PfSense syslog and ELSA
-
Thanks Jimp,
I will try to see if ELSA can parse the file better.
Looking forward to 2.2.
Thanks
-
When I modify a Firewall:Rule and hit "apply", I notice that all of the previous log entries in the pfSense Firewall log displays incorrect "Rule Names"?
All new entries in the Log will show the correct Rule name, but the older logs remain indicating an incorrect rule name?
Has anyone seen this issue before? I have tested this on two different boxes with the same issue.
I am using pfBlocker with aliases on pfSense v2.1.
I am also using a patch (http://files.pfsense.org/jimp/patches/pf-log-oneline-option.diff) to allow one line entries for a remote syslog.
Thanks
-
@BBcan17:
When I modify a Firewall:Rule and hit "apply", I notice that all of the previous log entries in the pfSense Firewall log displays incorrect "Rule Names"?
All new entries in the Log will show the correct Rule name, but the older logs remain indicating an incorrect rule name?
That is normal. When the rules reload, the rule numbers can change and the descriptions no longer match up. We're working on a way around that for 2.2 as well.
-
Thanks Jim,
I look forward to the next version.
-
Hello,
I'm using pcSense 2.1 with Elsa and your patches and it's working fine for TCP and UDP packages. However, ICMP doesn't get parsed. I was trying to locate the problem the last days but without luck. If pfSense is sending an ICMP log to Elsa, I receive this in my log:
00:00:02.504973 rule 317/0(match): block in on bge1: (tos 0x0, ttl 64, id 315, offset 0, flags [none], proto ICMP (1), length 84) 172.27.0.36 > 172.27.1.3: ICMP echo request, id 44331, seq 0, length 64
host=172.27.1.3 program=pf class=NONEI was looking at the parsing rules and run some tests by hand, it always get parsed. I then send the same message via netcat:
echo '<14>Jan 22 10:57:48 pf: 00:00:02.504973 rule 317/0(match): block in on bge1: (tos 0x0, ttl 64, id 315, offset 0, flags [none], proto ICMP (1), length 84) 172.27.0.36 > 172.27.1.3: ICMP echo request, id 44331, seq 0, length 64' | nc -v -u -w 0 172.27.50.77 514
and suddenly it gets parsed:
00:00:02.504973 rule 317/0(match): block in on bge1: (tos 0x0, ttl 64, id 315, offset 0, flags [none], proto ICMP (1), length 84) 172.27.0.36 > 172.27.1.3: ICMP echo request, id 44331, seq 0, length 64
host=172.27.0.162 program=pf class=FIREWALL_ACCESS_DENY proto=ICMP srcip=172.27.0.36 srcport=0 dstip=172.27.1.3 dstport=0 o_int=bge1 i_int= access_group=Does anyone else has this problem?
Thanks in advance,
fsom -
What are you using as the search Query in ELSA? Did you look at the Google ELSA group https://groups.google.com/forum/#!forum/enterprise-log-search-and-archive
Try the following example Querys -
class=FIREWALL_CONNECTION_END groupby:host
class=FIREWALL_ACCESS_DENY groupby:host
class=FIREWALL_ACCESS_DENY icmp or tcp or udp dstip="x.x.x.x" groupby:srcipAre you using a parser similar to this?
http://www.securitygrit.com/2013/03/pfsense-into-elsa.html
I have added the ruleset to the end of patterndb.xml between <patterndb></patterndb>and ran "service syslog-ng restart"
The syslogs are going into the existing database ELSA class/fields "FIREWALL_ACCESS_LOG" and "FIREWALL_CONNECTION_END".
Make sure you Select "Grid Display" to see it parsed properly.
