The bad data is NULLs (I did not have hexedit, but was introduced to hexedit mode in emacs which proved useful).
In the meantime, Muli Ben-Yehuda suggested to prevent the mess to begin with. The corrupted file is the output of a C program. The problem is that the program continues writing to the file, but it does not verify that the data is written. In a normal filesystem, I would not care, but mine fails several times a day. The NULLs are empty data, because the program did fseek forward, but the file was not written. The solution I am testing is syncing. The options I got were: 1. to mount the filesystem such that it will always sync, 2. to sync everything the user is running at a certain point, or 3. to fsync just the problematic file, when I stop writing to it. I am currently testing the third option, for one file only. It is likely to hurt the performance the least. On Mon, Jul 20, 2015 at 1:40 PM, Rabin Yasharzadehe <ra...@rabin.io> wrote: > can you provide a example of a bad lines and how do you like them to look > like after you fix them ? > > -- > Rabin > > On Mon, Jul 20, 2015 at 11:56 AM, Orna Agmon Ben-Yehuda < > ladyp...@gmail.com> wrote: > >> Hello everyone, >> >> I often have damaged text files (due to a lovely storage system). The >> files are of different formats, although I can usually assume they contain >> spaces. The files are structured as lines. >> >> Every once in a while, the lovely destruction (ahm....storage) system >> inserts binary garbage to the file. I wish to fix the files by removing the >> cancer without leaving any leftovers. That is, I want to lose partial lines. >> >> I tried using grep with all sorts of keys, but it did not do the trick. >> strings catches too little - it leaves partial lines. >> Is there an elegant way to do the trick line-wise? >> >> Thanks >> Orna >> >> -- >> Orna Agmon Ben-Yehuda. >> http://ladypine.org >> >> _______________________________________________ >> Linux-il mailing list >> Linux-il@cs.huji.ac.il >> http://mailman.cs.huji.ac.il/mailman/listinfo/linux-il >> >> > -- Orna Agmon Ben-Yehuda. http://ladypine.org
_______________________________________________ Linux-il mailing list Linux-il@cs.huji.ac.il http://mailman.cs.huji.ac.il/mailman/listinfo/linux-il
