0

OS = CentOS 7.3

If I try to run the 'df' command on this server with an option (e.g. -h -l) it hangs and leaves a zombie process. I cannot Ctrl+z out back to a prompt. If I run 'df' against specific mount points that are found in FSTAB the command works successfully (e.g. df /home).

How do I go about troubleshooting this?

5
  • I would start by running your command under strace or ltrace to see where the process is hanging up. Also, is it really a zombie, i.e., it's unkillable from another window?
    – doneal24
    Aug 20, 2018 at 16:12
  • And is it really a Zombie or a process hung in D state (uninterruptible sleep) ? I'd check for a remote filesystem not working anymore too
    – A.B
    Aug 20, 2018 at 16:13
  • 1
    Could you post dmesg and cat /etc/mtab?
    – Bob
    Aug 20, 2018 at 16:28
  • @Bob, what is the best way to post the dmesg and mtab output? Pastebin???
    – Heisenberg
    Aug 20, 2018 at 21:18
  • @Heisenberg Try pastebin.ubuntu.com
    – Bob
    Aug 21, 2018 at 4:36

2 Answers 2

0

This is the symptom of a system call locked on one of your local filesystem:

a local disk who is dying and cause any read or write to loop on retries.

You might be able to sort which of these is your culprit by looking at the system log within /var/log if the root or /var filesystem isn't the origin of your df locked.

0

A reboot resolved the problem. Whether the issue returns or not is yet to be seen. A complex configuration script ran that configured docker containers and made various IPTables changes was executed and no reboot had ever taken place. I will dig a bit deeper and perform and RCA if the issue returns.

I did notice an entry in the dmesg log - Thanks @Bob for the suggestion.

[252869.529789] INFO: task df:14958 blocked for more than 120 seconds. [252869.529838] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .