn't offering to assist me 
with closing down these renegade looped buckets... 

Best of luck guys. 

Original comment by jeremy.c...@gmail.com on 8 Apr 2011 at 3:14

[GoogleCodeExporter]

Original comment by dmoore4...@gmail.com on 8 Apr 2011 at 3:44

• Changed state: Invalid