NDR Attacks
My Exchange 2000 server is being hit with 3,000-5,000 NDR type of emails every day. By reading the various yet excelent posts here at EE, I have implemented the following steps:
- Relaying is turned off and has been verified using telnet, ehlo, mail from, rcpt to etc. The checkbox to "allow all authenticated computers to relay" is NOT checked.
- The number of connections to the SMTP server is limited to five connections with a 10-minute timeout.
- In Exchange System Manager, the ability to send NDR reports is disabled (not checked). This is still enabling NDR to our internal users as expected.
- The badmail directory is emptied nightly with a cleanup script.
As a result of implementing the above steps, two things have happened:
- The SMPT Queues have gone way way way down which is resulting in a lower CPU usage.
- The badmail directory is not getting as full. Instead of 4,000 badmails per day, only a few hundred are appearing in the directory.
So I have a few questions:
1) Are the above steps satisfactory to prevent relaying? Is five connections too little?
2) Since I can't stop the incoming NDRs into Exchange 2000 (unless I had an appliance), I can definitely stop the NDRs from leaving Exchange. Are the above steps satisfactory in achieving this?
3) And perhaps the most important question. If the NDRs that continue to come in are "bad mail", why is my badmail directory not filling up at the same rate BEFORE disabling the NDR features?
(heavy points due to some insightful advice sure to follow...)
- Relaying is turned off and has been verified using telnet, ehlo, mail from, rcpt to etc. The checkbox to "allow all authenticated computers to relay" is NOT checked.
- The number of connections to the SMTP server is limited to five connections with a 10-minute timeout.
- In Exchange System Manager, the ability to send NDR reports is disabled (not checked). This is still enabling NDR to our internal users as expected.
- The badmail directory is emptied nightly with a cleanup script.
As a result of implementing the above steps, two things have happened:
- The SMPT Queues have gone way way way down which is resulting in a lower CPU usage.
- The badmail directory is not getting as full. Instead of 4,000 badmails per day, only a few hundred are appearing in the directory.
So I have a few questions:
1) Are the above steps satisfactory to prevent relaying? Is five connections too little?
2) Since I can't stop the incoming NDRs into Exchange 2000 (unless I had an appliance), I can definitely stop the NDRs from leaving Exchange. Are the above steps satisfactory in achieving this?
3) And perhaps the most important question. If the NDRs that continue to come in are "bad mail", why is my badmail directory not filling up at the same rate BEFORE disabling the NDR features?
(heavy points due to some insightful advice sure to follow...)
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
more on SPF (tutorials, etc) here: http://www.openspf.org/
ASKER
yuja,
My server was always set to not allow relaying; therefore, we are not on any open relay lists.
1) I'll increase the connection count. You brought up a good point.
2) I have not heard of SPF so I will read up on that.
3) The NDR feature I was referring to was disabling (unchecking the box) the ability to send out NDRs (Exchange Manager, Internet Mail Delivery, Properties).
To clarify your confusion. Since open relay has always been disabled in my SMTP, I am taking the NDR attacks (all with fake email addresses) one step further by disabling the NDR sending. I'll let them come in but won't let them go out. Tests from my Hotmail account are successful in that it does not receive a NDR.
Since the time I unchecked the box, the amount of bad mail in the Badmail directory has dropped to less than a few hundred per day yet the amount of NDRs coming in have remained the same at a few thousand per day. Is this normal behavior of disabling NDRs? If not, then where would you suggest I further check in my environment?
I like the GFI products. Is MailEssentials installed on the Exchange server or on a gateway server? I do not have a gateway server since we are a small firm.
My server was always set to not allow relaying; therefore, we are not on any open relay lists.
1) I'll increase the connection count. You brought up a good point.
2) I have not heard of SPF so I will read up on that.
3) The NDR feature I was referring to was disabling (unchecking the box) the ability to send out NDRs (Exchange Manager, Internet Mail Delivery, Properties).
To clarify your confusion. Since open relay has always been disabled in my SMTP, I am taking the NDR attacks (all with fake email addresses) one step further by disabling the NDR sending. I'll let them come in but won't let them go out. Tests from my Hotmail account are successful in that it does not receive a NDR.
Since the time I unchecked the box, the amount of bad mail in the Badmail directory has dropped to less than a few hundred per day yet the amount of NDRs coming in have remained the same at a few thousand per day. Is this normal behavior of disabling NDRs? If not, then where would you suggest I further check in my environment?
I like the GFI products. Is MailEssentials installed on the Exchange server or on a gateway server? I do not have a gateway server since we are a small firm.
GFI Mail essentials has been good for us. I use MailEssentials and MailSecurity. Both are installed on the exchange server itself with active directory integration.
In the system manager/message delivery properties (sender filtering) do you have filter messages with blank sender and drop connection if address matches filter checked? (receipient filtering)-check filter receipients who are not in active directory?
Just some basic stuff that helps, but like others have said GFI Mail Essentials are pretty good.
kshays
In the system manager/message delivery properties (sender filtering) do you have filter messages with blank sender and drop connection if address matches filter checked? (receipient filtering)-check filter receipients who are not in active directory?
Just some basic stuff that helps, but like others have said GFI Mail Essentials are pretty good.
kshays
ASKER
kshays,
We have E2k which I believe does not have recipient filtering.
So to make up for this shortfal, will Mail Essentials handle the NDR attacks?
We have E2k which I believe does not have recipient filtering.
So to make up for this shortfal, will Mail Essentials handle the NDR attacks?
Dangit, sorry, i overlooked the 2000 and thought it was 2003.
My bad.
kshays
PS: have you went to www.gfi.com and looked through their product documentation yet. Some like GFI and some don't, but I find it very useful though.
:)
My bad.
kshays
PS: have you went to www.gfi.com and looked through their product documentation yet. Some like GFI and some don't, but I find it very useful though.
:)
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
"So to make up for this shortfal, will Mail Essentials handle the NDR attacks?"
MailEssentials is an antispam product. If you include the NDR keywords in it - it will block those messages (or discard them or whatever you want to do with them)
MailSecurity, on the other hand, does have massive email defense system - including NDRs
MailEssentials is an antispam product. If you include the NDR keywords in it - it will block those messages (or discard them or whatever you want to do with them)
MailSecurity, on the other hand, does have massive email defense system - including NDRs
ASKER
Got it,
We already have an antispam product in place with 98% success rate so it looks like I will be looking at MailSecurity since my focus is to prevent NDRs from coming into Exchange. If the user doesn't exist in Active Directory, then block it from even entering Exchange.
Question: some users have multiple email addresses and different domain names. For example, user1@domain1.com and user2@domain2.com. Will MailSecurity handle multiple alias filtering?
We already have an antispam product in place with 98% success rate so it looks like I will be looking at MailSecurity since my focus is to prevent NDRs from coming into Exchange. If the user doesn't exist in Active Directory, then block it from even entering Exchange.
Question: some users have multiple email addresses and different domain names. For example, user1@domain1.com and user2@domain2.com. Will MailSecurity handle multiple alias filtering?
Actually, if user doesn't exist in Active Directory, MailEssentials (rather than MailSecurity) will take care of it, as it is tightly integrated with Active Directory and the Exchange engine itself (this type of attack is called Directory Harvesting). And yes, it will work fine with two - or more - differend domains and multiple aliases.
98% is good, but we have achieved phenomenal percentage rate with MailSecurity + MailEssentials (99.7%). 90 percent is taken care of by checking against various live open relay databases (you can add your own, but there are a bunch of them there as it is) - not only for the originating mail server, but also for LINKS inside emails (if links point to a "bad" server which has been listed as a spam server, the email will get blocked as well). The rest is either picked up by Directory Harvesting module or Bayesian module (Bayesian basically blocks emails based on certain words, and it "teaches" itself as it goes). The rest is killed by the SPF check, Headers check and Keyword check. Our organization has 75% of incoming mail being spam, and out of thousands of emails of spam we receive each day, only about 10 make it through (for all the users in the organization).
Yes, all these checks do slow down things a little bit (your incoming email will be delayed for 4-5 seconds each), but the blocking rate is just tremendous.
You will need to add certain keywords to block most of the phishing (Paypal+suspended, eBay+suspended, account+suspended, Paypal+stolen, eBay+stolen, account+stolen seems to eliminate 99% of them) while MailSecurity completely protects against viruses in attachments, various exploits, blocks (or lets through) attachment of the specified type, tell you if the attachment extension does not match the file's mime type (for example, an exe file renamed to txt file).
98% is good, but we have achieved phenomenal percentage rate with MailSecurity + MailEssentials (99.7%). 90 percent is taken care of by checking against various live open relay databases (you can add your own, but there are a bunch of them there as it is) - not only for the originating mail server, but also for LINKS inside emails (if links point to a "bad" server which has been listed as a spam server, the email will get blocked as well). The rest is either picked up by Directory Harvesting module or Bayesian module (Bayesian basically blocks emails based on certain words, and it "teaches" itself as it goes). The rest is killed by the SPF check, Headers check and Keyword check. Our organization has 75% of incoming mail being spam, and out of thousands of emails of spam we receive each day, only about 10 make it through (for all the users in the organization).
Yes, all these checks do slow down things a little bit (your incoming email will be delayed for 4-5 seconds each), but the blocking rate is just tremendous.
You will need to add certain keywords to block most of the phishing (Paypal+suspended, eBay+suspended, account+suspended, Paypal+stolen, eBay+stolen, account+stolen seems to eliminate 99% of them) while MailSecurity completely protects against viruses in attachments, various exploits, blocks (or lets through) attachment of the specified type, tell you if the attachment extension does not match the file's mime type (for example, an exe file renamed to txt file).