PPPOE disconnect
-
Try with a pfSense 2.3 snapshot, it's most likely some interaction between FreeBSD and that board/BIOS if tweaking powerd brings out the problem.
Next step is to reproduce the problem with your hardware on plain FreeBSD 10.3 and if it also shows the problem there, then report it directly to FreeBSD.
-
I ran a live image with OPNsense 16.1.8-amd64/FreeBSD 10.2-RELEASE-p14/OpenSSL 1.0.2g on the same hardware and I did several speed tests and it worked without a problem.
-
Apples and oranges (Different FreeBSD version bases) – compare against pfSense 2.3.
-
I did that and it does not work. Pfsense 2.2.6 or 2.3 do not work with this hardware correctly.
-
Apples and oranges (Different FreeBSD version bases) – compare against pfSense 2.3.
I understand, I wanted to check if the hardware is too slow.
-
I have updated to 2.3-RC (amd64) built on Wed Apr 06 05:34:38 CDT 2016 /FreeBSD 10.3-RELEASE
The GUI certainly looks better.
And the extra help info next to the settings is very welcome.But the problem is still there.
-
Tried FreeBSD 10.3 with PPPoe and it stays up without any problem for 48h+. The problem is in pfSense itself somewhere.
-
I have a quite different hardware configuration:
pfsense 2.2.6 under ESXi 6.0U2 (ie. virtualised)
what I am seeing looks very similar in that the PPPoE disconnects frequently, and the disconnections appear to be load related. When there was little/no Internet usage (ie. nobody at home) the PPPoE link stayed up for 4 days. When the net starts getting used a lot it is up/down a lot - sometimes a few minutes apart. The log shows something like:
Apr 26 08:07:44 router ppp: [wan] IFACE: Up event
Apr 26 08:07:44 router ppp: [wan] IFACE: Rename interface ng0 to pppoe0
Apr 26 08:16:21 router ppp: Multi-link PPP daemon for FreeBSD
Apr 26 08:16:21 router ppp:
Apr 26 08:16:21 router ppp: process 52834 started, version 5.7 (root@pfSense_REL
ENG_2_2_amd64_amd64-pfSense_RELENG_2_2-job-03 20:05 21-Aug-2015)
Apr 26 08:16:21 router ppp: waiting for process 14050 to die…
Apr 26 08:16:21 router ppp: caught fatal signal TERM
Apr 26 08:16:21 router ppp: [wan] IFACE: Close eventI don't think that this is related to virtualisation. I also have a copy of pfsense 2.2.5 running under ESXi 5.5U3 and the PPPoE links stay up fine with no TERM signals encountered. I suppose it could relate to the 6.0U2 vs 5.5U3 for ESXi, however, given that these other reports are here I suspect it is more likely a 2.2.6 issue.
I think I will run up a 2.2.5 pfsense and see whether it has the same problems.
Tim
-
I have now tried 2.2.5/2.2.6/2.3 and all releases show similar behaviour in my vmware environment.
The 2.2.5 release does not include the "caught fatal signal TERM" in the logs, however, in all other respects the logs look the same.
So I have a working production configuration: pfsense 2.2.5 / ESXi 5.5U3 / Dell T320 server / TP-Link TD-8817 modem
The configuration with the PPPoE Disconnect issue is: ESXi 6.0U2 / Intel NUC / Apple USB Ethernet for PPPoe / TP-Link TD-8817 modem
So, I suppose my next investigation is to switch to using VLANs to obtain multiple network links and thus delete the USB Ethernet. The USB Ethernet connections get a lot of bad press, so perhaps it is the source of my problem.
-
Well, it looks like my problems with PPPoE are related to the use of the USB Ethernet adapter for the PPPoE link.
A) I moved the USB Ethernet to a 'charging' USB port on the Intel NUC. The 'charging' USB ports can supply up to 1.5A (or there abouts). The result of this was to significantly improve the stability of the PPPoE. However, there were still still multiple PPPoE dropouts with the 'caught signal TERM' message in the logs over the next 24 hours.
B) I then moved the PPPoE traffic onto a VLAN on the main Ethernet port. I have left the USB Ethernet plugged in and attached to pfsense - however, it is not being used for anything. The result of this is that there have been no PPPoE outages due to the 'caught signal TERM', all outages have been loss of LCP echo on the link. Link uptime is significantly better.
So, in summary, I have no idea of exactly where the 'problem' is with the USB Ethernet adapter, but it certainly looks like attempting to use it for a PPPoe link does not work. So my attempt to work around only having a single Ethernet port on the NUC by using a USB Ethernet for the PPPoE wasn't the best idea.
Tim
