Latest pfSense update causes interface/VLANs to crash on heavy traffic
-
Hello,
Prior to updating to 2.2.5-RELEASE I did not have any trouble transferring any files over 1GB cross VLAN. After updating, I am no longer able to transfer large files >1GB cross VLAN. Doing so will cause the network adapter to go into a down link state and cause all VLANs to become disconnected.
pfSense Specs:
Motherboard: ASRock H61MV-ITX LGA 1155
CPU: Intel Celeron G1610 Ivy Bridge Dual-Core 2.6GHz
RAM: 4GB
Onboard nic (LAN w/VLANs) & TP-LINK TG-3468 (WAN)Network Setup:
VLAN 10 - 192.168.1.0/24 GW: 192.168.1.1 (pfSense)
VLAN 11 - 10.1.2.0/24 GW: 10.1.2.1 (pfSense)File transfers from 192.168.1.200 (desktop) to 10.1.2.18 (file server) fail and cause pfSesnse to temporarily drop VLANs. File transfer from 10.1.2.14 (laptop) to 10.1.2.18 (file server) sucessfully transfers.
I've uploaded a video of the behavior:
https://www.youtube.com/watch?v=ufwO1Gp_OHw
-
RealTek driver regression of some sort?
I saw a watchdog mentioned. Could that be related to apinger or only directly to the interface? I know apinger has issues where it can falsely think the target is down and drop the link. I am just throwing around ideas because I don't understand PFSense that well.
-
RealTek driver regression of some sort?
I saw a watchdog mentioned. Could that be related to apinger or only directly to the interface? I know apinger has issues where it can falsely think the target is down and drop the link. I am just throwing around ideas because I don't understand PFSense that well.
I think you may be right on the Realtek driver issue.
I'm going to swap out my single NIC card for a dual port Intel one and see what happens.
-
File transfers from 192.168.1.200 (desktop) to 10.1.2.18 (file server) fail and cause pfSesnse to temporarily drop VLANs.
Why only temporarily? Or is there perhaps a rule that prevents this action?
File transfer from 10.1.2.14 (laptop) to 10.1.2.18 (file server) sucessfully transfers.
It is the same VLAN and Subnet (CIDR) and in normal the pfSense is not involved because
their must nothing be routet, all is done by the switch if one is in the game! If there was no
switch in the game perhaps the 4 GB was to less or the memory system was rendered or saturated.I'm going to swap out my single NIC card for a dual port Intel one and see what happens.
Would be glad to hear about how this change was working, if so, please keep us up to date.
-
Why only temporarily? Or is there perhaps a rule that prevents this action?
Because watchdog detects the interface is down and restarts it..I think.
It is the same VLAN and Subnet (CIDR) and in normal the pfSense is not involved because
their must nothing be routet, all is done by the switch if one is in the game! If there was no
switch in the game perhaps the 4 GB was to less or the memory system was rendered or saturated.Yep the only reason why I showed that in the video was to show that the switch wasn't causing the issue.
I'm hoping the new NIC card will fix the issues.
-
Replacing the RealTek device with an Intel EXPI9402PT solved the issues.