git rebase failing when using ZFS partition on Mac OS X

Alex Blewitt· May 6, 2009, 21:14 UTC · lore

I've found out why git rebase wasn't working for me. It appears to be something to do with the fact that the filesystem is running on ZFS (and frankly, could well be a ZFS bug). I'd be loathe to move away from ZFS generally but it would be good to find out why it's complaining.

apple:egit alex$ git status # On branch master # Your branch is ahead of 'origin/master' by 3 commits. # nothing to commit (working directory clean) apple:egit alex$ git rebase -i HEAD~1 Working tree is dirty

Here's what running a stat on TODO (an arbitrary file at the head of the project) gives on a ZFS mounted file:

apple:egit alex$ stat -r TODO 754974724 135333 0100644 1 1000 100 0 2179 1241639236 1241034917 1241639236 1241034917 2560 5 0 TODO apple:egit alex$ stat TODO 754974724 135333 -rw-r--r-- 1 alex bandlem 0 2179 "May 6 20:47:16 2009" "Apr 29 20:55:17 2009" "May 6 20:47:16 2009" "Apr 29 20:55:17 2009" 2560 5 0 TODO

Here's what running a stat on the same file on an HFS partition gives:

apple:egit alex$ stat -r TODO 234881026 930759 0100644 1 1000 0 0 2179 1241643695 1241034917 1241643690 1241034917 4096 8 0 TODO apple:egit alex$ stat TODO 234881026 930759 -rw-r--r-- 1 alex wheel 0 2179 "May 6 22:01:35 2009" "Apr 29 20:55:17 2009" "May 6 22:01:30 2009" "Apr 29 20:55:17 2009" 4096 8 0 TODO

These are repeatable; if I re-run this periodically, it doesn't seem to change. However, I wonder if there's any dependency on an 'inode' or similar, which doesn't really have a comparable concept in ZFS.

Is there any more information that I can provide to assist with finding out what's going on?

Alex

Robin Rosenberg· May 6, 2009, 23:26 UTC · re: Alex Blewitt · lore

Re: git rebase failing when using ZFS partition on Mac OS X

onsdag 06 maj 2009 23:14:11 skrev Alex Blewitt <alex.blewitt@gmail.com>:

Show 41 quoted lines

> I've found out why git rebase wasn't working for me. It appears to be  
> something to do with the fact that the filesystem is running on ZFS  
> (and frankly, could well be a ZFS bug). I'd be loathe to move away  
> from ZFS generally but it would be good to find out why it's  
> complaining.
> 
> apple:egit alex$ git status
> # On branch master
> # Your branch is ahead of 'origin/master' by 3 commits.
> #
> nothing to commit (working directory clean)
> apple:egit alex$ git rebase -i HEAD~1
> Working tree is dirty
> 
> Here's what running a stat on TODO (an arbitrary file at the head of  
> the project) gives on a ZFS mounted file:
> 
> apple:egit alex$ stat -r TODO
> 754974724 135333 0100644 1 1000 100 0 2179 1241639236 1241034917  
> 1241639236 1241034917 2560 5 0 TODO
> apple:egit alex$ stat TODO
> 754974724 135333 -rw-r--r-- 1 alex bandlem 0 2179 "May  6 20:47:16  
> 2009" "Apr 29 20:55:17 2009" "May  6 20:47:16 2009" "Apr 29 20:55:17  
> 2009" 2560 5 0 TODO
> 
> Here's what running a stat on the same file on an HFS partition gives:
> 
> apple:egit alex$ stat -r TODO
> 234881026 930759 0100644 1 1000 0 0 2179 1241643695 1241034917  
> 1241643690 1241034917 4096 8 0 TODO
> apple:egit alex$ stat TODO
> 234881026 930759 -rw-r--r-- 1 alex wheel 0 2179 "May  6 22:01:35 2009"  
> "Apr 29 20:55:17 2009" "May  6 22:01:30 2009" "Apr 29 20:55:17 2009"  
> 4096 8 0 TODO
> 
> These are repeatable; if I re-run this periodically, it doesn't seem  
> to change. However, I wonder if there's any dependency on an 'inode'  
> or similar, which doesn't really have a comparable concept in ZFS.
> 
> Is there any more information that I can provide to assist with  
> finding out what's going on?

Try mine or Linus patch at http://thread.gmane.org/gmane.comp.version-control.git/117649/ and so what that gives.

-- robin