oops... to many files and can't clean up...
John (EBo) David
plug-discuss@lists.plug.phoenix.az.us
Thu, 20 Jun 2002 20:34:43 -0700
sorry to reply to my own message, but a quick update...
I went to another xterm and tried get a second look at things. From
there I was able to see and remove the rest of the files (no idea what
happend to that one xterm to muck with ls/rm). Anyway, things are
cleaned up, and that even seemed to fix things in the original xterm
that was causing the problems...
I may well still reboot and FSCK... Any idea what could have screwed
up? I remember that there used to be a limit of 10,000 files/directory
or inode. That is why I was originally concerned with having more than
2.5 times that in a single directory.
EBo --
"John (EBo) David" wrote:
>
> ummm....
>
> I have a unit and regression test suite for my ecological modeling
> virtual machine. I needed to bump up one of the tests to run for a
> longer time for model testing. Problem was that I forgot that I am
> creating an image dump for *every* variable specified each and every
> iteration... start_time=0, stop_time=25, dt=0.01... that is 2,500 images
> for umm... looks like 8 variables, and there are 15 other unit tests...
>
> So now I find that I have over 25,000 files in a single directory.
> oops. Ok, off to clean them up....
>
> First, ls and rm complain that there are to many files to "rm *.pgm", so
> I go though and delete them by group name... ok, appears to go ok. Now
> I am finally able to "rm *.pgm" so they should be clean up. Problem is
> that once I do that I still have hundreds of pgm files in the directory
> that "ls" reports, but an "ls *meta_pop*" does not. I am affraid that I
> have corrupted the file system or something.
>
> any suggestions?
>
> thoughts:
>
> shut down the machine, reboot single user, fsck ever partition
> (including XFS partions), and recite some prayer to Boolean...
>
> other ideas, thoughts, intuitions as to what happens when creating 10's
> of thousands of files in a single directory by accident?
>
> EBo --