well…that’s most unusual. we don’t have widespread (or any) reports of this; so, as with most issues, we’ll be playing the game of “what’s semi-unique about you.” Not bad-unique, but unique. There are a lot of knobs that can be set, and a lot of ways a phone can get used/deployed in which testing doesn’t provide complete coverage.
One issue semi-related to this (failure to notify user of a pending restart, but even with that issue, phone shouldn’t have restarted while on-call) was addressed in 4_3_0, back in April of 2022, but you’re well beyond that if you’re on 4_13 or 4_15.
Another semi-related issue was introduced in 4_13_0 and was fixed in 4_13_4, whereby the phone could get into a bad state (restart) if the DNS servers became unreachable. But, if you’re on 4_13_5 or newer, you wouldn’t have seen that. That’s also something that probably would be seen before you started making calls - after boot before ready.
If you want to try to see what precipitates the restart yourself, you can turn on the phones’ remote syslogging capabilities and watch. You’d be looking for these parameters in your baseconfig:
<setting id="log_level" value="debug" />
<setting id="enable_logging" value="0" />
<setting id="log_server" value="10.1.2.3" />
<setting id="log_port" value="514" />
You’d have to turn logging on (it’s off by default) and have a syslog server prepared to receive the details at whatever address and port you define.
If you engage with Support, which is also fine as a first step, in lieu of doing your own logging, they’d want the information preceding the restart. That information can also be provided to them if you grab a debug dump (after the fact) from the phone (take a browser to http://ip of phone/cgi-bin/ptsr) and hand it to them. They’ll probably want the dump either way because it’ll contain other information useful for Support.
It’s less likely, but not impossible, that if what precipitates the problem occurs too far (timeline) before the restart manifests, the debug dump won’t grab it (the cause) - the phone has limited memory and only keeps so much rolling data in its local memory, so syslogging would be required.