-
Hello,
Thank you for your reply. ICMP packages coming from pfSense are going into the CLASS=NONE group, TCP/UDP into the FIREWALL_*. I tried your example queries but the only ICMP package I found was the one I spoofed via netcat. All the other ICMP packages are in the CLASS=none group. The parser I'm using is from http://www.securitygrit.com/2013/03/pfsense-into-elsa.html, before I added it, everything from pfSense was going into CLASS=NONE, after I applied and restarted syslogd they got assigned to the FIREWALL_ACCESS_DENY or FIREWALL_CONNECTION_END class. The thing I don't understand is why ICMP packages aren't going into these classes when coming from pfSense.The parsing itself is working:
[root@elsa ~]# /usr/local/syslog-ng/bin/pdbtool match -p /usr/local/elsa/node/conf/patterndb.xml -P pf -M "00:03:54.505077 rule 50/0(match): block in on bge0: (tos 0x0, ttl 107, id 25728, offset 0, flags [none], proto ICMP (1), length 28) 172.16.250.251 > xxx.yyy.21.150: ICMP echo request, id 768, seq 51812, length 8"
MESSAGE=00:03:54.505077 rule 50/0(match): block in on bge0: (tos 0x0, ttl 107, id 25728, offset 0, flags [none], proto ICMP (1), length 28) 172.16.250.251 > 194.180.21.150: ICMP echo request, id 768, seq 51812, length 8
PROGRAM=pf
.classifier.class=2
.classifier.rule_id=2
s0=bge0
i0=ICMP
i1=172.16.250.251
i3=xxx.yyy.21.150Does your ELSA handle ICMP like UDP/TCP ?
Thanks,
fsom -
I guess I've been sleeping, been waiting a long time for FW logs to be parse on a single line. Thanks Jim for the patch!! I'll have to give this a try or just wait for 2.2 ;)
-
ICMP packages coming from pfSense are going into the CLASS=NONE group
I tested an icmp block rule and can confirm that ELSA is logging this as "class=none" instead of "class=FIREWALL_ACCESS_DENY"
You should post a request to the Google ELSA group.
-
Hi Jimp,
I have followed the following document -
And it was working for about a month or so, but recently any Ubuntu machine that is in a Site-Site Ipsec tunnel can not get any connectivity when this Manual route is implemented. All other traffic from Windows based machines seems to be unaffected by this, Including the Syslogs from pfSense being directed to a remote syslog server.
I have tried this on two different tunnels with different Ubuntu machines.
When I delete the manual route and restart apinger, it doesn't clear the issue. A full reboot of the pfSense box seems to fix it.
Do you have any suggestions?
Thanks for your help.
-
Ditch the route, update to 2.1.1, then Status > System Logs, Settings tab, pick LAN for the source address. :-)
-
Is 2.1.1 stable enough now? If there are no other choices, i guess I will test it out.
This is the reason why they invented "Alcohol!!"
-
It's still got a couple issues yet but it may be good enough for most uses.
Otherwise track down the commit(s) for the syslog source address selection and apply them manually.
-
Otherwise track down the commit(s) for the syslog source address selection and apply them manually.
Hi Jim,
I found this revision, https://redmine.pfsense.org/projects/pfsense/repository/revisions/53c5407e646028a003b2765a87dd3316b21a9497
Would the steps involved be to replace the two files with the ones on this site.
/etc/inc/system.ini
/usr/local/www/diag_logs_setings.php -
You should be able to use the system patches package to apply that patch. Taking the whole files might get other changes that would have unintended consequences.
-
Do you have a link that you could share?
-
http://doc.pfsense.org/index.php/System_Patches
-
Ditch the route, update to 2.1.1, then Status > System Logs, Settings tab, pick LAN for the source address. :-)
Hi Jimp,
Thanks for the direction on getting the Syslog to work thru the VPN tunnel. Works well!
I believe that "System:NOTIFICATION / SMTP" has this same issue.
I have "DNS Forwarder" set to forward "mail.domain.com" to a 10.10.10.5, I have the Notification "Email server" set to "mail.domain.com" and the emails never go out.
If I change the "Email Server" in Notification to 10.10.10.5, the emails don't go out.
When i change "mail.domain.com" to the External IP address of the mail server, the email go thru, as this sends the email out thru the internet to get to my mail server.
Would prefer the mail to stay within my VPN tunnel if possible.
-
Not relevant to this thread, but that would require a route, the smtp client doesn't have a way to force the source address. Start a new thread if you want to discuss alternatives.
-
Not relevant to this thread, but that would require a route, the smtp client doesn't have a way to force the source address. Start a new thread if you want to discuss alternatives.
Hi Jimp,
I posted my question to the group without any replies, would you have any suggestions?
https://forum.pfsense.org/index.php/topic,72149.msg394065.html#msg394065
