Unofficial E2guardian package for pfSense
-
@Marcelloc why did you remove the E2 Guardian start log? It was pretty useful in situations like this.
I didn't. It was removed by robotman3000.
Ah yes, you're right. He removed the entire watch dog code because it caused problems like switching on E2G when it was set to disabled.
However, have you got any ideas why my E2 Guardian could be randomly crashing? It seemed to start when I started experimenting with Traffic Shaper, but I really can't be sure without anything useful in logs. Have you checked to see if it has crashed in your production environment at all?
Other than the hiccup here and there, I've been loving v5. Transparent proxy is awesome, and seems to work well!
EDIT: Found this in the logs now: status_services.php: The command '/usr/local/etc/rc.d/e2guardian.sh stop' returned exit code '1', the output was 'kern.ipc.somaxconn: 16384 -> 16384 kern.maxfiles: 131072 -> 131072 kern.maxfilesperproc: 104856 -> 104856 kern.threads.max_threads_per_proc: 20480 -> 20480 e2guardian not running? (check /var/run/e2guardian.pid)
-
Apr 11 13:32:07 check_reload_status Reloading filter
Apr 11 13:34:17 kernel pid 24878 (e2guardian), uid 106: exited on signal 11
Apr 11 13:37:50 kernel pid 14342 (e2guardian), uid 106: exited on signal 11
Apr 11 13:37:55 kernel pid 90156 (e2guardian), uid 106: exited on signal 11
Apr 11 13:38:04 kernel pid 5340 (e2guardian), uid 106: exited on signal 11
Apr 11 13:38:20 kernel pid 54590 (e2guardian), uid 106: exited on signal 11
Apr 11 13:38:39 kernel pid 84677 (e2guardian), uid 106: exited on signal 11
Apr 11 13:38:46 kernel pid 14612 (e2guardian), uid 106: exited on signal 11
Apr 11 13:39:07 kernel pid 36090 (e2guardian), uid 106: exited on signal 11
Apr 11 13:39:13 e2guardian 92621 I seem to be running already!
Apr 11 13:39:13 root /usr/local/etc/rc.d/e2guardian.sh: WARNING: failed to start e2guardian
Apr 11 13:44:19 kernel pid 95365 (e2guardian), uid 106: exited on signal 11Really unsure how to go about debugging this mess.. But eventually the watchdog wins the battle agains the E2Guardian crashing.
-
Are you using any authentication?
-
Are you using any authentication?
I'm just using IP Authentication, I tried DNS before but that would give me a seg fault instantly. So I'm sticking to IP for the moment.
-
Are you using any authentication?
I'm just using IP Authentication, I tried DNS before but that would give me a seg fault instantly. So I'm sticking to IP for the moment.
No suggestions Marcello? I've got the watch dog running so it does restart E2 Guardian but sometimes it's just a mess. And will crash 5-8x in a row reducing performance and causing disruption.
-
No suggestions Marcello? I've got the watch dog running so it does restart E2 Guardian but sometimes it's just a mess. And will crash 5-8x in a row reducing performance and causing disruption.
May be related to https://github.com/e2guardian/e2guardian/issues/375.
I'll compile a version with debug and include some instructions on how to use, so you can see if it's something already related or not.
-
No suggestions Marcello? I've got the watch dog running so it does restart E2 Guardian but sometimes it's just a mess. And will crash 5-8x in a row reducing performance and causing disruption.
May be related to https://github.com/e2guardian/e2guardian/issues/375.
I'll compile a version with debug and include some instructions on how to use, so you can see if it's something already related or not.
I just re-did my config as this was bothering me and other users a lot. It seems to be better, but I'll continue to monitor it. It's possible it could've been a config issue and without a meaningful error message it's hard to tell where the problem lies.
-
I just re-did my config as this was bothering me and other users a lot. It seems to be better, but I'll continue to monitor it. It's possible it could've been a config issue and without a meaningful error message it's hard to tell where the problem lies.
I've pushed a 5.0.2_3 binaries with some fixes from e2guardian team. check if you're on this latest binaries with pkg info under console.
-
pkg v 5.0.3 is out. Main changes on this version:
-
Apply firewall limiters to direct connect clients (wpad or proxy manual config).
-
New button to clear cert MITM cache data.(useful when you change MITM CA)
-
-
pkg v 5.0.3 is out. Main changes on this version:
-
Apply firewall limiters to direct connect clients (wpad or proxy manual config).
