What is the biggest attack in GBPS you stopped
-
Thanks man!
Over my head…. I need others to chime in on this. Spent the last mth using all my spare time to test this frigging thing.
After I did the cron job, then I can make it come alive again.
I have to settle with that right now...........
Currently migrating all the pf's in the datacenter to Mikrotik since I cant have this hosting customers. At home I dont care.
-
So is it worth starting a Getting Dtrace to work on pfsense thread?
We only have to ask the right questions, document everything we do so that others can learn from it as well if they have the thirst for knowledge.
There's plenty of examples online which show how to compile code onto platforms where a package doesnt exist for downloading.
There's generally parallels everywhere if you look. ;)
-
Posted a bounty for Dtrace package.
https://forum.pfsense.org/index.php?topic=94846.0
-
dtrace is installed already on pfSense.
Go to the FreeBSD forums now. You're not dealing with pfSense anymore when you start working with dtrace. This is the wrong place to look for that kind of guidance.
I think I'm having déjà vu….
-
OK.
But it still leaves me with no options to monitor pfsense deep under the hood with timestamps.
https://forum.pfsense.org/index.php?topic=94260.0
Any ideas?
-
You need to be more specific with what you want exactly, plus programmers are very specific and you might find Dtrace does what you want, so would it be fair for anyone to just compile and get Dtrace running?
Besides when you say what you say, have you considered the other app/conversation thread here https://forum.pfsense.org/index.php?topic=94843.msg527152#msg527152 or is your browser running single tab/single core/single thread and you have not locked/mutexed this thread and swapped to the other thread yet? Or if your browser is running multitabbed/multi core/multi threaded, have you locked this thread in your browser by switching tabs to read the other thread/multi core/multi thread if you catch my drift. ;D
There's parallels everywhere if you have an abstract look on life.
-
dtrace is installed already on pfSense.
Go to the FreeBSD forums now. You're not dealing with pfSense anymore when you start working with dtrace. This is the wrong place to look for that kind of guidance.
I think I'm having déjà vu….Significance might not have registered. Apologies. :o
-
dtrace stuff. Remember, dtrace is a component of the FreeBSD distribution, so you need to go there to understand it and get support from the community. These forums are more product-specific for pfSense–mostly dealing with networking versus deep coding. So you're questions are best answered in the FreeBSD forums.
https://wiki.freebsd.org/DTrace
https://wiki.freebsd.org/DTrace/Tutorial
http://www.freebsd.org/doc/en_US.ISO8859-1/books/handbook/dtrace.html
http://www.brendangregg.com/USEmethod/use-freebsd.html
-
Smoking gun?
https://forums.freebsd.org/threads/freebsd-10-1-network-unaccessible-after-high-traffic.51743/
Potential fix? (synproxy state)
https://forums.freebsd.org/threads/enabling-ipv6-eventually-freezes-the-gateway.47959/#post-267820
See #4 from this link:
http://www.onlamp.com/pub/a/bsd/2004/04/15/pf_developers.html
https://forums.freebsd.org/threads/freebsd-9-0-release-syn-flood.29071/#post-163533
Notice how all of this freakin' awesome information is coming from the FreeBSD forums. I strongly suggest you consolidate your findings in this thread and move the discussion over to the FreeBSD networking forums. They have dozens of discussions on SYN floods with some excellent guidance given.
-
Yes. It just doesnt work as expected.
Until the drop in traffic occurs. So I need to find a way to see whats initiated when the traffic drops.
-
Yes. It just doesnt work as expected.
Until the drop in traffic occurs. So I need to find a way to see whats initiated when the traffic drops.
https://forums.freebsd.org/forums/networking.7/
-
Yes….one more forum to watch....................... :D
-
Just a small update..
Changed NIC's to vmxnet3 in vmware and this happened.
Same traffic drop but curves are different. On E1000 they follow each other. On vmxnet3 they are totally seperated. Red is 15mbit/s and black is almost zero.
Suddenly then traffic drops on incoming WAN and it begins to flow traffic and respond to ping for Google DNS.
No change what so ever to anything.
vmxnet3 drivers are much worse handling this kind of attack than the E1000 drivers.
EDIT: When Core nr 3 drops, the FW comes alive and begins to route packets. Overall load is cut in half.
EDIT EDIT: Getting 100% interrupt in the top -HSP on the console. Again the E1000 almost had zero.
Logs are flooded with this.
| Jun 7 19:54:14 | check_reload_status: Reloading filter |
| Jun 7 19:54:14 | check_reload_status: Restarting OpenVPN tunnels/interfaces |
| Jun 7 19:54:14 | check_reload_status: Restarting ipsec tunnels |
| Jun 7 19:54:14 | check_reload_status: updating dyndns Yousee |
| Jun 7 19:54:01 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:53:49 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:53:43 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:53:38 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:53:33 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:53:28 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:53:22 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:53:15 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:53:06 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:53:01 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:52:56 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:52:51 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:52:45 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:52:40 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:52:35 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:52:28 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:52:23 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:52:18 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:52:13 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:52:07 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:52:02 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:51:56 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:51:50 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:51:46 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:51:40 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:51:34 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:51:29 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:51:24 | check_reload_status: Reloading filter |
| Jun 7 19:51:24 | check_reload_status: Restarting OpenVPN tunnels/interfaces |
| Jun 7 19:51:24 | check_reload_status: Restarting ipsec tunnels |
| Jun 7 19:51:24 | check_reload_status: updating dyndns Yousee |
| Jun 7 19:51:22 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:51:16 | kernel: vmx0: watchdog timeout on queue 0 |
| Jun 7 19:51:11 | kernel: vmx0: watchdog timeout on queue 0 |
-
And an update from me.
Currently downloading and getting 5.42Mbps according to the dashboard, yet when you did the ddos the otherday that was only giving showing 2.42Mbps.
So why does a ddos cause my bandwidth to drop down? Could it be related to this post https://forum.pfsense.org/index.php?topic=94903.0 ie a hw issue, or is freebsd or pfsense.
I have a huwei supplied router dont know if it can go into modem only mode, but if it can I will check the speeds again with this device if you dont mind doing a Britney Spears and hit me one more time? ;)
-
No problem.
Let me know when youre ready and what IP/port it is.
-
.-….........
-
How resilient? No longer an issue for established connections as long as there is enough bandwidth, kind of resilient? You tone makes it sound as if it is effectively fixed. Time to party? 8)
-
EDIT: It was my bandwith that caused to attacked IP to survive.
No change in resilience. (Sorry).
-
EDIT: It was my bandwith that caused to attacked IP to survive.
No change in resilience. (Sorry).
It was your bandwidth? Isn't that the goal? The weakest link should be the bandwidth, not the firewall, yes? What issue(s) still remain? Same?
-
As soon as I was able to push 10mbit/s out on my private line, then it failed again….