Ssh – How to logout users when reboot/halt over ssh, but so that their .bash_history is saved

bashssh

When I reboot my server over SSH, the session just hangs instead of nicely logging me out, so that I can continue to use the current terminal.
I tried the following to fix this: https://bbs.archlinux.org/viewtopic.php?id=50089
But then my bash history isn't saved before rebooting.

Is there a way to nicely log out all users (so that their bash history is saved) before a reboot/halt command? i.e. save bash history and end user sessions before reboot/halt?

OS: Ubuntu Server 11.04

Best Answer

You need to execute something like history -ain each user session. I can't think of a "nice" way to do this but using PROMPT_COMMAND in bash would work, PROMPT_COMMAND is an env variable for a command that is execute every time bash returns to a prompt, i.e. it automatically flushes the history to .bash_history on each command execute. Works but probably not the best solution. Put this in the .bashrc or in a profile.d file (if you are on profile.d compatible distro).

export PROMPT_COMMAND='history -a'

Related Solutions

Linux – SSH session logging

If you are using Linux or FreeBSD, check out sudosh2:

sudosh is an auditing shell filter and can be used as a login shell. Sudosh records all keystrokes and output and can play back the session as just like a VCR. Sudosh2 is a continuation of the development of sudosh.

This will store a log of all TTY input and output (all keystrokes, including backspaces and other control characters). The output can be stored on the local machine, or can be sent to another server using syslog.

You don't state if your engineers are SSHing from a central server, or if they are all connecting from their own workstations. sudosh2 is ideal for a central server, but might be harder to implement on dozens of workstations.

Some others have used rootsh, but I am not familiar with rootsh.

SSH Connection Silently Dies – Troubleshooting Guide

Check the timeout settings on the router at the server side (the system you're connecting to). I usually run into the 5-minute delay as a result of the default settings on Sonicwall firewalls. In these cases, I'll make the following changes on the ssh server IF I don't have access to correct this on the firewall side.

In /etc/ssh/sshd_config, I add or make sure the following directives are set:

TCPKeepAlive yes
KeepAlive yes
ClientAliveInterval 60

Restart the ssh daemon, and the issue should go away.

You can also make the change on your client system. This entails adding ServerAliveInterval 60 to /etc/ssh/ssh_config and reconnecting or adding -o ServerAliveInterval 60 to your ssh connection string.

Best Answer

Related Solutions

Linux – SSH session logging

SSH Connection Silently Dies – Troubleshooting Guide

Related Topic