fortunately I do not have MARS to play with but if you did not have to "set" a password for them to use with the expert account I find it very suspect.
When they accessed the box was it through a shared terminal where you were watching the session or did they access it over the network remotely? That you had to log into the system as pnadmin suggests that it was a shared session and then they performed a sudo or su to expert. This would then suggest that the expert account has a fixed password. If it required a local account first that indicates remote access is denied for expert. While this is preferable it is not fool proof. Any vulnerability presenting local shell access could then allow expert access if the password was known. I ask because it would not be the first time a Cisco product had an undocumented account with a default/predictable/easy to guess password. Perhaps someone from Cisco can clarify these points. Brent Stackhouse wrote: > It did cross my mind that there might be a backdoor/default account that > is remotely accessible but TAC said that "expert" access cannot be used > without having an existing, valid account on the system. To reiterate, > per TAC, you cannot simply login to a MARS appliance via SSH or SSL with > the "expert" account. I have not attempted to verify the veracity of > that statement but during the specific support issue I worked with TAC > on, I was instructed to login with the pnadmin account (and a password > known only to me) before TAC could use the expert mode. > > If you have a MARS, go to the CLI and type "expert" - I believe it'll > prompt for a password. > > Part of the point is that a similar issue will happen again which will > require TAC access to the MARS OS and I'm wondering what Cisco's plan is > to deal with that in the future. The MARS manager I spoke with during > this support issue provided this rationale: there is a lot of > easily-accessible intellectual property, due to their use of shell > scripts, Java, etc., that they'd prefer stay obscured. I mentioned that > someone could probably rip out the hard drive and access it anyway but > he said it would still be protected. Um, okay, maybe so and I'm not > really a forensics guy. I just know that this is not a typical Cisco > approach and it caused a major support headache for me and a major client. > > Brent Stackhouse, GSEC/GCIH > VP of Security > Solis Security, Inc. > Austin, Texas > 512-417-9772 > www.solissecurity.com > > Jason wrote: > >>> 3. The MARS OS is a Linux distro but users can't get to the actual >>> OS. This wouldn't normally be a problem but there was a bad MARS >>> build that was published recently, yanked within a day or so, and >>> then required a TAC engineer to remotely login to the MARS box to fix >>> it. This is contrary to every other Cisco device, including >>> Linux-based 42xx IDS/IPS, that I've worked with. >>> >> >> >> Can I read into that statement that there is a some form of capability >> that does allow access to the OS but only to Cisco TAC? Did you need to >> enable an account and password for that access or simply access to the >> system? ------------------------------------------------------------------------ Test Your IDS Is your IDS deployed correctly? Find out quickly and easily by testing it with real-world attacks from CORE IMPACT. Go to http://www.securityfocus.com/sponsor/CoreSecurity_focus-ids_040708 to learn more. ------------------------------------------------------------------------
