New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
FS#854 - Unstable Internet caused by frequent PPPoE reconnect, network.wan.keepalive=0 has no effect #7996
Comments
ppuzr: It turns out the last commands I tried below didn't help. pppd still didn't ignore echo failures |
ppuzr: Before a proper solution is found, it seems that I could work around it by setting a large lcp-echo-interval (86400) to reduce the echo request frequency and optionally a large echo failure threshold (1024) |
MomenMamdouh: the unstable pppoe connection still exists in lede-17.01.4 |
kochstefan: I have noticed this issue the first time at the begin of the year 2016 with OpenWRT. The issue occurs for me every time when connecting to my SSH-server behind the OpenWRT router and load (SSH/SCP server uploads to SSH/SCP downloading remote client) a big file. So the SSH/SCP client does this: To reproduce: The SSH server behind OpenWRT router needs a slower internet connection as the remote client (server's upload speed is slower than client's download speed). So the upload speed of the SSH servers internet connection is fully exhausted. I assume that the keepalive message does not get transmitted properly because of exhausted internet connection. The question is, why there is no prioritization or balancing. The workaround for me was to add "option keepalive '100 30'" to the WAN interface within /etc/config/network. Recently, I noticed a similar failure again. The issue occurs every time when loading a big file via SCP, too. So I have tried your suggestion of '1024 86400', too. But it seems to be another error (not the keepalive issue): If no keepalive option is set I get the log messages: With both options '100 30' and '1024 86400' there is no "No response to X echo-requests" message within the log. pppd was started with: I use OpenWrt SNAPSHOT, r6013-112f0469c4 on a lantiq xway vrx288 based router with a 25 MBit/s (down), 5 MBit/s (up) VDSL connection. The following workaround is possible, too. It solves both the keepalive and Modem hangup errors. Enabling SQM (simple.qos) on pppoe-wan, in my case with 25000 kbit/s at download and 5000 kbit/s at upload. A luci web page for SQM configuration is available, too. VDSL2 syncs with fixed rate down: 25.088 Mb/s and up: 5.056 Mb/s. |
Sven: I had the exact same problem on an o2 Box 6431 (Arcadyan VGV7510KW22) and managed to resolve it by activating QoS and limiting pppoe-wan's bandwidth (both uplink and downlink) to 97%. Many thanks @kochstefan! |
ppuzr:
daemon.info pppd[7738]: No response to 5 echo-requests daemon.notice pppd[7738]: Serial link appears to be disconnected. daemon.info pppd[7738]: Connect time 1.0 minutes. daemon.info pppd[7738]: Sent 28006 bytes, received 46170 bytes. daemon.notice pppd[7738]: Connection terminated. daemon.info pppd[7738]: Connect time 1.0 minutes. daemon.info pppd[7738]: Sent 28006 bytes, received 46170 bytes. daemon.info pppd[7738]: Sent PADT daemon.info pppd[7738]: Exit.
network.wan.keepalive is not set, and on luci "LCP echo failure threshold" shows a grey 0 and the description below it says "Presume peer to be dead after given amount of LCP echo failures, use 0 to ignore failures", but this description is not consistent with the behaviour.
After running the commands below, the grey 0 becomes darker, but the problem persists.
root@LEDE:
# uci set network.wan.keepalive=0# /etc/init.d/network restartroot@LEDE:
The value of network.wan.keepalive when set using luci has two numbers separated by a space (which appears to be in the format of '[threshold] [interval]'), but it should be a number according to the wiki [[https://lede-project.org/docs/user-guide/wan_interface_protocols?s[]=pppoe#protocol_pppoe_ppp_over_ethernet]]
I then tried the following commands, and there're no reconnects after one hour.
root@LEDE:
# uci set network.wan.keepalive='0 1'# /etc/init.d/network restartroot@LEDE:
Is it better to have a default of 0 (default=undefined network.wan.keepalive) instead of seemingly 5 which would also be consistent with luci?
The text was updated successfully, but these errors were encountered: