We are working on a web project which requires 8 servers. Some are web, some middleware, some database. How to best manage SSH/Security Access?<p>What are the best ways to manage access to ensure security?<p>Team members join in, leave later. How do you manage access at an individual user level? How do you make sure this does not get unwieldy as the no. of servers increase and the team members are transient?<p>Thanks
We've got 9-10 production servers. My methodology is allow what you want and deny everything else. So with iptables specific ports are open for public access and all else is dropped. For complete access to the servers we use openvpn. OpenVPN is setup on one server in the production environment that acts as a gateway to the rest of the servers. Only those with vpn keys can gain access. Each developer is given a specific IP and access is restricted through iptables based on what their requirements are. Knockd a port knocking daemon is used as a way to open a hole in the firewall for 60 seconds to make the connection to openvpn. I have a backdoor incase openvpn or knockd go down. The great thing with openvpn is later on you can disable keys if need be. User accounts are created across the servers for users however ssh keys are only added to the specific servers they may need access to. No passwords, people use insecure passwords, they can be cracked. SSH keys with 2048 bits work nicely.
Use a centralized id solution, such as ldap or kerberos, or a combination of the two. Both MIT and Heimdal kerberos implementations play nicely with OpenSSH. There is also a patch available (somewhere) for OpenSSH to allow one to put their SSH public keys into an LDAP directory.