Kroon Information Systems
Rant: Things not to do on remote servers

A list of things not to do remotely on servers

This is a list of things that has actually happened to me, one of my colueges or someone else that I know (In no particular order).

  1. On a Gentoo box:
    # vi /var/lib/iptables/rules-save
    # /etc/init.d/iptables reload
     * Flushing firewall...                                                   [ ok ]
    The rest of the output never came. Should have been
     * Loading iptables state and starting firewall...
     * Restoring iptables ruleset                                             [ ok ]
    This resulted in 2 hours on the road for 5 minutes worth of work.
  2. # yes | make oldconfig && make && yes | make install modules_install && reboot
    Some features can break you kernel, and since /boot wasn't mounted the kernel was broken - luckily the old kernel was used since grub was in use and the new kernel got installed in the wrong location (took a while to figure out why the kernel version stayd the same).
  3. # emerge depclean
    Forgetting the -av to this one can be deadly.
  4. emerge packages on a live server with little or no memory can be extremely deadly. I didn't have physical access - took me just over 12 hours to get the server back to an usable state from the point where I realised there is a problem (this was a long night).
  5. # emerge -uav glibc
    # logout

    breaks sshd.
  6. # shutdown -h now
    This happens. Especially when you are working on the local machine and the server simultaniously.
  7. # /etc/init.d/net.eth0 stop
    # ifconfig eth0 down
    Not good.
  8. # init 1
    Not unless net.eth0 or networking or whatever your system refers to for the networking stuff is in runlevel 1.
  9. # killall sshd
    I'll let you imagine the consequences if the server is in a DC that is 60km away ...

Actually I should consider myself lucky that not more things has broken on me or my colueges (There is probably a few that I cannot recall atm).