LCP echo failure threshold 0 does not behave as described in LuCI

[bill888uk]

For BT Home Hub 5A configured to use PPPoE or PPPoA protocol for example.

In LuCI→Network→Interfaces→WAN→Advanced Settings, there are two default settings displayed:

LCP echo failure threshold 0
LCP echo interval 5

On the page, the 'default' setting LCP echo failure threshold to Zero, implies all LCP echo failures would be ignored. (see attached image) But I found this to be incorrect/misleading.

There is no keepalive option found in the WAN interface section of /etc/config/network on Home Hub 5A when selecting PPPoE or PPPOA protocols. But there appears to be a fixed timeout (possibly 5 seconds) on the Home Hub 5A when I use PPPoE on VDSL connection. ie. not infinite as quoted in LuCI.

I don't know if this is the same behaviour for all devices.

If LCP echo failure threshold is set to a non-Zero value, then keepalive option is CREATED in the WAN interface within /e/c/network file.

The description displayed in LuCI for LEDE and OpenWRT 18:

'Presume peer to be dead after a given amount of LCP echo failures, use 0 to ignore failures'

perhaps needs to be changed?

eg. perhaps simplest solution is to remove the phrase 'use 0 to ignore failures' ?

This closed ticket may be of interest (I added my observations affecting HH5A running LEDE 17 to end of the closed ticket)
https://bugs.openwrt.org/index.php?do=details&task_id=1259

Would it be possible to add 'keepalive_adaptive' setting into LuCI mentioned in above ticket in the future?

Another ticket on same subject
https://bugs.openwrt.org/index.php?do=details&task_id=854

[jow-]

From ppp.sh:

        local lcp_failure="${keepalive%%[, ]*}"
        local lcp_interval="${keepalive##*[, ]}"
        ...
        [ "${lcp_failure:-0}" -lt 1 ] && lcp_failure=""
        ...
        proto_run_command "$config" /usr/sbin/pppd \
                ${lcp_failure:+lcp-echo-interval $lcp_interval lcp-echo-failure $lcp_failure $lcp_adaptive} \

Means if you set it to 0 in LuCI, option keepalive will be deleted. When no option keepalive is present, $lcp_failure will end up being empty which means that no lcp-echo-interval and no lcp-echo-failure is passed to pppd.

[jow-]

Default values in pppd lcp.c:

int     lcp_echo_interval = 0;  /* Interval between LCP echo-requests */
int     lcp_echo_fails = 0;     /* Tolerance to unanswered echo-requests */

So I am confused. The option seems to behave as intended?

[jow-]

To be perfectly clear: 0 is supposed to disable LCP stuff. If that does not work it is a bug that has to be fixed in OpenWrt.

[jow-]

Turns out the OpenWrt implementation is completely flawed and cannot be fixed in a way where omitting option keepalive would yield the expected behaviour without breaking the majority of existing setups.

Will change LuCI to emit an explicit option keepalive 0 but that'll still need fixes on the OpenWrt side.

[kiral6]

from master ppp.sh:

    [ -n "$keepalive" ] || keepalive="5 1"
    
    local lcp_failure="${keepalive%%[, ]*}"
    local lcp_interval="${keepalive##*[, ]}"
    local lcp_adaptive="lcp-echo-adaptive"

In the LuCI setting, the default value is to leave blank, so $keepalive is null. At this time, lcp-failure=5, lcp-interval=1, this is inconsistent with the description of LuCI. And if you fill in 0/5 in LuCI, you can't save it, because the value hasn't changed and doesn't need to be saved, then $keepalive will always be null, which means that you can't set lcp-echo-failure = 0 through LuCI.