# Block Rules / Log-Based Detection
There's no effective or rather gapless way to detect attacks that use log4shell due to the many ways to obfuscate the strings.
Don't put too much trust in any filter/detection pattern. All can be bypassed.
..
2/ # Behaviour Based Detection
We thought about network based detection, but it could be any remote port and any remote system. Java can have many legitimate outgoing connections & often has suspicious sub processes.
3/ # Vulnerability Detection
It's difficult to find vulnerable software. It could be the web app, the ticket mgmt that receives contact form content or the backup software. Vuln scanners won't give a complete picture.
Discovery could take months.
Try to use the Canary Tokens.
4/ # Patching
Since it's difficult to determine the vulnerable applications, it'll be difficult to patch what you don't know.
5/ # Post-Exploitation / Compromise Assessments
My best hope is that we can find enough evidence after a successful exploitation. Attackers have goals and try to persist their access on a compromised systems. Current tools and frameworks should help us with that.
6/ # Final thoughts
I hope that we can at least filter and detect the simple attempts and prevent some damage.
I wish you all the best in the coming months.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
I’d like to clarify my position on #Microsoft in general
Many things have improved over the last 10 years .. a lot .. especially with Windows 10/2016.
Today many fellow security researchers that I highly respect work there.
I criticize Microsoft’s response to recent ..
vulnerabilities (or design flaws) because I care about these things and believe that customers do care too.
I don’t think that it is fair / right to tell them to migrate to the cloud-based solution in order to get rid of these issues.
There are still few but good reasons ..
.. not to opt for the cloud.
I strongly believe that weaknesses in default configs that allow an attacker to escalate privs to Domain Admin should be addressed with a KB patch and not just a pointer to an advisory.
Many won’t read it.
1/x A #COVID19#OffTopic thread for my followers in countries that still enjoy the quiet before the storm.
It is serious. Don't listen to the voices that play it down.
But also don't panic.
The problem with SARS-CoV-2 is that the treatment of severe cases (~5-10%) require ..
2/x .. intensive care beds with respirators.
Here in Germany, we have 29k intensive care beds, most of them occupied long before COVID19.
If only 1% of the citizens get sick, that would be 830k citizens, 83k of them with the severe clinical course of the disease.
3/x I guess you can imagine what that means.
Italy is about 10 days ahead of us.
Doctors in Italy decide every morning in a so called "triage" who gets a bed with lung ventilator and who doesn't, which is basically a death sentence.
These patients slowly suffocate.
Log Sources Top 5
(ordered by cost-benefit ratio / volume > detectable threats)
1. Antivirus 2. Windows Eventlog (+Sysmon) 3. Proxy 4. Firewall 5. DNS
1/ I‘ll give some short comments to help you understand the order
In general: I included only those logs that can already be collected in most organizations, when you start a SecMon project.
Bro/Zeek, Suricata, Netflow, etc. would be somewhere between 2 and 4 if available. ..
2/ Some logs are more difficult (cost/effort) to tap into.
e.g. Antivirus logs can often be collected from a single console, while NSM requires high speed network Taps on mirroring ports in central locations (💵). If you have the budget and time, NSM is worth the effort.