On Thu, Aug 18, 2022 at 07:19:38PM +0100, Pádraig Brady wrote:
> On 18/08/2022 17:53, Rodolfo Aramayo wrote:
> > The result of df -T is:
> > Type    1K-blocks    Used  Available Use% Mounted on
> > ext4 Â Â Â Â 60782776 Â 10993500 Â Â 49772892 Â 19% /
> > ext4 Â Â 10401954720 4945781880 Â 4931868456 Â 51% /media/volume/sdb
> >
> > I am able to create more files in the same directory and in any other
> > directory on that drive
>
> Interesting.
>
> >
> > And again, the result of the df -i is:
> >
> > Inodes   IUsed   IFree     IUse%  Mounted on
> > Â 7741440 Â Â 312627 Â Â 7428813Â 5%Â Â Â Â Â /
> > 327680000Â 23300224Â 304379776Â 8%Â Â Â Â Â /media/volume/sdb
> >
> > I do not see any evidence of having run out of inodes on
> > '/media/volume/sdb' where the process was run
> > Unless csplit requires the 'root' partition inodes??
>
> I'm fairly sure the issue is not in csplit,
> as it's just doing open(), write(), close().
>
> I would do an `ef2sck -f` on the unmounted file system if possible.
>
> I would do the run on another (file) system if possible.
>
> cheers,
> Pádraig.
Maybe run the various df's periodically alongside in another terminal.
So you can see what grows.
You could use 'watch' for that or write a simple script.
Include a 'date' command in this. And also put a 'date' after the csplit.
This might help correlate the output.
Include both filesystems in the df's.
Succes!
--
Regards, Mike Jonkmans
It was all so different before everything changed.