3

I have a CentOS 6x server and running a Java application on it, the application is using a mount point /javapp which is an EXT4 filesystem.

suddenly I got the following error in the system logs:

kernel: attempt to access beyond end of device
kernel: dm-2: rw=0, want=34359697536, limit=112984064

I found this error may result from one of two possible scenarios:

  1. storage resize (LVM or SAN for example)
  2. corrupter filesystem

All the pointers suggests a filesystem corruption since I didn't change the size of any backing disks.

My question is, does an application able to corrupt a filesystem?

Uwe Keim
  • 2,370
  • 4
  • 29
  • 46
Zach
  • 131
  • 1
  • 1
    I'm not quite sure what's your asking, but if your question is if your application is the cause of the file-system corruption, then the answer is usually no. A write operation will trigger the error, but most frequently the actual cause is faulty hardware and/or memory errors. – HBruijn Jun 28 '16 at 14:13
  • @HBruijn, this is exactly my question -- so, a user space application is incapable of corrupting a filesystem, correct? – Zach Jun 28 '16 at 14:27

1 Answers1

1

It is generally not possible for an application to corrupt a filesystem: normal applications do not directly interact with disks, rather they use kernel-level syscalls for reading and writing.

Obviously, applications designed to interact with block devices (ie: fdisk) can corrupt filesystems, but this does not seems your case.

Can you provide the output of the following commands:

  • lsblk
  • fdisk -l -u <your_device>
  • pvs
  • vgs
  • lvs
  • df -h
shodanshok
  • 44,038
  • 6
  • 98
  • 162