bugGluster - Bugs: bug #25681, zero byte files created with...

 
 

bug #25681: zero byte files created with self-healing, data not healed

Submitter:  John Leach <johnleach>
Submitted:  Mon 23 Feb 2009 05:45:47 PM UTC
   
 
Category:  GlusterFS Severity:  3 - Normal
Priority:  5 - Normal Item Group:  Improper behaviour
Status:  Fixed Privacy:  Public
Assigned to:  None Originator Name:  * John Leach
Open/Closed:  Closed Release:  2.0.0qa17
Operating System:  GNU/Linux Reproducibility:  Every Time
* Mandatory Fields

Add a New Comment Rich Markup
   

Jump to the original submission

Thu 25 Jun 2009 02:05:23 PM UTC, comment #8: 

Fixed in 2.0.1 even.

Thanks!

John Leach <johnleach>
Tue 23 Jun 2009 07:56:50 PM UTC, comment #7: 

Please test with latest glusterfs (2.0.2) and re-open the bug, if issue is still seen.

--
Gowda

Basavanagowda Kanur <gowda>
Group Member
Tue 03 Mar 2009 03:12:47 PM UTC, comment #6: 

I can't. I have set up AFR, started only second server and client. First write to a new file crashed the server.

Configuration & server log attached. Again 32-bit ubuntu on physical machine. But this probably is a separate issue.

(file #17568, file #17569, file #17570, file #17571)

-Deleted Account- <findepi>
Tue 03 Mar 2009 01:50:09 PM UTC, comment #5: 

Piotr, can you reproduce this with rc3? (just noticed it in the QA released folder, released yesterday:

http://ftp.gluster.com/pub/gluster/glusterfs/qa-releases/

I'll be testing it myself, but I don't think I'll have time this week.

John Leach <johnleach>
Tue 03 Mar 2009 01:44:46 PM UTC, comment #4: 

I reproduced the bug exactly as you describe with 2.0.0rc2.
To get my behavior (file size replicated, file contents not) you may need to umount the client and mount again.

(Reproduced on 32-bit ubuntu on physical machine.)

-Deleted Account- <findepi>
Tue 03 Mar 2009 10:54:12 AM UTC, comment #3: 

I see. Maybe this is stupid, but what you mean by "0 bytes in size"?
If `du` reports 0, try `hexdump -C`. In 2.0.0rc2 I got replication of file size and meta data to work but NOT of file contents.

If you want just use it, try 1.4.0qa92 version downloadable from http://ftp.gluster.com/pub/gluster/glusterfs/qa-releases/ , it seems to work for me.  But developers would certainly appreciate your feedback for 2.0.0 series.

-Deleted Account- <findepi>
Tue 03 Mar 2009 10:41:15 AM UTC, comment #2: 

Piotr, I was using the ls flags "ls -lah" to list the files. I also read the entire file multiple times in the hope that would trigger replication.

John Leach <johnleach>
Tue 03 Mar 2009 08:23:35 AM UTC, comment #1: 

To force full replication (not only directory entry), you need to "ls -l" in the directory containing file, "ls -lR" anywhere above or any other operation touching the file (open, read).

However, in my scenario the above operations work only little better...

-Deleted Account- <findepi>
Mon 23 Feb 2009 05:45:47 PM UTC, original submission:  

I'm having trouble with self-healing.  I have a client cluster/replicate volume with two bricks as subvolumes.  Writes are replicated to both bricks.

The problem occurs when I shut down one brick (let's say remote2), create a new file and start the brick back up again.  Making a new lookup of the directory causes the self-heal to occur, and the files are created on remote2, but the files are 0 bytes in size.  Reads from the files work fine (assuming they're coming from remote1).

So the process to reproduce is:

start both servers, mount on /srv/share

mkdir /srv/share/test
cd /srv/share/test
cp /boot/memtest86+.bin test1

# all ok up to here

shutdown glusterfsd on remote2
cp /boot/memtest86+.bin test2
startup glusterfsd on remote2
cd .. && cd test # to force a new lookup

# test2 is created 0 bytes in size on remote2's filesystem


I've attached my server and client configs, and the debug output from the gluster client process, logging the above session from start (mkdir /srv/share/test) to end.

John Leach <johnleach>

 

(Note: upload size limit is set to 16384 kB, after insertion of the required escape characters.)

Attach Files:
   
   
Comment:
   

Attached Files
file #17568:  glusterfs-server-volume1.vol added by findepi (442B - application/octet-stream)
file #17569:  glusterfs-server-volume2.vol added by findepi (442B - application/octet-stream)
file #17570:  glusterfs-client.vol added by findepi (706B - application/octet-stream)
file #17571:  glusterfsd.log added by findepi (3KiB - application/octet-stream)
file #17515:  glusterfs.vol added by johnleach (465B - application/octet-stream)
file #17516:  glusterfsd.vol added by johnleach (324B - application/octet-stream)
file #17517:  glusterfs-client.log added by johnleach (46KiB - text/x-log)

 

Depends on the following items: None found

Items that depend on this one: None found

 

Carbon-Copy List
  • -email is unavailable- added by gowda (Posted a comment)
  • -email is unavailable- added by findepi (Posted a comment)
  • -email is unavailable- added by johnleach (Submitted the item)
  • -email is unavailable- added by johnleach
  •  

    There are 0 votes so far. Votes easily highlight which items people would like to see resolved in priority, independently of the priority of the item set by tracker managers.

    Only logged-in users can vote.

     

    Follow 11 latest changes.

    Date Changed by Updated Field Previous Value => Replaced by
    2009-06-23 gowda StatusNone Fixed
        Originator Name John Leach
        Open/ClosedOpen Closed
    2009-03-03 findepi Attached File- Added glusterfs-server-volume1.vol, #17568
        Attached File- Added glusterfs-server-volume2.vol, #17569
        Attached File- Added glusterfs-client.vol, #17570
        Attached File- Added glusterfsd.log, #17571
    2009-02-23 johnleach Attached File- Added glusterfs.vol, #17515
        Attached File- Added glusterfsd.vol, #17516
        Attached File- Added glusterfs-client.log, #17517
        Carbon-Copy- Added johnleach

    Back to the top

    Powered by Savane 3.13-02a9.
    Corresponding source code