-
New button to clear cert MITM cache data.(useful when you change MITM CA)
Unfortunately re-doing the configs didn't fix the crashing issue. I've just upgraded to the latest 5.0.3 version, and checked with pkg info. I'll see if this fixes it, thanks!
Also, could we apply the limiter to specific groups?
-
-
Also, could we apply the limiter to specific groups?
For transparent proxy, the rules are manual, so l guess yes.
-
@Marcelloc Just an update on the situation. The latest update fixed some of the crashes I had, everything's been running better today. 2500 workers, and I was able to use over 1000 without any huge problems. Its also been faster! Let's hope it stays that way, and improves until we can eradicate all crashes. There was time when my entire pfsense would crash, so I'm grateful for the improvements made.
I have a few suspicions that the crashes may perhaps be related to using pfBlockerNG along with transparent mode. I am trying to explore all the avenues to try locate the fault but it isn't easy without proper debugging. I have disabled pfBlockerNG for the meantime to try and isolate the problem, let's see if there's any improvement.
That being said, I think it's best to keep some debug code that can be enabled easily when required as we can come across quite a bit of unforseen issues with FreeBSD as the developers of E2G use Linux.
-
So disabling pfBlockerNG did nothing, I really thought that would've been it since the transparent proxy is sending all the pfBlocker domains through the proxy and crashing it or something.
I've checked and redone pretty much all the ACLs, removed all test and unnecessary ACLs and I'm having issues. I've even tried multiple restarts and reinstalls…
I'm pretty surprised and shocked no one else is having this.
-
I am also having issues with Transparent proxy mode. I messed with this for quite awhile and it would at times work with only https connections which seems weird. Then reinstalling and or rebooting some times would change things and it would work correctly for a little bit and then it would quit. Sometimes it would not catch any connections at all and connections would bypass E2 completly. But mostly it would not complete http connections while catching https. I know it would try because nothing would happen for awhile till browser would show error.
Also if I configure proxy settings on my computer things work correctly. It seems there are issues with transparent mode.
-
I am also having issues with Transparent proxy mode. I messed with this for quite awhile and it would at times work with only https connections which seems weird. Then reinstalling and or rebooting some times would change things and it would work correctly for a little bit and then it would quit. Sometimes it would not catch any connections at all and connections would bypass E2 completly. But mostly it would not complete http connections while catching https. I know it would try because nothing would happen for awhile till browser would show error.
Also if I configure proxy settings on my computer things work correctly. It seems there are issues with transparent mode.
I've never had issues with it oniy loading HTTPS sites, however I have come across instances where HTTPS traffic just bypasses E2 Guardian. Usually giving E2 Guardian a restart fixes it for the most part. I'm not sure if this is somehow conflicting with pfBlockerNG. I've tried various things but the fact that both E2G and pfBlockerNG both forward traffic to themselves could somewhat cause an issue.
@Marcelloc do you use pfBlockerNG and have you come across any issues?
-
On my labs I do not use pfblockerNG but on a network with 850 workstations yes.
pfblockerNG basically create firewall rules, so the only way I see that can conflict with e2guardian is preventing it to connect to a remote site.
-
Try to disable watch guard and see on access log which was the last site or url accessed before crash. Repeat few times to see if we can get close to a way to reproduce the crash on any system.
-
to install a debug version of e2guardian, paste this under console
pkg add -f https://e-sac.websiteseguro.com/e2g/e2guardian-5.0.2_5.txz
then run
/usr/local/sbin/e2guardian -QN
to reinstall current Unofficial repo version use
pkg install -f e2guardian
-
Try to disable watch guard and see on access log which was the last site or url accessed before crash. Repeat few times to see if we can get close to a way to reproduce the crash on any system.
Great idea, I'll definitely give that a go now. Thanks for your input!
-
to install a debug version of e2guardian, paste this under console
pkg add -f https://e-sac.websiteseguro.com/e2g/e2guardian-5.0.2_3.txz
then run
/usr/local/sbin/e2guardian -QN
to reinstall current Unofficial repo version use
pkg install -f e2guardian
Can we not integrate this into the unofficial rep? And allow the debug mode to be turned on or off? I think that maybe easier to manage and use.
I've tried a lot of things last night to try combat the crashes, I've even changed my pfBlocker ports again even though I've never set them up to conflict with each other. I did see in logs that E2 Guardian was saying that port 8080 was in use. I'm not sure if it was just ghost instances of itself or another service. Nothing else is set to use my E2G ports.