summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorAndreas Baumann <mail@andreasbaumann.cc>2019-03-29 14:36:48 +0100
committerAndreas Baumann <mail@andreasbaumann.cc>2019-03-29 14:36:48 +0100
commit8ef4eed8bd39ad4cd72ee48818a774aeebd485e1 (patch)
tree965884d4f5888790546acb64e955304717d61405
parentaeb0aab3e57e8f363152958e1a1f2a06aece8b0c (diff)
downloadwww-andreasbaumann-cc-8ef4eed8bd39ad4cd72ee48818a774aeebd485e1.tar.gz
www-andreasbaumann-cc-8ef4eed8bd39ad4cd72ee48818a774aeebd485e1.tar.bz2
published mail blog entry
-rw-r--r--content/blog/mail-disaster.md46
1 files changed, 28 insertions, 18 deletions
diff --git a/content/blog/mail-disaster.md b/content/blog/mail-disaster.md
index ea18bbe..d7c3545 100644
--- a/content/blog/mail-disaster.md
+++ b/content/blog/mail-disaster.md
@@ -3,7 +3,6 @@ title = "Mail Problems"
categories = [ "Mail", "Linux", "Security" ]
date = "2019-03-29T12:58:31+01:00"
thumbnail = "/images/blog/mail-disaster/mail-disaster.png"
-draft = true
+++
## History
@@ -29,17 +28,20 @@ went to sleep.
Of course this was not the case: I had a weak password in one account
of my mailserver (which allowed any legitimate Linux user to send
-emails). So, my thinking went along the lines: well, some weeks ago
+emails). This caused all those DNS lookups for my domain on the
+BuddNS DNS servers.
+
+So, my thinking went along the lines: well, some weeks ago
I replaced the SD card, because the old one was worn out, I cannot
remember whether I replaced all standard passwords. My suspicion got
-confirmed when I saw the maillog sending from the email address:
+confirmed when I saw the following line in the my mail log:
```
From: "George"<alarm@andreasbaumann.cc>
```
-Swearing big times about my own stupidity (the default password is - well -
-weak) I started cleaning up the mess.
+Swearing big times about my own stupidity (the default password for the
+'alarm' account is - well - weak) I started cleaning up the mess.
Checking my mail server logs I found that all attacks went via one single
IP (185.228.80.18). So just blocking the firewall was the fastest way to
@@ -48,7 +50,7 @@ fix the tousands of spam email being sent via my now-defacto-open mail relay.
## Checking status
There are various helpfull tools to check about the status of your mail
-server. I picked the https://mxtoolbox.com/. This is what I got:
+server. I picked https://mxtoolbox.com/. This is what I got:
```
dmarc andreasbaumann.cc DNS Record not found
@@ -62,9 +64,7 @@ server. I picked the https://mxtoolbox.com/. This is what I got:
mx andreasbaumann.cc DMARC Quarantine/Reject policy not enabled
```
-I also like the results from
-
-http://zy0.de/q/83.150.2.48
+I also like the results from http://zy0.de/q/83.150.2.48:
{{< figure src="/images/blog/mail-disaster/zy0_de.png" alt="zy0_de check resulst for 83.150.2.48" >}}
@@ -203,7 +203,7 @@ Record removed. The published list is updated hourly, so changes may not show im
https://www.spamcop.net/w3m?action=checkblock&ip=83.150.2.48
-I filed in the provided form.
+I filled in the provided form.
### IBM DNS
@@ -227,7 +227,8 @@ https://support.google.com/mail/contact/msgdelivery
The postmaster tools are not a big help, really, I registered nonetheless.
I got reject till March 28th, as far as I can tell the domain reputation below
-is one of the worst ones you can get and the only option is to wait:
+is one of the worst ones you can get and the only option is to wait some weeks
+after filling in the forms:
```
Our system has detected that this message
@@ -235,20 +236,29 @@ is 550-5.7.1 likely suspicious due to the very low
reputation of the sending 550-5.7.1 domain
```
-## Course of Action for better mail service
+## Course of Action for a better mail service
I made sure, I have some security standards in place, so that
-at least faking the domain is not so simple:
-[SPF](https://en.wikipedia.org/wiki/Sender_Policy_Framework),
-[DKIM](https://en.wikipedia.org/wiki/DomainKeys_Identified_Mail) and
-[DMARC](https://en.wikipedia.org/wiki/DMARC).
+at least faking the domain in the 'From:' field is not so simple:
+
+* [SPF](https://en.wikipedia.org/wiki/Sender_Policy_Framework): Sender Policy Framework
+* [DKIM](https://en.wikipedia.org/wiki/DomainKeys_Identified_Mail): Domain Keys Identified Mail
+* [DMARC](https://en.wikipedia.org/wiki/DMARC): Domain-based Message Authentication, Reporting and Conformance
Those things don't help against a broken account on the mail server,
as in my case, but they provide positive rating for emails being
judged in the future, and they are simple to implement.
-And of course, I deleted the 'alarm' account on the machine. :-)
+I also added a list of accounts/emails to the postfix configuration.
+Only those accounts are allowed to send emails from the host.
+Even if this means you have to generate the entry in '/etc/passwd'
+and another one in that postfix list. This makes sure,
+no "rogue" Linux account can be abused for sending emails, when
+compromised.
+I added myself to the [DNSWL](https://www.dnswl.org) white list too.
+
+And of course, I deleted the 'alarm' account on the machine. :-)
## References
@@ -256,4 +266,4 @@ And of course, I deleted the 'alarm' account on the machine. :-)
* http://zy0.de/
* https://en.wikipedia.org/wiki/Sender_Policy_Framework
* https://en.wikipedia.org/wiki/DomainKeys_Identified_Mail
-
+* https://en.wikipedia.org/wiki/DMARC