Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

I'd like to be notified via email if our Linux servers are under any kind of hacking attempt or service attack. Is there some kind of all-in-one solution that can monitor for suspicious activity and send reports of said activity to an email address?

share|improve this question

15 Answers 15

up vote 8 down vote accepted

If you're worried about attacks, simple monitoring isn't enough. Imagine an attack comes in at 2am, or whenever you're off the clock and asleep. How many password guesses can be made before you check your email?

Too damn many. fail2ban and other programs will automate the policy you're after. Logwatch can look for unusual behavior, but it's mainly intended for discovering logged errors.

share|improve this answer

Every software package I've used has already been listed, but I just wanted to bring up a point about your second request:

Do you really want to be emailed every time you get hit with an automated attack?

Any time you set something up to notify you instantly via email or sms, ask yourself two questions:

What can I do in response to this notification, and when wont I do that? If you'll respond the same way every time, you should probably be using a trigger system like Fail2Ban (blacklisting IPs for failed logins automatically). If you'll never immediately respond to a notification, then it's probably better as either a daily digest email or just a log in an alert system somewhere.

Keep the signal to noise ratio high in any sort of alert based system. If you get 5-10 emails a day letting you know about some mundane 'attack' that you do nothing about, its a lot easier to let something important slip through the cracks.

share|improve this answer

It is one thing to have a secure server, one that is not easily hackable. So basically, you want to know if the server is under attack, ie, DOS or DDOS.

I'd suggest looking into Google on that subject, or get people's input on that area. There is a lot out there.

share|improve this answer

I use a combination of Denyhosts, Logwatch, and restrictive firewalls on my Linux servers.

First of all, NEVER let Denyhosts send you individual emails for every IP it blocks. If you get hit by a botnet SSH attack (as my personal server has successfully weathered several times), that'll mean way too many notifications. If LogWatch is configured correctly, it will include a report of the denyhosts activity in the daily email. If you keep it at the default of Service = All, this should happen automatically.

Additionally, create a text file called .forward in root's home directory on your server. In it, put the address of where you'd like it to forward root's emails. Even if you've only got a few servers, it's much easier to get all the daily LogWatch messages in one place than it is to log into each one individually and use "Mail".

Finally, on sensitive work machines, I have iptables set up to block all traffic except for IP addresses I've specifically whitelisted. Traffic from the local VLAN is unrestricted, and I have a bash script that builds my iptables rules and whitelists certain IPs based on configuration files. (This is much easier than trying to maintain your whitelists in the script itself.) It's extensible enough for me to create whitelists for any port and/or protocol on the fly as necessary: for example, tcp_22_access contains the IPs for which SSH access is permitted.

share|improve this answer

I'm suprised OSSEC has not been mentioned--It is another Host-Bassed IDS.

ossec.net

Josh

share|improve this answer

Have a look at ossim, it's an OpenSource IDS/Analysis system for *nix server's has event correlation, and connection tracking, and session monitoring, plus gives you a nice overview view of your current security level.

share|improve this answer

I Use Denyhosts. Simple to setup and very efective.

it mails you something like this:

Added the following hosts to /etc/hosts.deny:

144.16.111.105 (unknown)

share|improve this answer

"Samhain is the comprehensive open-source solution for centralized host integrity monitoring." -- http://www.la-samhna.de

share|improve this answer

I think you're looking for snort

It's an intrusion detection system or ids

Takes a bit of configuring and then tuning to minimize false alerts (or to address problems), but there are many tools available to help with that. There are also websites where you can subscribe to new sets of "rules", in order to keep on top of the latest hacks.

Along with that you might want to use snort log analyzers such as base or acid. I think there's also an all-in-one gui solution called sguil as well, that you might want to look at.

share|improve this answer

I use sshdfilter to block ssh attacks and it's setup to send me an email every time an attack is detected. Most of the time, I receive two notifications per day.

share|improve this answer

If you are running with selinux enabled, and you have setroubleshoot running, then any attack that gets past the normal defenses will trigger a popup. I've had one attack make it this far and get shut down by selinux and if it weren't for the alert I would have had no idea. A little research from the alert brought me to the advisory for the package in question, which I didn't need so I uninstalled it.

share|improve this answer

I've heard some good things about Tripwire, never tried it myself though.

share|improve this answer

I do log monitoring with logwatch to look for suspicious login activity. I have the system locked down pretty tight so I mostly "catch" our security office doing their regular scans. There's also an open source version of TripWire that would be useful for monitoring changes to selected files, though this is only useful after they've broken in to let you know what's been compromised.

share|improve this answer

You should consider setting the kernel to monitor for the Evil Bit. I'm not sure about Linux but support is built directly into FreeBSD!

Please check the date on the RFC link before any further consideration of my suggestions...

share|improve this answer
    
We are using evil-bit based filtering with ebtables all the time, works without a hitch. ebtables: evil bit tables, an alternative to iptables –  hayalci May 23 '09 at 21:24

I think you may want to rethink being alerted for every unsuccessful hacking attempt. If you want an e-mail every time someone pokes around looking for a vulnerability, you are probably in for quite a deluge.

Four Linux computers with vulnerable passwords were left online for 24 hours to determine the trend of attacks on them. This was the preparation a researcher made to conduct a study at the University of Maryland. According to his observation, the computers received 270,000 hacking attempts. That comes to one attempt per 39 seconds.

share|improve this answer
    
One would think that you'd start with a reasonably secure configuration that would reject a lot of this traffic without tripping the sensor. For instance, my SSH configuration limits connections to a single network so I never see the login attempts from outside in my log -- they don't make it that far. –  tvanfosson May 1 '09 at 20:10
    
Well the OP did say "any kind of hacking attempt", not just those that have any chance of success. –  Chris Upchurch May 1 '09 at 20:14

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.